Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files Community
1
paperID
stringlengths
36
36
pwc_id
stringlengths
8
47
arxiv_id
stringlengths
6
16
⌀
nips_id
float64
url_abs
stringlengths
18
329
url_pdf
stringlengths
18
742
title
stringlengths
8
325
abstract
stringlengths
1
7.27k
⌀
authors
stringlengths
2
7.06k
published
stringlengths
10
10
⌀
conference
stringlengths
12
47
⌀
conference_url_abs
stringlengths
16
198
⌀
conference_url_pdf
stringlengths
27
199
⌀
proceeding
stringlengths
6
47
⌀
taskID
stringlengths
7
1.44k
areaID
stringclasses
688 values
embedding
stringlengths
9.26k
12.5k
umap_embedding
stringlengths
29
44
3287b254-306c-48a5-9d61-ac166674cb91
evaluation-methodologies-for-code-learning
2108.09619
null
https://arxiv.org/abs/2108.09619v2
https://arxiv.org/pdf/2108.09619v2.pdf
Impact of Evaluation Methodologies on Code Summarization
There has been a growing interest in developing machine learning (ML) models for code summarization tasks, e.g., comment generation and method naming. Despite substantial increase in the effectiveness of ML models, the evaluation methodologies, i.e., the way people split datasets into training, validation, and test sets, were not well studied. Specifically, no prior work on code summarization considered the timestamps of code and comments during evaluation. This may lead to evaluations that are inconsistent with the intended use cases. In this paper, we introduce the time-segmented evaluation methodology, which is novel to the code summarization research community, and compare it with the mixed-project and cross-project methodologies that have been commonly used. Each methodology can be mapped to some use cases, and the time-segmented methodology should be adopted in the evaluation of ML models for code summarization. To assess the impact of methodologies, we collect a dataset of (code, comment) pairs with timestamps to train and evaluate several recent ML models for code summarization. Our experiments show that different methodologies lead to conflicting evaluation results. We invite the community to expand the set of methodologies used in evaluations.
['Milos Gligoric', 'Raymond J. Mooney', 'Junyi Jessy Li', 'Jiyang Zhang', 'Pengyu Nie']
2021-08-22
null
https://aclanthology.org/2022.acl-long.339
https://aclanthology.org/2022.acl-long.339.pdf
acl-2022-5
['comment-generation']
['natural-language-processing']
[ 2.92584002e-01 7.60981515e-02 -4.02609974e-01 -4.57251757e-01 -8.78984153e-01 -7.45787680e-01 6.78771257e-01 7.18304157e-01 -1.29070327e-01 4.84493226e-01 5.73278189e-01 -5.41050494e-01 1.04799166e-01 -1.40779734e-01 -2.82803625e-01 -5.31061627e-02 2.55423993e-01 -1.05171660e-02 7.35909045e-02 1.71834044e-02 9.73478675e-01 -7.37945661e-02 -1.57463527e+00 6.25055432e-01 1.33491111e+00 -1.44283146e-01 5.57013613e-04 8.71880829e-01 -4.49182600e-01 1.02757132e+00 -1.23510742e+00 -6.48453176e-01 -2.23555237e-01 -7.88225651e-01 -1.15075421e+00 1.78835720e-01 6.73908353e-01 6.04074709e-02 3.76628220e-01 9.00456011e-01 5.62395811e-01 -1.97119966e-01 7.07242727e-01 -1.48111892e+00 -6.10124469e-01 1.02050614e+00 -5.54391086e-01 2.24404037e-01 7.62493193e-01 -2.39236817e-01 1.07214820e+00 -7.09722281e-01 9.31452990e-01 9.11643147e-01 9.92604136e-01 5.91408610e-01 -1.28979278e+00 -4.05874401e-01 7.62023078e-03 4.34612371e-02 -8.37924600e-01 -5.52392364e-01 5.45580745e-01 -1.02863491e+00 1.15986919e+00 4.05733496e-01 5.72220027e-01 9.66040969e-01 3.22398126e-01 1.05583692e+00 6.36054933e-01 -7.43164837e-01 1.50457725e-01 4.67087507e-01 5.12750030e-01 3.81741941e-01 6.69568419e-01 -7.27791607e-01 -3.18386942e-01 -5.58233559e-01 -2.72010025e-02 -2.35498250e-01 -1.46415085e-01 -2.54840463e-01 -1.19052064e+00 9.59361136e-01 -2.93911725e-01 5.29345751e-01 -2.49788426e-02 4.82415035e-02 1.02254391e+00 3.94022614e-01 6.73099697e-01 8.23577762e-01 -4.54698473e-01 -6.58847511e-01 -1.40029991e+00 6.21557176e-01 1.26302683e+00 1.19351792e+00 6.53514266e-01 -2.38669962e-02 -3.69051009e-01 1.02477050e+00 2.88448185e-01 -2.95163114e-02 5.25932610e-01 -1.09212899e+00 7.37405062e-01 8.74515176e-01 1.39748797e-01 -9.48961496e-01 -3.02418858e-01 -1.80115506e-01 -2.86266863e-01 5.09154126e-02 8.20058435e-02 -2.31237531e-01 -3.24734449e-01 1.31432998e+00 -2.71880239e-01 -5.74828327e-01 5.89739904e-02 4.31421362e-02 1.03742504e+00 3.04579586e-01 -7.25623500e-03 -4.34093863e-01 1.03719509e+00 -1.12819779e+00 -6.57567382e-01 -2.89032876e-01 1.33711934e+00 -1.28959095e+00 9.18583393e-01 2.58842975e-01 -1.11766887e+00 -4.50406432e-01 -1.10071325e+00 1.17510773e-01 -1.82067626e-03 2.32191265e-01 4.13760394e-01 8.03489804e-01 -1.15999031e+00 6.35978520e-01 -8.51437986e-01 -9.18785810e-01 1.29167631e-01 -1.29178792e-01 -8.39005932e-02 8.81421268e-02 -6.58551991e-01 9.14884388e-01 2.56914467e-01 -4.27379340e-01 -4.29359823e-01 -6.29938841e-01 -8.74354243e-01 -2.17402950e-01 -1.00431954e-02 -5.77956915e-01 1.97493887e+00 -1.18126345e+00 -7.62833059e-01 7.99588382e-01 -2.04039350e-01 -2.04694360e-01 3.65343869e-01 -3.40633661e-01 -1.89512268e-01 -3.23970526e-01 7.13262081e-01 3.37197095e-01 4.37230706e-01 -1.26012516e+00 -7.31409907e-01 6.63596988e-02 9.44873244e-02 5.92769049e-02 -4.25241500e-01 5.46751380e-01 -1.55188084e-01 -8.04664552e-01 -3.67235065e-01 -9.27630782e-01 -5.91576546e-02 -7.21134305e-01 -3.59343290e-01 -3.71982008e-01 4.59976465e-01 -8.09764206e-01 2.03186584e+00 -1.92941940e+00 8.23655874e-02 -5.19091606e-01 1.13689229e-01 1.17877193e-01 -1.41507834e-01 1.17447913e+00 -1.26625255e-01 6.86667264e-01 -3.98185372e-01 -7.32489944e-01 -3.69305313e-02 -1.65384650e-01 -3.57997090e-01 4.44614261e-01 2.09727697e-02 4.90274906e-01 -1.00785255e+00 -7.13146746e-01 -4.13005412e-01 -2.79123243e-02 -8.15224886e-01 1.45622805e-01 -1.88598886e-01 4.30247793e-03 -4.16006893e-01 7.21440196e-01 4.17898387e-01 -4.14331742e-02 3.55639942e-02 2.21261382e-01 -5.79396546e-01 5.90975225e-01 -7.60919392e-01 1.79985869e+00 -3.80132228e-01 1.12264800e+00 -3.98747474e-01 -6.06759667e-01 8.49457562e-01 4.90550607e-01 4.51406062e-01 -3.15093070e-01 -2.67709315e-01 3.60555798e-01 4.18595299e-02 -9.19209540e-01 9.33242500e-01 2.53318578e-01 -3.70259434e-01 1.08631039e+00 -1.19322866e-01 -3.70507985e-01 9.05647159e-01 4.06027913e-01 1.22994852e+00 1.41597793e-01 5.78539848e-01 -2.07996741e-01 3.68792742e-01 3.70234281e-01 4.93273735e-01 8.20270598e-01 -1.46258429e-01 8.77232373e-01 8.43887627e-01 -2.04724878e-01 -1.00743198e+00 -4.06521857e-01 -9.24662799e-02 1.21441352e+00 -1.88549414e-01 -1.08836949e+00 -9.74669516e-01 -9.95213866e-01 -8.13101307e-02 1.14285743e+00 -5.43271840e-01 -7.13396296e-02 -6.44533277e-01 -7.15723515e-01 8.21139812e-01 3.03649962e-01 4.48746048e-02 -1.13418102e+00 -8.78897786e-01 4.34151292e-01 -5.10834098e-01 -5.09918571e-01 -6.21748388e-01 -1.33571982e-01 -1.01073861e+00 -1.24627125e+00 -4.06176805e-01 -6.55384243e-01 5.45239270e-01 3.82133126e-01 1.48600042e+00 3.35383832e-01 -1.70106646e-02 7.33873785e-01 -1.02173519e+00 -8.36939573e-01 -1.10848451e+00 5.15071571e-01 -4.15657789e-01 -5.89505494e-01 4.48276609e-01 -2.86477894e-01 -2.65996486e-01 -1.77764576e-02 -1.15336776e+00 1.45432383e-01 6.11782968e-01 4.64274406e-01 -2.83067852e-01 -3.03537816e-01 7.33991683e-01 -1.30494106e+00 1.21696258e+00 -7.10739434e-01 2.41371002e-02 5.20287812e-01 -8.26804459e-01 6.86312392e-02 2.46433839e-01 -2.38907069e-01 -1.03312445e+00 -4.17116970e-01 8.27948600e-02 5.74488044e-01 1.94479153e-02 1.14068735e+00 4.21761870e-01 1.06742457e-01 9.53876436e-01 -6.07469231e-02 -8.98216292e-02 -5.67484140e-01 1.54315695e-01 1.06153810e+00 9.94154625e-03 -5.22717297e-01 4.79477137e-01 5.97519660e-03 -8.64688098e-01 -6.50373340e-01 -6.22339666e-01 -6.27045214e-01 -6.60482883e-01 -2.22868145e-01 5.56633472e-01 -7.65531480e-01 3.55205029e-01 4.11829740e-01 -1.57116163e+00 -1.88538179e-01 -1.29330531e-01 3.51624370e-01 -4.74390566e-01 7.00717688e-01 -4.43773657e-01 -6.49857521e-01 -4.80423838e-01 -1.27562296e+00 8.77039552e-01 1.75953969e-01 -1.16045749e+00 -1.03363359e+00 6.25149906e-01 4.28343385e-01 4.38777000e-01 5.03476202e-01 1.06778038e+00 -7.69756436e-01 -1.30102530e-01 -3.70565683e-01 2.69421674e-02 3.37206692e-01 2.79589772e-01 7.74373710e-01 -6.87643647e-01 -4.40910310e-01 1.57480184e-02 -2.77034968e-01 5.58213890e-01 2.32692391e-01 6.21746898e-01 -4.02876824e-01 -3.23926091e-01 6.51882291e-02 1.26966310e+00 2.67090470e-01 3.66142571e-01 7.17555344e-01 6.26773596e-01 7.95198262e-01 7.45806456e-01 7.44213462e-01 6.96243465e-01 5.14417887e-01 2.35402107e-01 2.81581253e-01 -2.60380972e-02 1.44344643e-01 7.16085792e-01 1.43669617e+00 1.64101601e-01 -3.75323117e-01 -1.25351143e+00 8.88934553e-01 -1.95120120e+00 -1.06926227e+00 -5.01128197e-01 2.07595682e+00 1.00189734e+00 3.24703343e-02 1.97120294e-01 -1.16264271e-02 6.15823150e-01 2.22963199e-01 -9.13051590e-02 -9.43198562e-01 2.66128838e-01 -2.69957095e-01 1.41236736e-02 1.92411602e-01 -8.39870095e-01 2.25281045e-01 6.40455294e+00 4.07150388e-01 -9.02768075e-01 1.48013830e-01 1.34523302e-01 5.80474772e-02 -6.55751646e-01 4.49739188e-01 -6.51793361e-01 4.60550845e-01 1.06459785e+00 -8.00710678e-01 -4.31362949e-02 9.20596480e-01 2.10047349e-01 -2.90561944e-01 -1.60960317e+00 6.65239036e-01 5.94684303e-01 -1.15592110e+00 -4.75587584e-02 -2.08960116e-01 1.21753263e+00 1.05544947e-01 -4.40581381e-01 7.18368590e-01 3.24488759e-01 -6.27935588e-01 1.06420660e+00 2.69544959e-01 3.91036898e-01 -4.84769821e-01 1.01537859e+00 4.03594226e-01 -9.40818310e-01 -1.77966535e-01 -1.27212942e-01 -1.99326634e-01 1.22870496e-02 4.72376376e-01 -9.68106031e-01 7.48185396e-01 4.94722545e-01 1.05359352e+00 -1.27053440e+00 1.55033541e+00 5.24189956e-02 6.98754787e-01 4.38224584e-01 -1.73793346e-01 5.78019284e-02 -2.16171164e-02 6.25458956e-01 1.76601577e+00 4.21110004e-01 -5.55440962e-01 2.36428708e-01 7.93733120e-01 5.43139055e-02 3.85130942e-01 -6.17506623e-01 -3.01704735e-01 5.09951055e-01 1.02473891e+00 -7.68172383e-01 -2.94824004e-01 -7.96174288e-01 6.12042546e-01 1.50600418e-01 3.97138357e-01 -6.66214705e-01 -8.56640160e-01 3.64898562e-01 2.60619253e-01 2.53128074e-02 -7.66090006e-02 -4.95318055e-01 -1.12202668e+00 2.94690043e-01 -1.24625313e+00 3.27801019e-01 -6.41584694e-01 -9.56154823e-01 5.99474847e-01 4.01873797e-01 -1.37695014e+00 -3.85162920e-01 1.69372693e-01 -1.02699316e+00 6.26264632e-01 -1.07453597e+00 -8.16489458e-01 -3.69230628e-01 -3.16986442e-01 8.99893403e-01 -8.68969411e-02 6.09178185e-01 4.39032167e-01 -6.32862985e-01 5.18996179e-01 2.40256011e-01 3.04567423e-02 1.18405199e+00 -1.39247882e+00 6.59842372e-01 1.01188898e+00 -2.89864419e-03 1.09991026e+00 1.09576416e+00 -7.99972296e-01 -8.48782599e-01 -1.15470374e+00 1.36544228e+00 -9.28217411e-01 5.77009141e-01 -1.44017681e-01 -1.00265801e+00 7.04526722e-01 6.71524405e-01 -7.52300799e-01 9.72273827e-01 1.96926519e-01 -2.81217396e-01 1.92451045e-01 -7.26159275e-01 5.46836197e-01 5.05796373e-01 -6.11380577e-01 -9.49622393e-01 4.32504743e-01 6.96999907e-01 -2.22715780e-01 -7.85506785e-01 1.41642421e-01 4.77880001e-01 -1.12358892e+00 3.77360612e-01 -4.06855553e-01 8.30377817e-01 -1.48271427e-01 2.07595900e-01 -1.35473621e+00 -1.49118066e-01 -4.93449420e-01 6.80335313e-02 1.92952800e+00 5.11308908e-01 -3.48771036e-01 4.31635290e-01 5.10580242e-01 -4.90109980e-01 -6.94661677e-01 -3.74094933e-01 -5.12779713e-01 1.91238210e-01 -3.57932299e-01 4.30043578e-01 1.07887709e+00 4.03090209e-01 3.20012748e-01 -9.07722209e-03 -3.56803715e-01 2.64333725e-01 2.57948816e-01 1.04000998e+00 -1.44259524e+00 -1.49479568e-01 -8.73837113e-01 -1.80385351e-01 -5.58357716e-01 1.37720302e-01 -1.08543348e+00 6.31413683e-02 -2.07375622e+00 7.10599601e-01 -1.63383842e-01 1.44009471e-01 3.77518475e-01 -3.96101743e-01 -2.12447867e-01 1.65838972e-01 5.55335760e-01 -7.75584698e-01 2.16997087e-01 5.69146693e-01 -2.00346589e-01 -4.28418577e-01 3.10632318e-01 -1.01276052e+00 5.78227282e-01 7.78447092e-01 -8.78412962e-01 -6.04319990e-01 -6.73802793e-01 6.72924876e-01 -4.86723036e-02 -1.64397061e-01 -1.02756810e+00 2.68061846e-01 -8.70008953e-03 -2.15159252e-01 -5.46471894e-01 -6.89697921e-01 -2.88946867e-01 2.57650137e-01 5.14641225e-01 -6.59275711e-01 5.31383216e-01 2.70573676e-01 1.32728562e-01 -3.56532812e-01 -1.00791192e+00 4.21818167e-01 -2.39839658e-01 -3.37966114e-01 -1.78612590e-01 -6.84604406e-01 4.38655317e-01 8.72901678e-01 -5.99139810e-01 -4.51872975e-01 -2.31583059e-01 -6.52932674e-02 2.17913136e-01 9.75308955e-01 6.85866594e-01 2.53117472e-01 -1.13539922e+00 -1.09072697e+00 -2.74252415e-01 6.92672372e-01 -2.70390123e-01 -4.29709032e-02 9.03355420e-01 -6.99063122e-01 3.72345597e-01 -1.40293300e-01 -4.63569492e-01 -1.51305568e+00 1.85960114e-01 -5.06736748e-02 -3.51189286e-01 -3.25848639e-01 5.06307304e-01 -2.21089303e-01 -5.65882981e-01 9.87597108e-02 -3.22292894e-01 -5.61669409e-01 6.68256164e-01 4.70676005e-01 6.79020107e-01 2.99848676e-01 -4.56742108e-01 -2.04610407e-01 2.94164628e-01 -3.76686305e-01 -1.04252789e-02 1.35066080e+00 -1.38719127e-01 -5.55395663e-01 9.92394745e-01 1.30895567e+00 2.00553536e-01 -6.63145185e-01 -6.16744976e-04 6.48568869e-01 -3.37520659e-01 -4.45075512e-01 -5.93774617e-01 -5.28219998e-01 5.80235541e-01 1.43870786e-01 7.66811430e-01 6.93929076e-01 -7.59741366e-02 3.77995551e-01 2.53714502e-01 1.83800444e-01 -1.16516936e+00 1.78991869e-01 6.03864074e-01 1.10975361e+00 -1.19777739e+00 4.28578615e-01 -7.49355853e-02 -7.23140299e-01 1.29219782e+00 6.21075690e-01 3.52347791e-01 1.89471200e-01 1.20792203e-01 6.26105368e-02 -3.37307274e-01 -1.02987742e+00 4.28071976e-01 1.81656539e-01 3.66271198e-01 1.27369189e+00 -1.25048742e-01 -7.46274054e-01 3.56815219e-01 -2.17342272e-01 2.78605130e-02 1.42855012e+00 1.50183451e+00 -3.87571007e-01 -1.31209481e+00 -3.01637560e-01 8.57141078e-01 -6.27159178e-01 -1.42687336e-01 -7.03956723e-01 8.26100290e-01 -1.40376473e-02 1.15124524e+00 -2.09228575e-01 -3.22221965e-01 4.93892133e-01 1.12185009e-01 1.56234741e-01 -1.41663086e+00 -1.18027246e+00 -3.92019719e-01 4.11428362e-01 -1.27381369e-01 -7.24167228e-01 -1.32291627e+00 -1.00504363e+00 -2.72304267e-01 -5.70763588e-01 6.08978808e-01 5.60721040e-01 7.67035246e-01 2.89403409e-01 5.25220454e-01 4.43494022e-01 -7.23386109e-01 -5.63560724e-01 -1.29406559e+00 -2.02758700e-01 5.48500717e-01 3.47479135e-01 -3.05375814e-01 -4.29396629e-01 4.76285726e-01]
[7.654990196228027, 7.937382221221924]
ed3c98a1-40e7-4b8d-b034-bb79d3aec377
amstertime-a-visual-place-recognition
2203.16291
null
https://arxiv.org/abs/2203.16291v2
https://arxiv.org/pdf/2203.16291v2.pdf
AmsterTime: A Visual Place Recognition Benchmark Dataset for Severe Domain Shift
We introduce AmsterTime: a challenging dataset to benchmark visual place recognition (VPR) in presence of a severe domain shift. AmsterTime offers a collection of 2,500 well-curated images matching the same scene from a street view matched to historical archival image data from Amsterdam city. The image pairs capture the same place with different cameras, viewpoints, and appearances. Unlike existing benchmark datasets, AmsterTime is directly crowdsourced in a GIS navigation platform (Mapillary). We evaluate various baselines, including non-learning, supervised and self-supervised methods, pre-trained on different relevant datasets, for both verification and retrieval tasks. Our result credits the best accuracy to the ResNet-101 model pre-trained on the Landmarks dataset for both verification and retrieval tasks by 84% and 24%, respectively. Additionally, a subset of Amsterdam landmarks is collected for feature evaluation in a classification task. Classification labels are further used to extract the visual explanations using Grad-CAM for inspection of the learned similar visuals in a deep metric learning models.
['Jan van Gemert', 'Ronald Maria Siebes', 'Seyran Khademi', 'Burak Yildiz']
2022-03-30
null
null
null
null
['visual-place-recognition']
['computer-vision']
[-1.94574699e-01 -2.00900018e-01 -2.18290334e-05 -8.07460606e-01 -1.16150951e+00 -8.56794596e-01 9.98855531e-01 2.68029302e-01 -6.26621366e-01 3.36437404e-01 3.04019809e-01 -2.42775511e-02 5.07172868e-02 -6.21255696e-01 -9.36233103e-01 -3.66608799e-01 8.15798640e-02 5.96701741e-01 1.33861184e-01 -2.71085322e-01 4.87817198e-01 6.79605067e-01 -1.63964140e+00 2.12769464e-01 5.18093705e-01 1.16895294e+00 2.57336080e-01 4.30244684e-01 2.36691460e-01 5.28312147e-01 -4.55869883e-01 -4.55218047e-01 7.55560219e-01 1.26526088e-01 -6.73244119e-01 -1.16490654e-03 1.39573967e+00 -3.57827872e-01 -7.03420162e-01 8.50523472e-01 4.11360979e-01 4.05603290e-01 8.23571980e-01 -1.60058618e+00 -1.34124732e+00 2.61571128e-02 -4.55007315e-01 1.47372097e-01 7.02378213e-01 1.66562781e-01 1.06415367e+00 -1.50661755e+00 1.00623012e+00 1.01518071e+00 6.14028931e-01 1.25616252e-01 -1.09825659e+00 -6.74366772e-01 -2.50832159e-02 5.93646288e-01 -1.95481646e+00 -6.65377259e-01 5.74531734e-01 -4.48274463e-01 1.09530222e+00 9.09880362e-03 4.02687728e-01 1.09531558e+00 -1.40169233e-01 6.25311315e-01 9.04424489e-01 -1.74206167e-01 3.39679450e-01 3.22512150e-01 -2.52688557e-01 7.79062271e-01 -1.32239088e-01 3.34830105e-01 -8.43859851e-01 3.28098647e-02 3.31246257e-01 2.51387000e-01 -3.96660268e-01 -6.28192127e-01 -1.42319763e+00 6.32952154e-01 1.18205404e+00 1.26660359e-03 -1.66658804e-01 -1.70224622e-01 6.05964735e-02 3.07765812e-01 2.36694068e-01 4.87430543e-01 -4.26682144e-01 3.65647972e-01 -1.07453465e+00 4.03525352e-01 3.17780584e-01 1.39177823e+00 1.11117113e+00 -8.93648043e-02 -2.60552287e-01 6.91983342e-01 6.21647418e-01 9.83799160e-01 4.81242716e-01 -7.61926591e-01 7.27886438e-01 8.16421032e-01 3.06202412e-01 -1.43192101e+00 -3.38949353e-01 -3.57096978e-02 -4.67739642e-01 2.54965216e-01 4.08664197e-01 4.85041678e-01 -1.27273345e+00 1.41931486e+00 1.66421920e-01 2.03331307e-01 7.17009157e-02 1.16524649e+00 1.39425635e+00 5.09424925e-01 -1.49714842e-01 7.80311406e-01 1.13048625e+00 -1.27479625e+00 -1.20860219e-01 -6.86358333e-01 4.16026920e-01 -8.30331981e-01 1.19265139e+00 -1.24543235e-01 -4.86345947e-01 -5.66161036e-01 -1.26109052e+00 -5.28753459e-01 -1.02209425e+00 4.31116313e-01 1.45562172e-01 1.59635618e-01 -1.37122393e+00 4.02593344e-01 -3.37304533e-01 -7.23651528e-01 4.94530261e-01 -4.66461889e-02 -1.12935019e+00 -4.64178771e-01 -8.32854390e-01 1.13011122e+00 -4.29093093e-02 1.26359880e-01 -1.33907223e+00 -8.68596077e-01 -1.40392363e+00 -2.16837183e-01 -3.37483883e-01 -1.51327252e-01 9.49069381e-01 -5.59214592e-01 -7.78841794e-01 1.71778858e+00 -2.06071869e-01 -4.60219562e-01 8.36455226e-01 3.38787101e-02 -6.14246845e-01 1.40744880e-01 7.97615409e-01 1.17348194e+00 7.36111522e-01 -1.27619803e+00 -7.04738975e-01 -6.02155328e-01 2.09522322e-02 1.90597698e-01 1.18032835e-01 -2.35038340e-01 -5.75082600e-01 -4.62883860e-01 3.95382434e-01 -1.11282969e+00 -3.45146582e-02 4.10189182e-01 -5.75045049e-01 1.50149584e-01 6.56535506e-01 -8.71059000e-01 3.26023012e-01 -2.26407290e+00 -1.38288021e-01 2.59423763e-01 6.97669685e-02 -1.13132767e-01 -5.79823494e-01 4.38456655e-01 -1.05555765e-01 -1.23476908e-01 -1.18667632e-01 -7.37134099e-01 1.82904750e-01 9.32290182e-02 -5.54582953e-01 8.53979886e-01 -1.54136179e-03 1.04177797e+00 -1.06501567e+00 -3.61022234e-01 5.41736543e-01 3.36126924e-01 -1.88301653e-01 2.51766711e-01 1.35562837e-01 5.11808336e-01 3.54673155e-02 1.06737983e+00 8.95720124e-01 -1.52565703e-01 -4.31747288e-01 -1.36533946e-01 -5.40929176e-02 1.82210684e-01 -1.05199170e+00 2.19566131e+00 -3.79758596e-01 1.24295855e+00 -5.94234228e-01 -5.52411973e-01 1.13602376e+00 -2.34883010e-01 1.11261625e-02 -1.34480143e+00 -2.12497458e-01 2.38924965e-01 -8.26916933e-01 -3.32664907e-01 9.75076437e-01 5.48115551e-01 -2.30034888e-01 4.21268754e-02 1.78856850e-01 -2.62389593e-02 -5.13127856e-02 2.86406040e-01 1.02532637e+00 1.88068762e-01 -8.58435687e-03 -2.44855762e-01 3.46311182e-01 5.81313193e-01 3.57705683e-01 7.79244542e-01 -4.32118148e-01 1.38163388e+00 -2.88492113e-01 -8.01789165e-01 -1.25949669e+00 -1.35051727e+00 -1.86397523e-01 9.73809659e-01 6.49919808e-01 -3.22394460e-01 -3.15099448e-01 -6.41455591e-01 3.21802229e-01 6.95607662e-01 -9.39750373e-01 -2.83902008e-02 -2.00791851e-01 -6.79787248e-02 5.89071453e-01 6.12420022e-01 8.75881553e-01 -9.93444979e-01 -5.97451091e-01 -3.82176369e-01 -1.25594243e-01 -1.26689792e+00 -4.81920689e-01 -2.05219820e-01 -1.76976725e-01 -1.43253314e+00 -8.38408113e-01 -1.03555429e+00 8.65231216e-01 7.67829180e-01 1.15988517e+00 1.17257677e-01 -2.58022428e-01 6.88505530e-01 -2.68911242e-01 -1.37560666e-01 -2.14638691e-02 3.63837406e-02 2.74141431e-02 5.60766831e-02 6.87164783e-01 -2.80803561e-01 -8.57160568e-01 5.81310034e-01 -4.11590815e-01 -1.98762447e-01 2.02191710e-01 6.88049376e-01 6.75592422e-01 -6.90769374e-01 -8.18120390e-02 -1.28690600e-01 -2.86667999e-02 -5.61148584e-01 -8.38112175e-01 4.75729167e-01 -5.13214350e-01 -1.25930026e-01 2.34257996e-01 -5.77202253e-02 -7.46959925e-01 2.42536917e-01 2.80938476e-01 -6.66528940e-01 -4.07347441e-01 1.84275374e-01 2.43829340e-02 -3.71349990e-01 8.96963537e-01 2.87140697e-01 -3.75944018e-01 -1.89331025e-01 5.71165085e-01 6.07489526e-01 9.45783019e-01 -1.70635715e-01 1.24401915e+00 7.28333056e-01 -2.10796461e-01 -5.26109040e-01 -6.47882700e-01 -6.28019214e-01 -8.43064666e-01 -1.33843094e-01 8.98845851e-01 -1.30290008e+00 -3.43132973e-01 3.45781177e-01 -1.05128419e+00 -2.60651916e-01 -2.84080990e-02 3.26209784e-01 -3.53165925e-01 2.84118112e-02 1.88923460e-02 -3.08562607e-01 -1.40669823e-01 -1.04403698e+00 1.54396105e+00 4.63628799e-01 -6.65427148e-02 -7.94846356e-01 2.05569237e-01 4.49393004e-01 3.71892422e-01 4.10800129e-01 3.25638860e-01 -8.28948677e-01 -8.66001666e-01 -4.86031115e-01 -5.22561967e-01 -3.98605205e-02 -9.62472484e-02 3.79170962e-02 -1.42374194e+00 -3.04414451e-01 -6.72016323e-01 -4.88625735e-01 8.26019168e-01 -9.96303037e-02 8.12166989e-01 5.69533333e-02 -4.34593588e-01 9.12501454e-01 1.51400769e+00 5.62135056e-02 6.86444402e-01 9.46139872e-01 8.07196259e-01 5.35279572e-01 8.28364849e-01 2.26607069e-01 1.02338326e+00 8.25434089e-01 6.85963273e-01 -2.45764211e-01 -1.11246884e-01 -7.69045174e-01 7.61243775e-02 4.85752299e-02 2.83936739e-01 7.32196420e-02 -1.41431260e+00 1.11822367e+00 -1.83345020e+00 -8.00484836e-01 -9.74316746e-02 2.37801480e+00 1.73636228e-01 -3.43631297e-01 -3.50594014e-01 -2.64867693e-01 6.31669104e-01 2.99123347e-01 -5.12097001e-01 2.03162581e-01 -4.65438485e-01 -2.31469750e-01 9.18633044e-01 3.67529243e-01 -1.34334600e+00 1.02661836e+00 6.12589455e+00 3.72933239e-01 -1.03902817e+00 9.20657516e-02 4.57143873e-01 -4.03659455e-02 -1.05355002e-01 1.11584283e-01 -6.28948033e-01 2.59824276e-01 6.62556946e-01 8.61710031e-03 4.97310668e-01 1.18570971e+00 -1.87201262e-01 -1.11584641e-01 -1.45934010e+00 1.40894365e+00 5.02629161e-01 -1.60019577e+00 -2.93242326e-03 1.33871799e-02 1.01256692e+00 7.14774966e-01 2.88573593e-01 3.52500737e-01 1.85894668e-01 -1.16495323e+00 1.13062060e+00 4.76627886e-01 6.84572577e-01 -4.60312068e-01 6.25440300e-01 -4.39862683e-02 -1.32132459e+00 -1.06907420e-01 -6.48034453e-01 1.31207973e-01 8.43808576e-02 -3.95832844e-02 -1.06167066e+00 5.74931741e-01 1.25387526e+00 1.18308914e+00 -1.40623200e+00 1.29674983e+00 -5.63987732e-01 -2.59410173e-01 -2.26323113e-01 2.86136687e-01 3.44408095e-01 4.02697064e-02 4.04947698e-01 8.83696795e-01 3.19210291e-01 -4.09059107e-01 1.43770546e-01 9.62157667e-01 -1.05274357e-01 -7.80847073e-02 -1.09863567e+00 5.19627035e-01 9.77235138e-01 1.27877402e+00 -4.83098865e-01 -3.12585920e-01 -2.63642520e-01 1.09191692e+00 5.63527465e-01 5.20589888e-01 -8.56235743e-01 -3.20086569e-01 9.11197901e-01 4.44921032e-02 1.29395381e-01 -1.74499303e-01 -6.75806254e-02 -1.19125235e+00 2.22009733e-01 -4.46488559e-01 3.25842083e-01 -1.44910264e+00 -1.31138718e+00 7.68694758e-01 -1.97321232e-02 -1.59414995e+00 -1.08516127e-01 -6.83830082e-01 -5.65818071e-01 8.35086644e-01 -1.94959688e+00 -1.43419611e+00 -1.04531515e+00 9.37958777e-01 4.12322074e-01 -4.89716619e-01 8.72459292e-01 4.05133814e-01 -3.28155160e-01 7.45375037e-01 2.39737600e-01 5.95447183e-01 1.09770143e+00 -1.24025917e+00 9.07319248e-01 9.40765321e-01 5.14979064e-01 5.06361425e-01 4.83187199e-01 -3.17960739e-01 -1.29103363e+00 -1.39968216e+00 1.07152700e+00 -1.02411330e+00 4.96808887e-01 -5.29614270e-01 -7.08264351e-01 9.48590517e-01 1.84129074e-01 4.07491177e-01 4.98786062e-01 -2.73754686e-01 -7.50459850e-01 -1.42021358e-01 -1.41759837e+00 6.41731143e-01 1.20809412e+00 -1.00490069e+00 -7.97905564e-01 5.87063253e-01 5.37795186e-01 -8.81402194e-01 -6.64413393e-01 1.21258013e-01 4.08557385e-01 -8.82440448e-01 1.23496711e+00 -4.95661139e-01 5.31025648e-01 -7.38428652e-01 -8.35398078e-01 -1.39701068e+00 -1.71002522e-01 5.10348240e-03 7.57470012e-01 1.21156895e+00 5.65721929e-01 -6.44115806e-01 5.53909063e-01 7.92639792e-01 -7.03922808e-02 2.01117713e-03 -1.07573557e+00 -8.66646469e-01 -2.14673713e-01 -4.05633986e-01 8.60548854e-01 1.16113019e+00 -3.00248891e-01 3.89352292e-02 -1.73071325e-01 5.74303210e-01 7.13727832e-01 1.95022419e-01 1.15362251e+00 -1.13686037e+00 4.17728126e-01 -1.95029512e-01 -1.02187455e+00 -7.07636237e-01 3.38135421e-01 -1.17801476e+00 1.59074441e-01 -1.72182417e+00 -1.26020372e-01 -3.86431932e-01 -2.47386515e-01 6.62052333e-01 3.25889051e-01 6.74861789e-01 2.56497502e-01 4.74939406e-01 -7.10210025e-01 6.84029877e-01 5.95772862e-01 -5.96483171e-01 -6.22337237e-02 -5.97635627e-01 -3.57503235e-01 5.44242620e-01 5.50380468e-01 -5.28441608e-01 -2.69139856e-01 -8.15364242e-01 1.58836007e-01 -3.19880128e-01 8.67767215e-01 -1.16900170e+00 4.12087739e-01 9.10736546e-02 7.69169807e-01 -8.38500917e-01 3.87664080e-01 -1.00521243e+00 3.12951095e-02 2.75573563e-02 -4.73731130e-01 6.16158962e-01 2.00083047e-01 6.46782517e-01 -2.09308922e-01 7.09103122e-02 5.26812673e-01 -9.88467187e-02 -1.45357049e+00 3.50445688e-01 1.30340591e-01 1.38418540e-01 1.09812939e+00 -4.25549746e-01 -8.04269373e-01 -3.07786524e-01 -5.52138209e-01 5.24450779e-01 9.08497751e-01 1.01422846e+00 9.61306512e-01 -1.62567532e+00 -5.88076174e-01 3.53971630e-01 1.11377811e+00 3.00762728e-02 2.12056279e-01 4.43932503e-01 -7.74643421e-01 4.21533614e-01 -4.97450918e-01 -9.33904111e-01 -9.46558952e-01 5.81422985e-01 4.11069036e-01 5.24358869e-01 -7.34189093e-01 7.93924809e-01 -3.31476741e-02 -9.60802138e-01 3.07260960e-01 -2.73475200e-01 -8.76303539e-02 9.91195589e-02 7.74617791e-01 2.61452347e-01 2.87719876e-01 -1.24674559e+00 -9.62233841e-01 7.21034408e-01 1.99375346e-01 -1.91138238e-01 1.20529270e+00 -3.56523782e-01 8.56992975e-02 1.89941004e-01 1.41455388e+00 -1.97860301e-01 -1.31013191e+00 -5.12301326e-01 7.22488314e-02 -8.71966004e-01 -7.56858587e-02 -8.74315321e-01 -1.12520289e+00 7.03751445e-01 9.93537009e-01 -3.47360253e-01 6.49642169e-01 -1.42169546e-03 2.61960804e-01 6.90273583e-01 7.30741143e-01 -1.02239072e+00 4.33729554e-04 4.38677996e-01 1.26085889e+00 -1.88292110e+00 -2.09509254e-01 1.89349934e-01 -7.91770399e-01 8.43241036e-01 7.13380992e-01 -3.81989509e-01 5.51978827e-01 -4.82368708e-01 4.72592711e-01 -2.76111960e-01 -2.47163028e-01 -2.46645391e-01 5.80828965e-01 1.08259761e+00 -1.65569723e-01 6.45334721e-02 6.48570120e-01 4.55099344e-02 -5.11948407e-01 -4.76585895e-01 4.07275915e-01 8.68535340e-01 -1.31752148e-01 -3.51680905e-01 -4.02877957e-01 -8.40786174e-02 4.37134206e-01 -1.65828243e-01 -6.47782862e-01 9.06383693e-01 -8.72263983e-02 9.99655604e-01 2.84227163e-01 -4.73172903e-01 6.09422982e-01 -1.94939837e-01 1.30015135e-01 -4.24278378e-01 -4.46759224e-01 -7.26241291e-01 -9.90829542e-02 -8.44252467e-01 -2.63094276e-01 -5.26373386e-01 -1.07173777e+00 -2.33939737e-01 2.73942858e-01 -1.79512441e-01 9.33164835e-01 9.36913431e-01 5.90706706e-01 -1.54471360e-02 5.68932891e-01 -1.31741905e+00 -7.37811550e-02 -8.04129899e-01 -4.17414546e-01 6.52360022e-01 5.76010644e-01 -8.68578494e-01 -4.66370642e-01 -1.23166271e-01]
[7.670629024505615, -1.835938572883606]
ef0da43c-d607-435d-b58f-7db19936e72b
larnet-lie-algebra-residual-network-for
2103.08147
null
https://arxiv.org/abs/2103.08147v2
https://arxiv.org/pdf/2103.08147v2.pdf
LARNet: Lie Algebra Residual Network for Face Recognition
Face recognition is an important yet challenging problem in computer vision. A major challenge in practical face recognition applications lies in significant variations between profile and frontal faces. Traditional techniques address this challenge either by synthesizing frontal faces or by pose invariant learning. In this paper, we propose a novel method with Lie algebra theory to explore how face rotation in the 3D space affects the deep feature generation process of convolutional neural networks (CNNs). We prove that face rotation in the image space is equivalent to an additive residual component in the feature space of CNNs, which is determined solely by the rotation. Based on this theoretical finding, we further design a Lie Algebraic Residual Network (LARNet) for tackling pose robust face recognition. Our LARNet consists of a residual subnet for decoding rotation information from input face images, and a gating subnet to learn rotation magnitude for controlling the strength of the residual component contributing to the feature learning process. Comprehensive experimental evaluations on both frontal-profile face datasets and general face recognition datasets convincingly demonstrate that our method consistently outperforms the state-of-the-art ones.
['Wei Liu', 'Zhifeng Li', 'Dong-Ming Yan', 'Dihong Gong', 'Xiaohong Jia', 'Xiaolong Yang']
2021-03-15
null
null
null
null
['robust-face-recognition']
['computer-vision']
[ 2.01816037e-01 9.99628305e-02 3.00629139e-02 -6.56663537e-01 -1.21816531e-01 -3.41510415e-01 5.49953341e-01 -1.17134333e+00 2.52148672e-03 1.73055768e-01 1.30011201e-01 -1.45643830e-01 -1.10280201e-01 -5.85653007e-01 -1.00234044e+00 -1.05228782e+00 7.66762793e-02 -9.39824656e-02 -5.82411706e-01 -3.45626563e-01 1.38130710e-01 1.00389147e+00 -1.54636979e+00 9.89424065e-02 1.91843942e-01 1.23271322e+00 -3.29776973e-01 1.44989192e-01 2.14701459e-01 5.00172436e-01 -2.69024462e-01 -2.77466416e-01 7.25845516e-01 -4.59723175e-01 -5.56384265e-01 2.21423507e-01 7.98576176e-01 -1.80374593e-01 -6.36003375e-01 1.01467967e+00 7.04723835e-01 -4.76020575e-03 7.36097157e-01 -1.15358579e+00 -9.28037584e-01 4.06248271e-01 -5.85370243e-01 2.58572120e-02 2.48168752e-01 1.09343901e-02 7.25147069e-01 -1.27623975e+00 5.61348975e-01 1.69779646e+00 5.41904628e-01 9.89849865e-01 -9.57118690e-01 -7.90925682e-01 1.14906661e-01 1.47350490e-01 -1.56421816e+00 -9.06285584e-01 1.10226953e+00 -3.51299018e-01 6.71209991e-01 1.04046270e-01 3.75247598e-01 1.07623339e+00 2.39623889e-01 2.61389017e-01 8.35381567e-01 -2.32873261e-01 -1.29957914e-01 -4.00720328e-01 -1.82136253e-01 1.05899394e+00 1.70779273e-01 1.59576863e-01 -6.55716419e-01 1.93166777e-01 1.21630001e+00 1.22244641e-01 -5.36281943e-01 -5.05035996e-01 -7.46054173e-01 5.89870930e-01 7.87838340e-01 1.20261848e-01 -1.95193321e-01 1.98106319e-01 -8.00897330e-02 3.45927924e-01 1.92300871e-01 2.01562513e-02 -3.89092267e-01 6.09406829e-01 -5.95815599e-01 3.46432962e-02 6.06953204e-01 6.61714196e-01 7.38991916e-01 3.57734531e-01 -1.83958009e-01 6.77225411e-01 7.70138562e-01 4.91324931e-01 3.73328894e-01 -4.99655336e-01 1.96118847e-01 6.38381898e-01 -6.06447458e-01 -1.13873875e+00 -4.00459319e-01 -4.63884264e-01 -9.87360895e-01 7.81962648e-02 3.20680648e-01 -5.22886179e-02 -8.95632386e-01 1.99929619e+00 3.56804281e-01 4.02667791e-01 -1.22794271e-01 9.37944353e-01 8.99637818e-01 2.94480592e-01 -2.35354275e-01 -2.79538095e-01 1.45647144e+00 -5.12027800e-01 -5.63604236e-01 -3.45460474e-02 1.72887340e-01 -5.97288668e-01 5.04145265e-01 9.97690298e-03 -9.23421264e-01 -6.81853712e-01 -1.27044952e+00 -2.12335408e-01 -1.52561948e-01 3.72987300e-01 4.88073081e-01 7.30100572e-01 -1.08301389e+00 5.70385098e-01 -5.34250438e-01 -1.31698206e-01 7.33893812e-01 8.02100003e-01 -8.03031921e-01 -8.55465159e-02 -9.21784997e-01 6.71344578e-01 -9.50297639e-02 6.94302201e-01 -9.20404077e-01 -7.24170744e-01 -1.13617885e+00 7.28150606e-02 1.45707846e-01 -6.64106607e-01 1.03671467e+00 -9.45132434e-01 -1.69950140e+00 9.31047082e-01 -4.11155492e-01 -7.02285720e-03 3.20037901e-01 2.74537485e-02 -3.10043246e-01 7.83489272e-03 -2.32188180e-01 6.27622545e-01 1.53400183e+00 -1.08767176e+00 6.69230521e-02 -9.97196376e-01 -1.55648381e-01 9.00066867e-02 -2.10885227e-01 5.53831086e-02 -3.89752746e-01 -5.22796214e-01 5.35013318e-01 -1.07538855e+00 9.40242037e-02 -1.08198710e-02 -2.53533959e-01 -3.74476343e-01 9.92344618e-01 -4.04145390e-01 6.97022676e-01 -2.25216055e+00 3.63396049e-01 3.30041498e-01 3.71394195e-02 3.62090439e-01 -3.75619799e-01 -1.66472241e-01 -7.03318357e-01 4.79047671e-02 1.24566339e-01 1.86141990e-02 -1.61564082e-01 -1.09204024e-01 -6.03239715e-01 8.54773343e-01 8.32698226e-01 1.10133278e+00 -3.90966594e-01 6.34155869e-02 -5.20552099e-02 1.11121786e+00 -7.46127605e-01 1.13339774e-01 5.78084290e-02 7.47951210e-01 -3.48801136e-01 6.69061899e-01 1.12701285e+00 5.75614758e-02 4.53219801e-01 -5.97990990e-01 1.64241031e-01 2.07990527e-01 -1.15593696e+00 1.34300399e+00 -3.79322410e-01 4.04064655e-01 1.35659650e-01 -1.06954992e+00 1.09173322e+00 3.36053997e-01 3.59746724e-01 -5.98702848e-01 4.37468231e-01 3.15966271e-02 2.95569330e-01 -3.74295950e-01 -1.39712885e-01 -1.91952035e-01 3.25152040e-01 1.80787444e-01 5.01522064e-01 1.17274478e-01 -2.15821356e-01 -4.34360236e-01 7.31142104e-01 1.41550839e-01 5.45762740e-02 -4.59868968e-01 9.59411323e-01 -1.10465872e+00 6.25201762e-01 1.22823656e-01 -2.85883248e-01 5.52428186e-01 7.73958564e-01 -7.34195352e-01 -5.66082954e-01 -7.97382295e-01 -4.22874033e-01 9.46781218e-01 -1.29415348e-01 -2.06508115e-01 -1.05299735e+00 -6.36953592e-01 -7.20780641e-02 -1.86536998e-01 -8.33876729e-01 -5.03152430e-01 -8.61536324e-01 -6.82416141e-01 5.56007266e-01 4.43539679e-01 9.33780611e-01 -1.05357468e+00 -1.58993796e-01 -2.81127930e-01 3.02468538e-01 -1.15486598e+00 -6.01453185e-01 -9.17286128e-02 -6.60595536e-01 -1.29312992e+00 -3.65043044e-01 -9.07230377e-01 1.08184242e+00 5.18017769e-01 6.41416788e-01 1.42614007e-01 -6.11220598e-01 2.31538773e-01 5.68403676e-02 -2.25054592e-01 1.92282185e-01 -8.77202675e-03 3.66062999e-01 5.43921232e-01 4.11045015e-01 -5.01276016e-01 -5.83211601e-01 5.26576161e-01 -9.16122437e-01 -3.34725887e-01 5.42733610e-01 8.70412946e-01 2.77115405e-01 -1.00224495e-01 6.23889029e-01 -5.04025280e-01 2.40711659e-01 -2.31785819e-01 -6.49267495e-01 1.25466481e-01 -9.53380689e-02 4.49301630e-01 4.19676483e-01 -2.60987371e-01 -9.95069206e-01 4.46632326e-01 -2.06792727e-01 -4.94491845e-01 -4.96038329e-03 2.01634377e-01 -7.81873047e-01 -6.11792624e-01 5.91141284e-01 2.46476885e-02 2.41842121e-01 -2.91796774e-01 2.31114239e-01 3.47943336e-01 3.98943782e-01 -5.09471536e-01 1.27329016e+00 5.98426223e-01 6.35338247e-01 -1.14520419e+00 -8.30193758e-01 -7.34076276e-03 -8.56362522e-01 -1.83807418e-01 7.26679444e-01 -1.00815749e+00 -1.21360159e+00 5.64113319e-01 -1.11119938e+00 1.38838172e-01 2.42388368e-01 2.26506740e-01 -4.80851144e-01 -4.05669510e-02 -3.28279406e-01 -6.04857564e-01 -2.34940678e-01 -1.40941775e+00 1.22519398e+00 3.65898728e-01 4.47808295e-01 -6.44530833e-01 -2.02782780e-01 2.50342309e-01 3.38171571e-01 1.44736737e-01 8.00813377e-01 -2.36663163e-01 -7.14145958e-01 -4.34260666e-02 -3.33664298e-01 3.89325857e-01 1.08985275e-01 4.06457484e-02 -1.28523922e+00 -4.26620066e-01 2.71416515e-01 -2.00570822e-01 1.12706268e+00 1.93031445e-01 1.34279311e+00 -4.04438436e-01 -5.28543741e-02 1.07392561e+00 1.21404457e+00 1.02163263e-01 9.21529055e-01 -3.05886090e-01 1.00750160e+00 6.96717143e-01 -1.32693410e-01 8.21576118e-02 1.66134074e-01 8.30231130e-01 4.83240217e-01 2.15227738e-01 -2.24466994e-01 -2.67599374e-01 7.39682138e-01 4.96392965e-01 -3.03491563e-01 3.71619731e-01 -5.45055091e-01 2.17576269e-02 -1.58936417e+00 -1.07438600e+00 3.56752217e-01 2.12430096e+00 6.29610956e-01 -3.66697013e-01 -3.07006449e-01 2.21714720e-01 6.62851632e-01 2.15097070e-01 -4.47870612e-01 -1.97509870e-01 -1.57049775e-01 5.49989760e-01 2.51907080e-01 3.90158981e-01 -1.18531549e+00 1.07979906e+00 5.85917902e+00 3.82132173e-01 -1.64529538e+00 -2.59357661e-01 5.73906541e-01 1.41858920e-01 1.05296448e-01 -2.91538179e-01 -1.11428690e+00 -5.85870855e-02 6.16811812e-01 6.36310205e-02 7.06720650e-01 9.03699696e-01 -2.89575625e-02 6.98890150e-01 -1.34608042e+00 1.19761431e+00 4.81433094e-01 -1.09716868e+00 3.53756875e-01 2.65550017e-01 6.50850892e-01 -1.75039545e-01 6.14375591e-01 1.73251912e-01 -3.30707163e-01 -1.60974848e+00 4.95776743e-01 4.45147663e-01 1.03855217e+00 -8.21777761e-01 3.71625185e-01 -1.22373141e-01 -1.28928530e+00 -1.83335051e-01 -5.74325264e-01 -1.21867254e-01 -5.44421375e-01 3.15625101e-01 -7.16089129e-01 4.02637124e-01 5.03636420e-01 8.63338351e-01 -6.02498949e-01 5.15199900e-01 -4.77237493e-01 2.42310554e-01 -3.25397849e-02 4.48220491e-01 -9.25731659e-02 -1.36934295e-01 3.49601537e-01 7.20370770e-01 2.92087495e-01 2.00392112e-01 -2.80197054e-01 9.51973736e-01 -7.18673706e-01 -5.20792753e-02 -9.90845025e-01 1.49262533e-01 2.59772807e-01 1.53562045e+00 -3.61536026e-01 3.36295933e-01 -2.58599877e-01 6.52692318e-01 3.74987364e-01 4.21677947e-01 -5.45990109e-01 -1.89179093e-01 1.14740467e+00 1.06263891e-01 5.86895943e-01 -2.48186290e-01 1.90528843e-03 -1.34592199e+00 2.49781251e-01 -8.61276090e-01 -1.06583811e-01 -3.07927191e-01 -9.40891266e-01 6.77883565e-01 -2.46277720e-01 -9.90113676e-01 -2.27224588e-01 -1.31734395e+00 -6.84167147e-01 9.37284112e-01 -1.55295157e+00 -1.28008974e+00 -3.75071228e-01 9.36310172e-01 1.74052879e-01 -3.29641372e-01 6.97180688e-01 3.25703114e-01 -9.02641296e-01 9.77847934e-01 -3.68224740e-01 5.11467040e-01 5.08909941e-01 -6.18456364e-01 3.71949852e-01 7.97118008e-01 1.74856082e-01 1.12201118e+00 1.43801942e-01 -2.00225025e-01 -2.13971233e+00 -1.19412339e+00 7.08552301e-01 -4.53669935e-01 1.77672714e-01 -6.04953229e-01 -7.38198698e-01 8.42725277e-01 -1.01928666e-01 5.67049742e-01 5.25088608e-01 -5.52272238e-02 -9.80792403e-01 -3.99901658e-01 -1.13661301e+00 6.22355402e-01 1.39176738e+00 -7.51600385e-01 -3.73337269e-01 9.16290209e-02 3.89447600e-01 -2.98832983e-01 -5.51449895e-01 7.66862750e-01 6.88453794e-01 -7.68953502e-01 1.02126598e+00 -8.47871065e-01 3.40134501e-01 -4.63978082e-01 -1.73695818e-01 -1.27212799e+00 -3.08776557e-01 -7.20519483e-01 -7.02244043e-02 9.99914348e-01 -3.51404548e-02 -6.20272279e-01 6.12660229e-01 2.53579438e-01 1.12429775e-01 -6.92592442e-01 -1.22127903e+00 -5.55235565e-01 2.45301813e-01 -1.75116494e-01 6.78957045e-01 7.47274220e-01 -2.25542709e-01 5.52602828e-01 -1.34043947e-01 2.67249763e-01 6.24598563e-01 -2.37968564e-01 6.46726251e-01 -1.39095187e+00 1.84551761e-01 -4.23539400e-01 -7.41307676e-01 -8.79937172e-01 7.98254192e-01 -1.09604871e+00 -9.88677293e-02 -8.42126608e-01 1.53065220e-01 5.37957959e-02 -1.33517399e-01 6.00979865e-01 9.44072753e-02 5.77467084e-01 2.11226985e-01 -6.84238076e-02 -1.57766476e-01 6.82378173e-01 1.38678372e+00 6.95650503e-02 3.44872996e-02 -1.38237536e-01 -8.99721861e-01 8.09368312e-01 6.23183608e-01 7.81505257e-02 -3.10511261e-01 -5.07026553e-01 1.32450640e-01 -2.62526810e-01 5.85263908e-01 -8.57446969e-01 1.32896706e-01 3.98147590e-02 8.84999871e-01 -6.74806982e-02 2.14119524e-01 -8.28071296e-01 -9.16300192e-02 3.47851992e-01 -1.33163139e-01 -1.37823552e-01 8.95338953e-02 2.14018866e-01 -7.60376528e-02 1.19824961e-01 1.16240239e+00 9.34896097e-02 -3.59871954e-01 7.71412790e-01 1.11056387e-01 -2.24395677e-01 8.09805036e-01 4.00770046e-02 -3.38288873e-01 -2.18576014e-01 -6.17910266e-01 -2.46029243e-01 1.44060478e-01 6.73898637e-01 7.73354590e-01 -1.65543079e+00 -6.77564442e-01 1.06998563e+00 -7.71183372e-02 -9.47383977e-03 1.69652179e-01 5.53870142e-01 -3.87797326e-01 7.66826510e-01 -4.88256425e-01 -4.13514704e-01 -1.14971268e+00 4.46369171e-01 7.32037783e-01 1.98695570e-01 -1.46333039e-01 9.68643010e-01 4.09953892e-01 -5.52057981e-01 2.41348416e-01 -2.89442427e-02 -2.93250978e-01 9.20364112e-02 6.82637036e-01 9.52833965e-02 3.38593721e-01 -1.36600971e+00 -5.30628145e-01 1.00902462e+00 -9.93747041e-02 1.26390439e-02 1.27598000e+00 3.61502826e-01 -5.48743725e-01 -2.60827303e-01 1.70060396e+00 -3.28186721e-01 -1.25912714e+00 -3.35838050e-01 -2.21066281e-01 -3.96147460e-01 9.35932435e-03 -2.45557785e-01 -1.58092344e+00 9.57943738e-01 6.32068455e-01 -2.79118717e-01 1.14415324e+00 1.04678003e-02 1.87198147e-01 5.66906691e-01 2.95323670e-01 -6.38896763e-01 2.13980004e-01 8.16350818e-01 1.36948502e+00 -1.17974007e+00 -1.90859050e-01 -4.89901155e-01 -1.32966816e-01 1.35247910e+00 7.88954139e-01 -3.76408041e-01 1.21049213e+00 -1.05559789e-01 -1.20130023e-02 -1.56847700e-01 -6.69675589e-01 -1.25292957e-01 7.21675694e-01 5.42356014e-01 7.16528654e-01 -1.31713510e-01 -1.13947853e-01 5.08871198e-01 -4.07882959e-01 -2.53177047e-01 2.19442528e-02 5.83085895e-01 -1.38969228e-01 -1.01754022e+00 -4.23846841e-01 3.10726725e-02 -5.86843014e-01 1.68952316e-01 -5.68569124e-01 4.63320225e-01 3.13526303e-01 7.00213253e-01 2.67373145e-01 -4.29029942e-01 2.03923181e-01 2.17632324e-01 1.04390609e+00 -4.83142436e-01 -2.78522581e-01 -3.10333185e-02 -5.83128333e-01 -6.54392719e-01 -2.55980581e-01 -6.01212919e-01 -1.19146466e+00 -1.84132069e-01 -7.29318783e-02 -2.33063236e-01 9.25864577e-01 8.90029848e-01 4.02130604e-01 4.29973513e-01 1.17680955e+00 -1.18129671e+00 -8.24733555e-01 -9.04790938e-01 -5.20513892e-01 2.61560589e-01 5.02507269e-01 -8.59281003e-01 -3.16966146e-01 7.82985240e-02]
[13.201752662658691, 0.5036616921424866]
0f39dcb3-4bad-4eaf-8a59-50b81e44a182
once-is-enough-a-light-weight-cross-attention
2210.05261
null
https://arxiv.org/abs/2210.05261v2
https://arxiv.org/pdf/2210.05261v2.pdf
Once is Enough: A Light-Weight Cross-Attention for Fast Sentence Pair Modeling
Transformer-based models have achieved great success on sentence pair modeling tasks, such as answer selection and natural language inference (NLI). These models generally perform cross-attention over input pairs, leading to prohibitive computational costs. Recent studies propose dual-encoder and late interaction architectures for faster computation. However, the balance between the expressive of cross-attention and computation speedup still needs better coordinated. To this end, this paper introduces a novel paradigm MixEncoder for efficient sentence pair modeling. MixEncoder involves a light-weight cross-attention mechanism. It conducts query encoding only once while modeling the query-candidate interaction in parallel. Extensive experiments conducted on four tasks demonstrate that our MixEncoder can speed up sentence pairing by over 113x while achieving comparable performance as the more expensive cross-attention models.
['Chuanyi Liu', 'Qifan Wang', 'Yulan He', 'Zenglin Xu', 'Cuiyun Gao', 'shiyi qi', 'Yuanhang Yang']
2022-10-11
null
null
null
null
['sentence-pair-modeling', 'answer-selection']
['natural-language-processing', 'natural-language-processing']
[ 1.60324350e-02 -9.84285399e-02 -1.08077556e-01 -7.51039505e-01 -1.15240371e+00 -2.52755284e-01 5.27339935e-01 4.86166090e-01 -7.52187371e-01 5.17404318e-01 2.15771586e-01 -6.62213027e-01 2.57923335e-01 -8.22509408e-01 -8.86912704e-01 -2.36389413e-02 2.77308315e-01 8.06357741e-01 -3.30850249e-03 -4.76884067e-01 8.95361602e-02 -1.84142664e-01 -1.47509921e+00 7.03040063e-01 9.65226412e-01 7.80080080e-01 3.02267432e-01 8.87962461e-01 -3.27231556e-01 1.19165623e+00 -6.86439812e-01 -1.01184964e+00 -9.11138430e-02 -3.02258074e-01 -1.22140479e+00 -7.71523952e-01 5.12137175e-01 -7.53560185e-01 -5.73592067e-01 7.65448809e-01 8.44535053e-01 2.48372793e-01 1.73031315e-01 -1.15878820e+00 -8.77795279e-01 1.06629753e+00 -3.32463235e-01 2.85986215e-01 6.90525055e-01 7.76089430e-02 1.55772901e+00 -1.06558597e+00 1.58428892e-01 1.38848615e+00 5.65687895e-01 4.12552446e-01 -1.41862738e+00 -6.87755346e-01 -5.43783046e-03 5.76424062e-01 -1.28797197e+00 -3.93765956e-01 5.78010201e-01 6.29733503e-02 1.78619969e+00 4.45351362e-01 2.86129385e-01 9.01704490e-01 3.14374954e-01 1.33478367e+00 4.68036771e-01 -4.78124827e-01 -2.86105126e-01 1.25414521e-01 5.00040710e-01 6.84056580e-01 -3.01931858e-01 -1.14708744e-01 -5.56288362e-01 -3.35300833e-01 3.67787778e-01 8.89848918e-02 -2.31515989e-01 2.63208836e-01 -9.70724165e-01 8.88896286e-01 5.74961543e-01 4.65141892e-01 -2.10161731e-01 3.37991536e-01 7.26655900e-01 7.42036760e-01 3.33555430e-01 4.62382883e-01 -5.43412030e-01 -2.25671262e-01 -6.00922048e-01 4.60630268e-01 7.68783450e-01 1.12643731e+00 6.39448285e-01 -3.15136254e-01 -6.49819911e-01 9.60252047e-01 1.42559782e-01 1.14634283e-01 4.83028680e-01 -6.71301603e-01 9.36309278e-01 7.78127432e-01 -2.87145615e-01 -1.01950216e+00 -1.02163233e-01 -3.60625029e-01 -9.91454601e-01 -3.09640676e-01 1.39885202e-01 4.21157200e-03 -3.15343827e-01 1.75263631e+00 1.60917506e-01 -6.37414977e-02 7.43371015e-03 7.16347277e-01 1.15420341e+00 6.08816922e-01 1.59621879e-01 2.14043081e-01 1.53690493e+00 -1.31744671e+00 -7.34362960e-01 -2.44117692e-01 1.03686655e+00 -7.68293381e-01 1.29030585e+00 -1.57513931e-01 -1.60264349e+00 -9.42676306e-01 -8.49461257e-01 -1.00890088e+00 -8.49135518e-02 9.75512937e-02 8.45280826e-01 8.56369436e-02 -1.03912592e+00 6.27912819e-01 -7.24810243e-01 4.89323512e-02 4.40370813e-02 7.35080838e-01 -1.80233330e-01 1.25693372e-02 -1.41888237e+00 8.63673508e-01 2.59735763e-01 9.56379548e-02 -3.37698966e-01 -1.02058220e+00 -8.82876158e-01 6.07670665e-01 9.63126943e-02 -1.16015983e+00 1.70419276e+00 -5.46782315e-01 -1.48421228e+00 7.37073004e-01 -4.61346537e-01 -7.92360008e-01 7.97328949e-02 -5.07334113e-01 -2.28637129e-01 -9.62000862e-02 -5.43204695e-02 9.00611341e-01 5.30266345e-01 -6.31689191e-01 -3.46649230e-01 -2.96260446e-01 5.92477381e-01 3.79442871e-01 -3.97770703e-01 4.53582257e-01 -4.78296012e-01 -5.00762522e-01 -1.76187277e-01 -6.82979703e-01 -1.65303320e-01 -2.93105036e-01 -3.08362395e-01 -7.78963447e-01 5.15626192e-01 -5.58091402e-01 1.52059484e+00 -1.96781111e+00 2.13447943e-01 -5.19729257e-01 4.23798084e-01 4.21832591e-01 -3.52248818e-01 6.36388421e-01 -1.73858821e-01 -5.94182611e-02 -3.82508710e-02 -7.38497853e-01 9.23380703e-02 9.90202501e-02 -4.23235267e-01 -8.96797404e-02 3.78413111e-01 1.51962745e+00 -8.02269220e-01 -5.92041790e-01 -3.35587449e-02 3.03353935e-01 -1.06910455e+00 6.55283928e-01 -2.67025679e-01 -8.43090937e-02 -9.21264514e-02 3.42578828e-01 5.09985566e-01 -4.61048663e-01 1.81937784e-01 -4.61749017e-01 3.44769269e-01 8.81775856e-01 -5.09581983e-01 1.90310407e+00 -8.27364683e-01 7.65082777e-01 -6.73095286e-02 -9.41006899e-01 7.11313426e-01 3.25493932e-01 1.24807477e-01 -9.49615002e-01 2.81392276e-01 9.55915004e-02 1.21424414e-01 -4.74307299e-01 8.40671003e-01 1.88734960e-02 -4.50152829e-02 5.81782341e-01 1.60470933e-01 4.23199404e-03 1.49742663e-01 4.55888003e-01 9.44249809e-01 -1.85301721e-01 3.69434021e-02 -7.85009004e-03 7.24203050e-01 -2.76512921e-01 3.28740746e-01 7.00116217e-01 6.18842319e-02 2.52986968e-01 2.39357203e-01 -3.99152517e-01 -6.21815741e-01 -7.97650874e-01 1.44048139e-01 1.46630418e+00 4.94254418e-02 -8.18268359e-01 -6.51378751e-01 -6.14394963e-01 9.60884914e-02 8.16047251e-01 -4.06718820e-01 -4.66723204e-01 -1.00348556e+00 -4.44558442e-01 6.87311769e-01 9.42607105e-01 4.68834072e-01 -9.28955376e-01 -4.49243098e-01 4.05178875e-01 -5.75082302e-01 -1.03327131e+00 -7.44527042e-01 2.04832390e-01 -8.70446503e-01 -8.35670352e-01 -4.61223871e-01 -1.06910813e+00 5.54851532e-01 3.81074458e-01 1.76030779e+00 3.74047309e-01 -1.66824311e-01 -1.14706241e-01 -2.37030730e-01 -1.98192567e-01 -1.31996647e-01 3.68168801e-01 -1.60212606e-01 -3.01985949e-01 7.90718496e-01 -5.13813317e-01 -5.03167927e-01 1.21516325e-01 -6.60077870e-01 9.89124551e-02 5.04201710e-01 1.51700592e+00 2.77168006e-01 -4.48346198e-01 5.09136617e-01 -7.35010505e-01 9.64590788e-01 -4.64695841e-01 -3.64222825e-01 5.99703848e-01 -3.77648741e-01 3.16359580e-01 6.85442924e-01 -3.49343181e-01 -1.22065902e+00 -4.76287454e-01 -6.03525281e-01 -3.87930572e-01 2.58440524e-01 7.07744122e-01 -1.66590954e-03 1.31250266e-02 3.49596024e-01 1.97015524e-01 -8.06507468e-02 -5.17315567e-01 1.83996230e-01 7.78759181e-01 5.10342062e-01 -5.19855857e-01 3.74380946e-01 -1.60966963e-01 -5.82000792e-01 -3.83335084e-01 -8.77745390e-01 -3.51277173e-01 -2.84596920e-01 3.06877196e-01 7.73897588e-01 -1.03005600e+00 -1.25316179e+00 2.15721160e-01 -1.74095798e+00 -2.66729891e-01 -1.15587607e-01 4.89255279e-01 -2.75039077e-01 2.64484257e-01 -9.81189549e-01 -6.75157905e-01 -8.55376840e-01 -1.26143026e+00 1.24890399e+00 8.10123309e-02 -3.99912268e-01 -1.00792468e+00 -2.50767283e-02 7.57480204e-01 4.38386232e-01 -6.34827077e-01 1.13848591e+00 -6.34416580e-01 -7.20605493e-01 -2.09337533e-01 -4.72172141e-01 9.94780734e-02 -4.29678231e-01 -3.54491144e-01 -8.99516821e-01 -3.17791790e-01 8.22542831e-02 -6.20729983e-01 8.84000599e-01 6.08028695e-02 1.51388490e+00 -1.75278738e-01 -2.60624975e-01 5.51385045e-01 1.28560293e+00 1.49280831e-01 5.53148448e-01 -7.45607167e-02 5.27167976e-01 4.67499703e-01 3.74474585e-01 2.60542125e-01 7.95674443e-01 9.11451280e-01 1.95900202e-01 -9.01207402e-02 -5.19034564e-02 -3.68472517e-01 2.06533447e-01 1.32221735e+00 4.45584357e-01 -4.29194689e-01 -8.95106792e-01 4.90694016e-01 -1.91405725e+00 -1.01703966e+00 -1.63723454e-01 1.89480984e+00 9.62258637e-01 1.47274926e-01 -1.47247136e-01 2.00737372e-01 4.34002966e-01 2.20680274e-02 -3.50305796e-01 -7.59521067e-01 6.10100254e-02 6.06537998e-01 -1.64834852e-03 8.29424143e-01 -7.62054324e-01 8.97442520e-01 6.58514500e+00 8.55073869e-01 -9.04511154e-01 3.05531144e-01 7.23863661e-01 -1.52294829e-01 -5.41638613e-01 -1.41395018e-01 -9.70527887e-01 3.47823381e-01 1.14086199e+00 -4.27423209e-01 2.94104397e-01 6.30850732e-01 -4.17018205e-01 -5.68174981e-02 -1.37229621e+00 1.23581922e+00 1.73036858e-01 -1.34579265e+00 -8.97216126e-02 -3.75111789e-01 2.91673154e-01 1.12757549e-01 -1.37877226e-01 9.54980195e-01 2.29373962e-01 -8.64094675e-01 2.53478348e-01 4.37895894e-01 6.42522633e-01 -8.30529332e-01 8.13903928e-01 4.68791097e-01 -1.25565743e+00 -2.24660024e-01 -2.99541861e-01 -4.25319195e-01 3.26248169e-01 3.94153535e-01 -6.15988791e-01 5.44668257e-01 5.50898671e-01 3.52137178e-01 -5.14985859e-01 6.86737180e-01 -4.75793220e-02 4.82144535e-01 -2.95641810e-01 -3.34023893e-01 8.64543766e-02 -9.22229420e-03 1.51748389e-01 1.15158010e+00 -4.62572835e-02 1.90862030e-01 8.02447200e-02 7.97694266e-01 -4.70144391e-01 1.93914637e-01 -5.50239623e-01 -2.77968943e-02 5.17068505e-01 9.09505844e-01 9.60252434e-02 -5.93869746e-01 -5.06830871e-01 1.33416557e+00 8.40435445e-01 2.35896900e-01 -1.00213170e+00 -6.81654274e-01 7.48591959e-01 -2.08494738e-01 4.75761555e-02 -2.13971138e-01 -4.94158506e-01 -1.46286547e+00 3.05299401e-01 -9.09565747e-01 3.72565925e-01 -5.84632993e-01 -1.17615438e+00 8.38511467e-01 -4.07351553e-02 -8.47929418e-01 -5.76503336e-01 -2.91984826e-01 -7.86818624e-01 1.03500462e+00 -1.28771770e+00 -1.14971268e+00 -1.16877057e-01 3.71963024e-01 7.39058256e-01 3.49757411e-02 1.08655679e+00 8.36918175e-01 -6.47880435e-01 1.33750367e+00 -1.02690764e-01 1.28999576e-01 5.06344140e-01 -1.21824706e+00 8.92184973e-01 4.13408488e-01 3.50045413e-01 9.80582952e-01 2.71197557e-01 -1.94649965e-01 -1.53797269e+00 -8.38210702e-01 1.91914415e+00 -3.75161260e-01 4.94188458e-01 -5.61530888e-01 -1.10691631e+00 6.42702639e-01 7.30073333e-01 -2.46719986e-01 9.43239450e-01 5.77168763e-01 -4.11098182e-01 -1.80370986e-01 -7.53375828e-01 6.83018625e-01 8.25465739e-01 -1.21212935e+00 -6.38478220e-01 4.52219009e-01 1.16615343e+00 -4.79471207e-01 -9.19943154e-01 3.33738297e-01 4.37429041e-01 -9.38671291e-01 1.14697111e+00 -8.68820667e-01 6.29534841e-01 1.89666748e-01 -1.66983046e-02 -1.01295900e+00 -4.16142941e-01 -5.50631642e-01 -6.19904220e-01 1.26100254e+00 4.86822397e-01 -2.99711287e-01 8.17822397e-01 9.03773308e-01 -7.74469227e-02 -1.45721698e+00 -6.47439003e-01 -5.28393507e-01 1.62910387e-01 -4.99232978e-01 9.25711870e-01 6.97686136e-01 2.13259473e-01 1.09507644e+00 -2.71338671e-01 -3.01694125e-01 2.51753390e-01 2.62846202e-01 7.31715083e-01 -9.05514598e-01 -7.97266364e-01 -5.97317636e-01 -4.05335613e-02 -1.69282687e+00 5.14642537e-01 -1.08986974e+00 -1.55443296e-01 -1.34808505e+00 2.41392821e-01 -1.46913111e-01 -5.67749478e-02 3.61245424e-01 -7.48816550e-01 1.56012201e-03 1.61257803e-01 -1.74342945e-01 -5.79942405e-01 9.71499562e-01 9.29027736e-01 -4.19805378e-01 6.63331384e-03 -1.80952132e-01 -6.60507739e-01 4.05379772e-01 8.27939928e-01 -2.63542742e-01 -5.14771521e-01 -1.41039276e+00 3.36842388e-01 3.32341284e-01 1.98133126e-01 -8.52823615e-01 7.06701279e-01 3.23515803e-01 -2.76164431e-02 -9.15140152e-01 5.28199434e-01 -5.76032102e-01 -3.17418396e-01 3.72313797e-01 -8.46511006e-01 6.72760487e-01 2.42008463e-01 3.09871882e-01 -7.00444162e-01 -3.02807122e-01 3.99920821e-01 2.83937948e-03 -2.52844989e-01 2.01951802e-01 -1.27071947e-01 2.19283402e-01 5.99329948e-01 1.82383135e-01 -2.38443036e-02 -4.65938598e-01 -3.75648648e-01 5.16894221e-01 -1.73968315e-01 4.78454679e-01 8.43445003e-01 -1.37351251e+00 -8.42279613e-01 2.61008769e-01 4.07635495e-02 4.54710890e-03 4.26246047e-01 8.03518474e-01 -3.16703558e-01 8.03009391e-01 2.19464332e-01 -3.55551302e-01 -1.50847304e+00 5.44556141e-01 2.92108327e-01 -7.26158738e-01 -2.61474073e-01 1.52028453e+00 8.19345340e-02 -6.73511744e-01 4.99167234e-01 -2.94363290e-01 3.44030857e-02 -7.98889920e-02 6.58089757e-01 2.56149679e-01 2.55566657e-01 -2.24251732e-01 -3.37453544e-01 1.02709860e-01 -6.67366326e-01 2.00175002e-01 9.82827604e-01 3.34374756e-02 -3.34508449e-01 9.70265269e-02 1.80807471e+00 -4.93718565e-01 -3.67642015e-01 -5.33167064e-01 1.00292958e-01 -4.52440560e-01 1.73123717e-01 -5.48402727e-01 -8.58351052e-01 1.19068170e+00 1.33051366e-01 -2.54995376e-02 1.14190996e+00 -2.47529969e-01 1.57522058e+00 7.72014558e-01 2.13680193e-01 -7.94960558e-01 9.57984254e-02 7.86921203e-01 9.94966865e-01 -1.46105087e+00 -4.69096869e-01 -3.54340643e-01 -3.48763674e-01 8.64849091e-01 1.10927057e+00 -7.60026425e-02 5.59360802e-01 4.24838006e-01 -2.05888391e-01 -1.13375261e-01 -1.47549212e+00 -4.08197232e-02 3.10007304e-01 5.73766641e-02 1.06178308e+00 -1.09189145e-01 -4.25984681e-01 6.60037994e-01 -1.69330597e-01 6.12563603e-02 -2.36788183e-01 8.82874966e-01 1.06546670e-01 -1.46457279e+00 1.20711066e-01 6.75415337e-01 -4.78781283e-01 -6.61605060e-01 -2.67525315e-01 2.33701646e-01 -1.11500159e-01 1.06425190e+00 3.60240906e-01 -6.78219378e-01 3.87775868e-01 6.21267557e-02 5.26395619e-01 -4.96180445e-01 -1.44068003e+00 -4.97209787e-01 3.14001381e-01 -5.57817042e-01 -7.66224042e-02 -2.53891259e-01 -1.12083542e+00 -4.78458136e-01 -6.81175411e-01 4.85644013e-01 3.24486703e-01 8.27850521e-01 9.09429252e-01 6.50620461e-01 5.29411256e-01 -4.14482355e-01 -1.00752485e+00 -1.04890847e+00 1.69223547e-01 2.79309988e-01 2.26478904e-01 -2.71025777e-01 7.79627413e-02 -3.42460632e-01]
[11.239624977111816, 8.314621925354004]
2e138576-666c-4d92-b985-37364106875a
sketch-less-for-more-on-the-fly-fine-grained-1
null
null
http://openaccess.thecvf.com/content_CVPR_2020/html/Bhunia_Sketch_Less_for_More_On-the-Fly_Fine-Grained_Sketch-Based_Image_Retrieval_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Bhunia_Sketch_Less_for_More_On-the-Fly_Fine-Grained_Sketch-Based_Image_Retrieval_CVPR_2020_paper.pdf
Sketch Less for More: On-the-Fly Fine-Grained Sketch-Based Image Retrieval
Fine-grained sketch-based image retrieval (FG-SBIR) addresses the problem of retrieving a particular photo instance given a user's query sketch. Its widespread applicability is however hindered by the fact that drawing a sketch takes time, and most people struggle to draw a complete and faithful sketch. In this paper, we reformulate the conventional FG-SBIR framework to tackle these challenges, with the ultimate goal of retrieving the target photo with the least number of strokes possible. We further propose an on-the-fly design that starts retrieving as soon as the user starts drawing. To accomplish this, we devise a reinforcement learning based cross-modal retrieval framework that directly optimizes rank of the ground-truth photo over a complete sketch drawing episode. Additionally, we introduce a novel reward scheme that circumvents the problems related to irrelevant sketch strokes, and thus provides us with a more consistent rank list during the retrieval. We achieve superior early-retrieval efficiency over state-of-the-art methods and alternative baselines on two publicly available fine-grained sketch retrieval datasets.
[' Yi-Zhe Song', ' Tao Xiang', ' Timothy M. Hospedales', ' Yongxin Yang', 'Ayan Kumar Bhunia']
2020-06-01
null
null
null
cvpr-2020-6
['sketch-based-image-retrieval', 'on-the-fly-sketch-based-image-retrieval']
['computer-vision', 'computer-vision']
[ 1.38438940e-01 -6.20708466e-01 -3.64417017e-01 -1.38808802e-01 -1.37134206e+00 -7.90986478e-01 7.90322959e-01 -7.49179944e-02 -3.09913665e-01 4.70492095e-01 2.64421642e-01 2.80230314e-01 -4.59008545e-01 -7.85730064e-01 -5.36022425e-01 -5.31275272e-01 3.98496091e-01 5.83254755e-01 4.79805395e-02 -3.56492698e-02 7.63530314e-01 8.54409993e-01 -1.54039240e+00 3.23897123e-01 4.34094578e-01 9.53784287e-01 3.91836226e-01 6.59878969e-01 -1.12579256e-01 4.67052430e-01 -6.79747999e-01 -5.83339095e-01 3.02754313e-01 -1.49567306e-01 -6.98743641e-01 -1.11542270e-01 1.08085191e+00 -1.01833880e+00 -7.58864462e-01 6.69397473e-01 7.03837335e-01 2.87069112e-01 8.94674182e-01 -1.16191339e+00 -1.01640248e+00 1.35781392e-01 -5.85438371e-01 -1.74876869e-01 5.43697536e-01 -6.73272088e-02 1.34582865e+00 -1.06717682e+00 9.75341380e-01 1.37484479e+00 4.44513820e-02 7.96100378e-01 -1.20404160e+00 -7.38912880e-01 3.94300222e-01 -1.42552541e-04 -1.67252159e+00 -4.36294079e-01 9.97958481e-01 -1.74408257e-01 5.05987585e-01 2.64188826e-01 6.24459028e-01 1.06087875e+00 -4.59949970e-01 1.38794804e+00 7.05909431e-01 -3.73555779e-01 2.21653998e-01 -3.55653256e-01 -1.39013529e-01 6.65221632e-01 -7.55944178e-02 1.39095202e-01 -8.57819378e-01 -3.21823537e-01 1.22125304e+00 5.10542452e-01 -2.48449594e-02 -7.90518343e-01 -1.19072425e+00 6.83478117e-01 5.13325810e-01 3.67494255e-01 -4.76827949e-01 7.13970542e-01 2.09542990e-01 2.41106465e-01 1.59774184e-01 4.74629730e-01 1.03124432e-01 -6.53927624e-02 -1.43628681e+00 6.43771470e-01 4.67771500e-01 1.07565534e+00 6.97150946e-01 -3.88585627e-01 -7.30508268e-01 1.13307369e+00 1.62149280e-01 6.68454409e-01 1.41035775e-02 -1.24180591e+00 2.96768427e-01 3.09360445e-01 4.67702836e-01 -9.41703320e-01 5.80821276e-01 -4.11148258e-02 -6.79656446e-01 1.48469657e-01 2.19968185e-01 5.65519631e-01 -7.86516309e-01 1.64287722e+00 -7.60631263e-02 -2.66863592e-02 -2.65561849e-01 1.17011273e+00 7.02143967e-01 5.91575384e-01 2.43322983e-01 2.87289947e-01 1.13780916e+00 -1.11310327e+00 -4.82897639e-01 -5.92446029e-02 -3.01964343e-01 -8.28303576e-01 1.29661691e+00 4.36442822e-01 -1.22898078e+00 -5.32544076e-01 -1.04737914e+00 -3.89817953e-01 -3.17535341e-01 5.37269711e-01 6.04528069e-01 2.30390728e-01 -1.05182576e+00 8.41662109e-01 -3.22243333e-01 -2.56844580e-01 5.75130403e-01 1.66472733e-01 -3.93964946e-01 -5.98141313e-01 -7.95446098e-01 6.44977748e-01 -1.23072669e-01 6.81708306e-02 -1.04165208e+00 -6.60188913e-01 -3.57281327e-01 2.28547052e-01 4.53283578e-01 -9.29448664e-01 1.27220500e+00 -7.05338418e-01 -1.48298061e+00 8.95192742e-01 -2.24356845e-01 1.62599921e-01 7.53719747e-01 -5.75751364e-01 1.37969539e-01 4.67711508e-01 1.73039883e-01 1.11228311e+00 1.36375904e+00 -1.39902282e+00 -2.43868887e-01 -3.18326384e-01 2.18976066e-01 2.42001995e-01 -2.46556774e-01 -2.32678711e-01 -1.22444296e+00 -9.08487678e-01 -1.70721039e-01 -7.94162214e-01 -7.92528875e-03 6.87286735e-01 -1.22667357e-01 -7.00021684e-01 5.70893526e-01 -3.15879792e-01 1.25692642e+00 -2.20520997e+00 3.55161071e-01 2.45149732e-01 1.19184360e-01 2.38438278e-01 -7.28740752e-01 9.36994493e-01 3.53482097e-01 2.30107568e-02 1.68780059e-01 -4.36973453e-01 2.96720535e-01 1.15395352e-01 -6.96824372e-01 9.48354304e-02 2.87643254e-01 1.27342439e+00 -1.14781380e+00 -6.77197039e-01 2.44779021e-01 5.13011038e-01 -4.20447558e-01 4.70658809e-01 -2.53910959e-01 -1.67969577e-02 -8.56261134e-01 7.80572474e-01 5.28604746e-01 -3.38670313e-01 5.69609888e-02 -3.57765675e-01 2.11377442e-01 -2.72491127e-01 -1.07939947e+00 2.42075658e+00 -4.75836545e-01 4.88935441e-01 -1.26766995e-01 -4.33491200e-01 9.01335835e-01 -2.65590269e-02 2.87956268e-01 -1.04431486e+00 -3.33573371e-01 1.65763840e-01 -8.15212250e-01 -1.68693975e-01 8.86033714e-01 -5.47446217e-03 -1.43264502e-01 7.20669031e-01 -1.77218601e-01 -3.82239699e-01 1.81887984e-01 6.24760866e-01 9.65345263e-01 4.93742645e-01 7.69161060e-02 4.98782732e-02 3.11655581e-01 -3.33354712e-01 -1.24841578e-01 1.18564200e+00 3.86997238e-02 8.02633762e-01 1.86794817e-01 -4.90520239e-01 -1.13394320e+00 -1.36785030e+00 3.16565901e-01 1.27044094e+00 4.60225761e-01 -4.00159389e-01 -4.31011885e-01 -4.78708148e-01 3.51181448e-01 2.60561705e-01 -5.86079359e-01 2.83488967e-02 -5.33048391e-01 2.79184282e-01 4.49441403e-01 5.02867401e-01 3.66359562e-01 -1.39216650e+00 -6.27693832e-01 5.96397780e-02 -1.94753572e-01 -6.52691960e-01 -1.00958014e+00 -5.10656774e-01 -7.47600913e-01 -1.04616499e+00 -1.48011410e+00 -7.55239546e-01 8.26305926e-01 7.04852700e-01 1.44169605e+00 5.29189765e-01 -6.65728509e-01 8.14892650e-01 -2.65136540e-01 1.95442915e-01 1.38815716e-01 4.93560135e-02 -4.29429352e-01 -2.17858195e-01 5.85562661e-02 -1.84946105e-01 -1.14480329e+00 1.85374573e-01 -1.08260489e+00 -1.15701616e-01 8.90749991e-01 1.02116764e+00 8.06100309e-01 -3.61145645e-01 7.51156926e-01 -5.71931541e-01 9.95571375e-01 1.00815058e-01 -4.45835054e-01 8.34104061e-01 -5.26733696e-01 3.97604108e-01 2.93127507e-01 -5.06664097e-01 -7.74192512e-01 4.48103696e-02 1.54376373e-01 -9.52791095e-01 1.65579811e-01 3.71465385e-02 1.41402110e-01 -1.37936948e-02 3.29160988e-01 3.37021828e-01 -2.73639143e-01 -5.41170120e-01 8.22830737e-01 5.18033326e-01 5.24217069e-01 -1.04473341e+00 7.74481654e-01 3.80828679e-01 -1.57966062e-01 -6.54282331e-01 -6.68265522e-01 -6.82943702e-01 -5.02503753e-01 -3.26567650e-01 4.46382850e-01 -8.64899933e-01 -1.12526977e+00 2.35123411e-01 -1.16440105e+00 -3.29306483e-01 -2.24445000e-01 -1.60103187e-01 -6.35884345e-01 5.49508214e-01 -4.76579010e-01 -1.03634167e+00 -6.85371280e-01 -9.43600893e-01 1.93268394e+00 6.08798154e-02 -1.73384011e-01 -3.21838945e-01 1.53228238e-01 9.68350917e-02 4.19705451e-01 -1.17445752e-01 9.49228764e-01 1.02371715e-01 -1.28485692e+00 -3.47768873e-01 -7.52448440e-01 -9.49217007e-02 -7.68361688e-02 7.32647628e-02 -8.10070932e-01 -5.27629375e-01 -9.87294674e-01 -6.70058072e-01 1.08833790e+00 1.22727588e-01 1.31468916e+00 -1.59301862e-01 -3.04047108e-01 1.32216871e-01 1.67047930e+00 -7.83421919e-02 6.75038517e-01 1.35663571e-02 3.87473881e-01 3.90440166e-01 9.04064953e-01 4.18916911e-01 1.59185037e-01 8.61740768e-01 1.28935769e-01 4.03120741e-02 -3.27619553e-01 -8.42780292e-01 -2.29745328e-01 2.32622892e-01 4.25975174e-02 -3.01781505e-01 -2.05536470e-01 7.11066067e-01 -1.92684805e+00 -1.11672819e+00 7.91274548e-01 2.36784101e+00 7.00018048e-01 -2.33452588e-01 8.86611566e-02 -9.64815840e-02 5.24560034e-01 4.73183364e-01 -6.76122129e-01 2.26681959e-03 1.32425666e-01 4.14296120e-01 4.37926911e-02 4.35654581e-01 -8.86003077e-01 1.21325123e+00 6.12740850e+00 1.16779792e+00 -8.99992108e-01 -4.33757424e-01 4.07270253e-01 -3.38261575e-01 -4.85589087e-01 -1.69616938e-01 -4.64794278e-01 1.20941892e-01 1.16856359e-01 -2.68056169e-02 1.00387061e+00 7.77408540e-01 -2.28232920e-01 -1.45940885e-01 -1.42504585e+00 1.48834252e+00 2.23314404e-01 -1.38405967e+00 6.14789546e-01 -2.10535720e-01 6.23325408e-01 -3.60514224e-01 1.34848565e-01 2.86954701e-01 1.55868351e-01 -9.49770987e-01 8.25125158e-01 9.27073598e-01 1.34279215e+00 -7.54460752e-01 -4.69230749e-02 5.91675453e-02 -1.21368682e+00 6.10522889e-02 -3.93041223e-01 3.21367204e-01 -1.19556285e-01 1.49878696e-01 -4.66819316e-01 5.29255927e-01 4.43394363e-01 6.30842566e-01 -5.45752883e-01 9.62939978e-01 -1.59323081e-01 -2.40244702e-01 -6.13883473e-02 -3.36274087e-01 2.63113528e-01 -1.48238614e-01 1.73956051e-01 1.05142963e+00 2.30167165e-01 2.04798415e-01 2.24060923e-01 8.64597797e-01 -4.78567094e-01 1.13797329e-01 -6.81177676e-01 -3.34531844e-01 7.96608269e-01 1.24329686e+00 -5.68939447e-01 -4.83261168e-01 -3.40606384e-02 1.52277672e+00 6.72494888e-01 5.40311694e-01 -2.96625167e-01 -4.95269120e-01 4.19298857e-01 -1.04950577e-01 4.18697447e-01 -1.14229783e-01 2.30339676e-01 -1.06440890e+00 1.45609692e-01 -7.77694583e-01 3.24942946e-01 -1.19841588e+00 -1.69726455e+00 4.33192044e-01 -2.13525102e-01 -1.13152921e+00 -3.38111490e-01 -2.81154394e-01 -3.86537522e-01 9.13524210e-01 -1.84799910e+00 -1.40771794e+00 -4.61274445e-01 6.60737038e-01 8.02913427e-01 7.41379112e-02 9.69476342e-01 4.58592474e-01 -7.34305307e-02 6.61666453e-01 6.99573709e-03 1.14935316e-01 1.17063117e+00 -1.15559781e+00 5.12614489e-01 3.10473382e-01 4.00046855e-01 8.93104672e-01 2.16432527e-01 -6.16345882e-01 -1.76822662e+00 -7.06692517e-01 7.52975523e-01 -3.94863605e-01 3.44399184e-01 -2.44160488e-01 -5.64523518e-01 8.16695094e-02 -7.50466541e-04 3.34081426e-02 1.60719499e-01 9.89989191e-02 -6.86850488e-01 -2.87372768e-01 -9.49185610e-01 9.15205956e-01 1.13486516e+00 -1.15443885e+00 -4.51930970e-01 1.56050220e-01 2.58464009e-01 -1.80368438e-01 -6.57721817e-01 -9.86434519e-02 1.34832132e+00 -6.45038962e-01 1.49462223e+00 -5.29006600e-01 5.75185835e-01 -2.09856495e-01 -2.24022776e-01 -9.43606317e-01 -3.56879026e-01 -5.97731233e-01 -6.33894131e-02 9.69013870e-01 -1.67033151e-01 6.39360920e-02 9.64451134e-01 6.46295428e-01 5.03654540e-01 -7.79237628e-01 -5.71583509e-01 -6.63402796e-01 -1.18293792e-01 5.53087005e-03 6.64222777e-01 3.89460444e-01 -2.77755409e-01 2.47082055e-01 -6.09566033e-01 -2.00884983e-01 8.08405221e-01 7.28076279e-01 9.24294174e-01 -1.23084223e+00 -1.70870274e-01 -6.66922390e-01 -6.25865385e-02 -1.51741838e+00 1.88517094e-01 -6.09258115e-01 1.87614411e-01 -1.68368363e+00 5.69494486e-01 -6.08239591e-01 -4.13248569e-01 4.14952070e-01 -2.94250131e-01 5.22848845e-01 6.01163268e-01 5.77939272e-01 -1.07287323e+00 5.71884692e-01 1.47825170e+00 -3.74347836e-01 4.44058366e-02 -2.76747763e-01 -6.18667185e-01 -1.23311067e-02 2.21287325e-01 -2.63459951e-01 -5.33557117e-01 -4.89213705e-01 3.02690476e-01 5.03622949e-01 6.32259727e-01 -4.65603262e-01 3.98492396e-01 -4.68953885e-02 6.55737817e-01 -9.92056429e-01 6.55095041e-01 -7.39929497e-01 7.05498978e-02 1.70251608e-01 -6.95275545e-01 1.90137886e-02 -1.02056183e-01 8.19314778e-01 -1.38635904e-01 -2.89155155e-01 4.55399990e-01 -2.61013329e-01 -6.75389707e-01 5.84575772e-01 6.04824275e-02 -7.20161125e-02 6.10575736e-01 5.35479300e-02 -3.15722108e-01 -5.12453794e-01 -4.20940787e-01 1.28788009e-01 6.83360875e-01 6.26083910e-01 8.81630719e-01 -1.66926396e+00 -5.65377712e-01 2.85522770e-02 4.31152821e-01 -3.82645547e-01 3.67720515e-01 -6.10293597e-02 -2.62955278e-01 6.30371630e-01 -2.65219122e-01 -2.33199567e-01 -1.23315048e+00 8.37084234e-01 -8.01560730e-02 -3.92102093e-01 -6.05802655e-01 6.60895348e-01 2.24853024e-01 -1.16360784e-02 5.01077294e-01 1.02897234e-01 1.97592780e-01 9.03363079e-02 5.98166764e-01 3.84414732e-01 -1.05281584e-01 -1.05754241e-01 -2.17413679e-01 9.21090603e-01 -2.53552049e-01 -3.44172448e-01 1.25620162e+00 1.23545132e-03 1.40562400e-01 1.60510927e-01 1.19138455e+00 2.98568401e-02 -1.46384549e+00 -2.90921360e-01 -1.50030494e-01 -9.84298646e-01 5.29396487e-03 -1.15341783e+00 -8.84367347e-01 8.73442233e-01 5.22114515e-01 -2.09081575e-01 9.83918071e-01 -1.36222709e-02 8.34373593e-01 8.57451081e-01 5.86873293e-01 -9.51113462e-01 6.30818546e-01 -8.48082975e-02 1.29765797e+00 -1.12996125e+00 3.03059012e-01 6.22347221e-02 -5.12928009e-01 1.19767165e+00 2.86849380e-01 -5.38596094e-01 2.11096808e-01 -3.00077289e-01 -1.48503304e-01 -4.04208362e-01 -6.86214805e-01 -1.36532202e-01 6.10452473e-01 2.47609615e-01 2.43005797e-01 2.99026426e-02 -9.05764699e-02 1.03074670e-01 4.57428932e-01 3.25945020e-01 -1.37509748e-01 9.50114489e-01 -2.29994863e-01 -1.45148933e+00 -1.08116321e-01 4.40341860e-01 -1.08467750e-01 -1.83772400e-01 -7.15247631e-01 5.53339303e-01 -6.32950962e-01 5.97761095e-01 -8.02273154e-02 1.73266426e-01 4.53213692e-01 4.56858464e-02 1.02134907e+00 -2.91798979e-01 -3.56655300e-01 1.39731288e-01 -2.52630204e-01 -7.54681170e-01 -3.69038820e-01 -3.29611659e-01 -6.68196380e-01 -2.70613134e-01 -2.79968411e-01 6.27630949e-02 6.22302055e-01 5.37698448e-01 5.38739562e-01 9.72194970e-02 6.31753623e-01 -1.33470559e+00 -5.63648105e-01 -6.10241354e-01 -5.38416982e-01 5.85630715e-01 4.56463248e-01 -6.64729476e-01 3.66546996e-02 -1.42381757e-01]
[11.651995658874512, 0.5691048502922058]
5d6fa8cf-ce35-4552-9d3b-95c68aeacff6
generalization-bounds-for-neural-belief
2305.1054
null
https://arxiv.org/abs/2305.10540v1
https://arxiv.org/pdf/2305.10540v1.pdf
Generalization Bounds for Neural Belief Propagation Decoders
Machine learning based approaches are being increasingly used for designing decoders for next generation communication systems. One widely used framework is neural belief propagation (NBP), which unfolds the belief propagation (BP) iterations into a deep neural network and the parameters are trained in a data-driven manner. NBP decoders have been shown to improve upon classical decoding algorithms. In this paper, we investigate the generalization capabilities of NBP decoders. Specifically, the generalization gap of a decoder is the difference between empirical and expected bit-error-rate(s). We present new theoretical results which bound this gap and show the dependence on the decoder complexity, in terms of code parameters (blocklength, message length, variable/check node degrees), decoding iterations, and the training dataset size. Results are presented for both regular and irregular parity-check matrices. To the best of our knowledge, this is the first set of theoretical results on generalization performance of neural network based decoders. We present experimental results to show the dependence of generalization gap on the training dataset size, and decoding iterations for different codes.
['Tamal Bose', 'Bane Vasic', 'Ravi Tandon', 'Xin Xiao', 'Sudarshan Adiga']
2023-05-17
null
null
null
null
['generalization-bounds']
['methodology']
[ 3.24572802e-01 2.28051528e-01 -4.84307736e-01 -1.41203299e-01 -5.49149394e-01 -8.64949003e-02 3.99106413e-01 3.81226212e-01 -1.54523045e-01 7.30135143e-01 -8.00931305e-02 -9.47329164e-01 9.23124179e-02 -7.09600806e-01 -7.07895815e-01 -8.71450007e-01 -6.06031299e-01 3.22509766e-01 3.36630106e-01 -3.56999516e-01 4.84464079e-01 2.91503310e-01 -1.12967205e+00 6.87142322e-03 7.29951978e-01 1.01874208e+00 1.45121142e-01 1.14397562e+00 4.15777802e-01 5.93451619e-01 -3.81127328e-01 -6.51725650e-01 7.19361976e-02 -6.59668684e-01 -5.90041518e-01 -1.36063352e-01 -3.99848908e-01 -1.13783188e-01 -8.21640491e-01 1.22460413e+00 2.83391595e-01 -4.02488649e-01 8.77776265e-01 -1.03260863e+00 -7.08370686e-01 9.11243618e-01 -2.93059140e-01 2.52991766e-01 2.61509996e-02 -2.96300828e-01 6.44882798e-01 -6.15295410e-01 3.78475152e-02 1.01453471e+00 1.00282991e+00 6.01182938e-01 -1.03882623e+00 -6.20225012e-01 -5.46888232e-01 3.66070390e-01 -1.51122117e+00 -5.28457522e-01 4.27244782e-01 -3.59231800e-01 1.08992064e+00 -2.22742558e-03 3.69796664e-01 7.76812673e-01 5.43267310e-01 4.29068059e-01 6.00617170e-01 -9.81976032e-01 4.84032303e-01 5.16220704e-02 1.23400666e-01 1.06667995e+00 4.40931678e-01 2.76394606e-01 -3.04485261e-01 -1.12813957e-01 7.95148313e-01 -5.50728023e-01 -5.10301292e-01 -1.05101503e-01 -8.40443015e-01 9.22144353e-01 2.02985585e-01 2.18619198e-01 -1.02935098e-01 5.96703708e-01 2.99307764e-01 4.39063013e-01 1.76949844e-01 6.26055151e-02 -3.58625323e-01 -4.96481121e-01 -9.77635741e-01 -2.26125807e-01 1.27506864e+00 9.65689242e-01 5.66493571e-01 1.55263260e-01 3.86189111e-02 8.43696415e-01 6.91397905e-01 4.76698369e-01 6.09245360e-01 -7.78997064e-01 5.37316203e-01 3.23084712e-01 -3.79566997e-01 -1.04929662e+00 -4.31238204e-01 -7.49342680e-01 -1.26597822e+00 -1.66922450e-01 2.59109527e-01 -4.89040136e-01 -9.61589158e-01 1.54560566e+00 -4.06182557e-01 1.14704877e-01 3.30038458e-01 4.58845943e-01 4.19241250e-01 9.96923387e-01 -5.28947830e-01 -2.58466274e-01 8.43368530e-01 -8.61062109e-01 -5.14085174e-01 -1.00476451e-01 1.00727069e+00 -4.11203742e-01 1.14128157e-01 4.39314127e-01 -1.40774536e+00 -2.82006353e-01 -1.49174666e+00 1.34256780e-01 -4.61072549e-02 3.95360321e-01 3.57928157e-01 1.33898199e+00 -1.29469931e+00 6.04894042e-01 -1.01340926e+00 -7.42040873e-02 1.92440987e-01 7.38442540e-01 7.89544284e-02 -2.99766250e-02 -1.25601900e+00 7.77095079e-01 9.20550227e-01 2.32830346e-01 -1.00068712e+00 -1.47741169e-01 -7.28152812e-01 3.40453386e-01 -1.14532694e-01 -2.20120877e-01 1.27054417e+00 -7.49229312e-01 -1.46443713e+00 5.10317147e-01 -2.55005747e-01 -1.00702941e+00 -4.60312068e-02 5.26368439e-01 -6.14946723e-01 -1.01723120e-01 -6.20459497e-01 5.41583598e-01 6.82859659e-01 -1.09989703e+00 -5.95683575e-01 -8.56567174e-02 -1.59323588e-01 -2.24719852e-01 -3.66245776e-01 -2.42442027e-01 -7.80635178e-01 -3.62754643e-01 3.05095553e-01 -1.05801451e+00 -2.67335594e-01 -2.84858555e-01 -4.73666370e-01 -2.95427088e-02 2.65417039e-01 -8.86149883e-01 1.63274825e+00 -2.33239532e+00 2.08782077e-01 5.79999983e-01 7.50074536e-02 2.54826307e-01 1.85300142e-01 5.58707356e-01 1.04058221e-01 -7.23844022e-02 -5.66488326e-01 -9.01402384e-02 -1.26045629e-01 3.92439753e-01 -2.00663041e-02 5.51218033e-01 -1.43896684e-01 5.39285779e-01 -6.56518817e-01 -1.77603945e-01 -3.55020612e-01 3.82144570e-01 -5.81322491e-01 -1.52757332e-01 -3.48975547e-02 1.30448952e-01 4.23373058e-02 4.11644280e-01 6.79856956e-01 -5.13952076e-01 3.27644259e-01 2.15685889e-01 7.52490461e-02 2.94948637e-01 -6.57198370e-01 1.21262765e+00 -3.80882859e-01 1.13135922e+00 -1.80233002e-01 -1.27783060e+00 1.03861272e+00 5.23807764e-01 -1.16778620e-01 -5.33893585e-01 1.85487822e-01 5.12827992e-01 4.00419772e-01 -1.30298898e-01 5.63739836e-01 -7.50968382e-02 1.08978771e-01 1.94308773e-01 8.29991326e-02 2.97083795e-01 5.01690924e-01 8.03546682e-02 1.07262743e+00 -7.26984382e-01 3.16807866e-01 -3.04327816e-01 5.84390640e-01 -3.10455948e-01 3.84550542e-01 9.49137688e-01 9.56661068e-03 4.48987663e-01 9.33462024e-01 -2.02041477e-01 -1.38315940e+00 -8.81238699e-01 -3.94683838e-01 8.18803906e-01 1.83110729e-01 -4.52177614e-01 -8.30628335e-01 -1.30400836e-01 -2.45153472e-01 6.47830904e-01 -2.91381210e-01 -4.79176730e-01 -4.73548681e-01 -1.13093555e+00 1.04508066e+00 4.09410149e-01 5.64520836e-01 -5.89213789e-01 -1.73030734e-01 2.52039731e-01 1.19167425e-01 -1.13517547e+00 -5.88771440e-02 5.13469875e-01 -1.02696908e+00 -7.66930640e-01 -8.21446598e-01 -1.18385696e+00 8.16105068e-01 -3.88904303e-01 9.29327488e-01 2.92647660e-01 2.42507234e-01 -3.71627584e-02 -3.73585552e-01 -3.29990298e-01 -1.25922835e+00 2.08831757e-01 2.95068547e-02 -3.04063767e-01 6.83317259e-02 -5.05352259e-01 -3.66255850e-01 2.21568689e-01 -6.85559511e-01 4.49745029e-01 8.03436220e-01 8.20801556e-01 1.24059677e-01 4.57517952e-01 2.58881778e-01 -4.60578471e-01 7.16302097e-01 -7.44295955e-01 -7.43697941e-01 3.65424931e-01 -7.44699061e-01 4.92499828e-01 5.44420362e-01 -1.45374060e-01 -6.74008191e-01 -2.28755161e-01 -5.60097754e-01 3.66444319e-01 2.66793698e-01 9.21218097e-01 2.88776636e-01 -3.49656433e-01 7.88488328e-01 6.64584756e-01 -1.19295195e-01 -2.41969660e-01 -1.14716403e-01 1.06175637e+00 4.87589628e-01 -2.92443335e-01 4.04733479e-01 1.72388420e-01 2.65389621e-01 -8.21386099e-01 -3.04918736e-01 -7.58698210e-03 -2.41428569e-01 -1.17348373e-01 4.27693218e-01 -7.58018553e-01 -6.58936083e-01 5.33918619e-01 -1.07999647e+00 -2.97571212e-01 4.21056777e-01 6.27060950e-01 -5.44656754e-01 5.37453294e-01 -9.58303690e-01 -8.45813990e-01 -3.59692603e-01 -1.27459455e+00 5.24391532e-01 1.25054568e-01 2.97052592e-01 -1.17643666e+00 -2.13652793e-02 -1.36184126e-01 3.35999370e-01 -1.12520501e-01 1.19066441e+00 -5.34813166e-01 -3.84592801e-01 -3.48066777e-01 -4.80283946e-01 6.66043401e-01 -3.32157880e-01 -3.99355382e-01 -4.39600617e-01 -5.30969560e-01 -1.10343277e-01 -4.81716134e-02 9.17980969e-01 4.24640626e-01 1.28558373e+00 -3.44561607e-01 -4.62410629e-01 6.48414135e-01 1.57560503e+00 4.55931991e-01 8.17163467e-01 7.93003738e-02 3.77786636e-01 -2.64480025e-01 -2.68237591e-01 6.42486632e-01 1.42904326e-01 4.32978421e-01 4.74262506e-01 2.71462500e-01 4.51246239e-02 3.49356607e-02 4.58858639e-01 1.28811824e+00 -2.00299934e-01 -5.83213031e-01 -1.35462713e+00 3.61728549e-01 -1.59705281e+00 -5.37216604e-01 -2.80575693e-01 2.29348755e+00 1.04590380e+00 5.21939099e-01 -2.47070640e-01 5.40458798e-01 8.56291533e-01 -2.89099395e-01 -2.82877117e-01 -7.66028762e-01 6.30604103e-02 2.05376834e-01 1.11146295e+00 6.12692654e-01 -7.17004001e-01 4.83956963e-01 6.91739178e+00 1.01588285e+00 -9.47032213e-01 6.05768785e-02 8.42164695e-01 2.03944251e-01 -1.73989460e-02 3.87950316e-02 -1.06287694e+00 6.82825983e-01 1.57875514e+00 3.02892923e-02 6.06641531e-01 6.15246892e-01 -2.27174446e-01 -1.65612604e-02 -1.24368620e+00 1.14581907e+00 1.70465052e-01 -1.44759536e+00 -2.86803901e-01 2.32757673e-01 8.76460135e-01 3.27514708e-01 -4.51466888e-02 4.20686901e-01 1.15511246e-01 -1.02208257e+00 4.51911062e-01 3.90582889e-01 6.68822825e-01 -8.62884820e-01 8.90516281e-01 4.77123886e-01 -7.14638948e-01 -3.80087107e-01 -4.50500995e-01 -1.88023776e-01 1.80585936e-01 9.38639522e-01 -9.39487994e-01 1.82684571e-01 3.39197993e-01 4.71310645e-01 -3.05619299e-01 1.37254059e+00 7.09637254e-02 9.91749763e-01 -4.42840010e-01 -3.90882939e-01 3.76766980e-01 9.63654220e-02 3.85822684e-01 1.39579988e+00 7.52520323e-01 -1.20042942e-01 -2.13397980e-01 4.09223467e-01 -3.79449666e-01 -3.56234133e-01 -1.25595614e-01 -5.52858561e-02 5.38289487e-01 4.45866019e-01 -6.12042189e-01 -2.74626523e-01 -2.76628762e-01 8.70946825e-01 1.67455167e-01 3.82568926e-01 -7.31571555e-01 -3.58902603e-01 2.21857920e-01 -7.37066045e-02 6.78778410e-01 -5.62772334e-01 -4.84907150e-01 -7.99608052e-01 -1.11353928e-02 -5.94611943e-01 -1.04359396e-01 -4.66873825e-01 -6.67131662e-01 3.11986506e-01 -1.21049322e-01 -9.08508301e-01 -2.41979957e-01 -7.83991635e-01 -1.02711424e-01 7.03395426e-01 -1.34564495e+00 -3.73494446e-01 2.28814349e-01 8.92027393e-02 1.96822315e-01 -5.27966738e-01 8.28675807e-01 2.34909609e-01 -6.88291132e-01 9.91279006e-01 1.03778243e+00 3.36690426e-01 -1.72994137e-01 -9.46124196e-01 4.84231174e-01 7.75459647e-01 -6.87500685e-02 4.09359127e-01 8.68086517e-01 -4.56418246e-01 -1.50246441e+00 -6.74300969e-01 1.03512371e+00 2.51783401e-01 6.06171131e-01 -2.39451706e-01 -7.51335919e-01 4.57421571e-01 1.28717378e-01 -3.04292500e-01 6.80122137e-01 2.09301800e-01 -1.10873975e-01 1.00725755e-01 -8.34742069e-01 6.61431193e-01 6.47892177e-01 -3.30507129e-01 -7.49557689e-02 3.35388660e-01 3.00161272e-01 -6.52030110e-01 -9.07384217e-01 4.96131092e-01 5.86705804e-01 -1.08933306e+00 7.25835860e-01 -8.88899155e-03 5.87328315e-01 -2.77457871e-02 -2.20443740e-01 -1.14497888e+00 -4.00827616e-01 -5.83978832e-01 -7.05106795e-01 5.20776749e-01 9.20308173e-01 -3.56534243e-01 9.43351030e-01 2.78790623e-01 -3.42237622e-01 -1.15917218e+00 -1.11166441e+00 -8.86668742e-01 9.19497386e-02 -6.99040473e-01 2.62579501e-01 4.35925931e-01 3.73634964e-01 4.74375039e-02 -5.74365497e-01 4.18142945e-01 3.11004698e-01 -5.04117668e-01 1.35245055e-01 -1.05615985e+00 -7.45580792e-01 -5.31941891e-01 -7.92489290e-01 -1.54998982e+00 -2.63135284e-01 -9.71146107e-01 2.41889834e-01 -1.22878671e+00 1.67721227e-01 -5.30292571e-01 -3.74188632e-01 2.60290086e-01 1.89027578e-01 1.45663917e-01 -1.71967104e-01 2.62886137e-01 -4.36910003e-01 3.33277911e-01 7.73303807e-01 1.22595271e-02 -1.51550427e-01 1.73006773e-01 -3.86893213e-01 5.62540889e-01 1.36158884e+00 -4.65922356e-01 -3.31224799e-01 -5.85265815e-01 8.11638474e-01 4.31856185e-01 9.19350907e-02 -1.42173541e+00 5.09902596e-01 4.33825880e-01 2.28835002e-01 -5.96244991e-01 2.79803395e-01 -4.97546583e-01 -6.25765175e-02 9.11911428e-01 -4.54864293e-01 2.66175568e-02 1.51410311e-01 8.22522044e-01 -2.26508528e-02 -7.17917740e-01 6.74672484e-01 2.71145284e-01 -5.33546329e-01 8.77436474e-02 -8.42146099e-01 -2.88542956e-01 7.43809164e-01 -4.43967640e-01 -1.50050903e-02 -5.16124964e-01 -5.57012618e-01 -6.78353012e-02 1.34096846e-01 3.17532779e-03 7.57246792e-01 -1.23940170e+00 -8.81109476e-01 5.50368130e-01 8.52981508e-02 -4.81811076e-01 -1.79646090e-02 9.07851398e-01 -1.07738686e+00 7.91779697e-01 6.33202940e-02 -5.81055462e-01 -1.19192886e+00 3.09690922e-01 3.94713044e-01 -2.53317177e-01 -4.11077708e-01 1.04431486e+00 -3.57233882e-01 7.33224824e-02 5.71902990e-01 -3.12906265e-01 1.22939944e-01 -7.69496858e-01 5.39192379e-01 3.57914984e-01 2.18060941e-01 -2.82175720e-01 -6.93285614e-02 1.14772610e-01 -1.86297730e-01 -1.27327666e-01 8.02225113e-01 -9.40844044e-02 -2.77244568e-01 2.33217329e-01 1.45088792e+00 -5.09998262e-01 -9.42780256e-01 -2.80235648e-01 3.20415199e-02 1.11145601e-01 3.69921237e-01 -5.79883695e-01 -1.09105122e+00 8.19471359e-01 7.89905131e-01 5.16738296e-01 1.23415995e+00 6.12211972e-03 9.22923505e-01 6.31844282e-01 3.16440940e-01 -1.07033074e+00 -5.02013147e-01 9.76972759e-01 4.15874392e-01 -9.96798992e-01 -2.11593956e-01 -2.27911383e-01 -7.31779858e-02 1.45238328e+00 5.61552942e-02 4.61593941e-02 1.02501500e+00 3.67899597e-01 -4.85762626e-01 1.48691565e-01 -7.07919836e-01 2.08400875e-01 2.39878669e-01 4.96398896e-01 5.62958837e-01 1.84197694e-01 -5.34461141e-01 2.53980875e-01 -3.98318857e-01 4.00866903e-02 6.17282569e-01 7.45221078e-01 -8.81628096e-01 -1.17090142e+00 -4.81418937e-01 5.42839170e-01 -3.19263220e-01 -5.53919792e-01 -4.22865897e-02 2.47744143e-01 2.94624176e-02 8.96995425e-01 3.20158079e-02 -5.20938277e-01 -4.24234688e-01 -1.37782902e-01 6.56436145e-01 -2.19836310e-01 -5.39745763e-02 -5.94901741e-01 1.91982731e-01 2.31674492e-01 6.65438408e-03 -2.68353224e-01 -1.28996742e+00 -7.04569757e-01 -4.45314020e-01 2.94202805e-01 1.13851690e+00 1.08171511e+00 2.85207182e-01 3.64292949e-01 4.51915830e-01 -5.26822746e-01 -5.10094166e-01 -8.97734582e-01 -5.43162465e-01 -2.50130355e-01 4.73168880e-01 -1.61594570e-01 -2.00192884e-01 4.14345004e-02]
[6.505362033843994, 1.6522228717803955]
d8d7e06b-a058-4fac-97bd-02f3134a0e9f
expression-analysis-based-on-face-regions-in
1911.05188
null
https://arxiv.org/abs/1911.05188v1
https://arxiv.org/pdf/1911.05188v1.pdf
Expression Analysis Based on Face Regions in Read-world Conditions
Facial emotion recognition is an essential and important aspect of the field of human-machine interaction. Past research on facial emotion recognition focuses on the laboratory environment. However, it faces many challenges in real-world conditions, i.e., illumination changes, large pose variations and partial or full occlusions. Those challenges lead to different face areas with different degrees of sharpness and completeness. Inspired by this fact, we focus on the authenticity of predictions generated by different <emotion, region> pairs. For example, if only the mouth areas are available and the emotion classifier predicts happiness, then there is a question of how to judge the authenticity of predictions. This problem can be converted into the contribution of different face areas to different emotions. In this paper, we divide the whole face into six areas: nose areas, mouth areas, eyes areas, nose to mouth areas, nose to eyes areas and mouth to eyes areas. To obtain more convincing results, our experiments are conducted on three different databases: facial expression recognition + ( FER+), real-world affective faces database (RAF-DB) and expression in-the-wild (ExpW) dataset. Through analysis of the classification accuracy, the confusion matrix and the class activation map (CAM), we can establish convincing results. To sum up, the contributions of this paper lie in two areas: 1) We visualize concerned areas of human faces in emotion recognition; 2) We analyze the contribution of different face areas to different emotions in real-world conditions through experimental analysis. Our findings can be combined with findings in psychology to promote the understanding of emotional expressions.
['Ming-Yue Niu', 'Jian Huang', 'Jian-Hua Tao', 'Ya Li', 'Zheng Lian']
2019-10-23
null
null
null
null
['facial-emotion-recognition']
['computer-vision']
[ 2.70082746e-02 -1.76561475e-01 1.90788269e-01 -7.12382615e-01 -1.28193889e-02 -3.54930431e-01 2.25547954e-01 -4.19467598e-01 -2.07458496e-01 6.77056849e-01 -9.91316810e-02 3.56364459e-01 1.33348480e-01 -4.58873093e-01 -2.32000470e-01 -8.59482527e-01 -1.46271372e-02 -2.46566847e-01 -3.37614745e-01 -4.33989525e-01 2.13542476e-01 8.68634343e-01 -1.92413950e+00 4.50878710e-01 3.83289337e-01 1.40172040e+00 -5.27514696e-01 1.43609926e-01 -9.59494933e-02 4.36543822e-01 -6.14728510e-01 -4.45415229e-01 1.02196515e-01 -4.94571060e-01 -4.61983085e-01 1.84505969e-01 3.08691841e-02 -3.34991934e-03 2.97111243e-01 1.12902236e+00 6.47255182e-01 4.87357788e-02 6.22281194e-01 -1.71751273e+00 -3.16001773e-01 -2.70282120e-01 -8.48160386e-01 -8.35886896e-02 5.90181649e-01 1.78555995e-02 4.23528969e-01 -1.15874076e+00 6.44325614e-01 1.29165173e+00 5.21373272e-01 7.21047223e-01 -8.63626599e-01 -1.08377671e+00 1.74912959e-01 3.53800416e-01 -1.53071487e+00 -7.33840942e-01 9.76982892e-01 -4.47236240e-01 6.12910748e-01 3.20356816e-01 5.85177898e-01 1.20629239e+00 2.93555170e-01 3.86967808e-01 1.55010831e+00 -3.24973881e-01 2.08825350e-01 5.63820899e-01 3.55102941e-02 5.18546283e-01 -1.60462126e-01 1.25950336e-01 -5.59650123e-01 -2.60029715e-02 3.57783437e-01 -1.24094717e-01 -3.70651484e-01 -8.28707218e-02 -5.02520740e-01 6.99247003e-01 2.23335996e-01 4.85568613e-01 -6.01057768e-01 -3.39343816e-01 5.08237660e-01 3.66244078e-01 4.12910640e-01 2.86957055e-01 -4.21260744e-01 -6.10083118e-02 -6.38292313e-01 -1.49981782e-01 8.12947154e-01 5.59216499e-01 8.27709675e-01 9.68790054e-02 4.72193100e-02 1.06548476e+00 4.05447632e-01 3.30816746e-01 4.93355483e-01 -7.20518470e-01 -1.77667171e-01 5.71184993e-01 7.75909564e-03 -1.41446078e+00 -5.92873514e-01 1.20852649e-01 -8.62937927e-01 6.39842451e-01 2.11342350e-01 -4.35329169e-01 -5.05223453e-01 1.94918895e+00 4.99809206e-01 -5.86459935e-02 1.01091966e-01 1.05984390e+00 9.33868587e-01 5.73285758e-01 1.97435543e-01 -5.70242286e-01 1.64687049e+00 -6.58036768e-01 -1.10840034e+00 -1.45348236e-01 3.06307524e-01 -8.06131303e-01 8.57720673e-01 5.38122892e-01 -7.68798947e-01 -5.81202745e-01 -8.90877306e-01 3.30512732e-01 -7.83801675e-01 4.15722668e-01 6.77906573e-01 8.12223375e-01 -9.68747556e-01 2.53149569e-01 -2.64778852e-01 -6.40853822e-01 3.84684265e-01 5.03623962e-01 -9.23776090e-01 2.58171409e-01 -1.17432678e+00 1.06695747e+00 -9.41149518e-02 5.50803065e-01 -3.99590403e-01 -2.44748726e-01 -7.90501595e-01 -6.81895316e-02 9.56076980e-02 -7.43786246e-02 7.96118140e-01 -1.78488767e+00 -1.66743159e+00 1.15116990e+00 -4.28480417e-01 2.86511242e-01 8.65261257e-02 6.36353046e-02 -8.36990595e-01 7.40613416e-02 -3.20596129e-01 6.05842113e-01 7.36192465e-01 -1.28417659e+00 -2.33899251e-01 -9.11794603e-01 -4.03331578e-01 -6.34234622e-02 -1.46823615e-01 5.32515466e-01 -9.36866999e-02 -4.01448697e-01 2.34920793e-04 -8.86839509e-01 7.67950788e-02 2.59578913e-01 -3.28966416e-02 -3.06934237e-01 8.58252347e-01 -6.19798839e-01 1.00166726e+00 -2.53821921e+00 -2.00044557e-01 4.06044632e-01 1.96613986e-02 2.84594715e-01 -1.13793641e-01 1.33557454e-01 -5.45328557e-01 1.54088527e-01 8.06275904e-02 -2.08766162e-02 -1.39774121e-02 5.21802679e-02 -2.10653529e-01 5.17816067e-01 5.29956460e-01 7.11065948e-01 -4.00209785e-01 -6.11063421e-01 -1.88878868e-02 6.64112926e-01 -3.10853541e-01 2.02194795e-01 4.79245603e-01 3.11732680e-01 -3.65263939e-01 1.06997430e+00 9.10370886e-01 4.62869555e-01 6.69894442e-02 -4.77936178e-01 6.04692264e-04 -4.61944401e-01 -1.17908275e+00 1.01717877e+00 -2.99022317e-01 6.14343047e-01 5.25876343e-01 -1.01150894e+00 1.34829926e+00 5.32697380e-01 4.78511214e-01 -7.62824714e-01 6.51007056e-01 1.46821514e-01 1.36342481e-01 -7.60689497e-01 2.87433058e-01 -4.16899472e-01 1.30038694e-01 3.17049176e-01 2.13097259e-01 1.18505046e-01 -1.33788690e-01 -3.27996671e-01 4.21290010e-01 -4.65266667e-02 4.92160738e-01 -2.78860807e-01 6.35631979e-01 -6.18415236e-01 6.33270383e-01 -1.61425084e-01 -8.66615474e-01 3.14228177e-01 8.85265708e-01 -4.72561598e-01 -4.47719485e-01 -7.83454478e-01 -3.18704933e-01 9.17966127e-01 1.39932245e-01 -9.22743976e-02 -8.11260164e-01 -5.56020319e-01 -2.73656905e-01 3.33658308e-01 -9.03965890e-01 -3.28552723e-01 -8.22954401e-02 -9.00006831e-01 4.37895030e-01 3.88641089e-01 5.37084520e-01 -1.43263113e+00 -6.19086564e-01 -2.36241803e-01 -8.42947736e-02 -1.06289387e+00 -2.22602556e-03 -1.53578622e-02 -4.81795102e-01 -1.10919642e+00 -5.14778018e-01 -6.58586681e-01 7.57293403e-01 -9.54090282e-02 9.23044741e-01 -8.68482813e-02 -4.48321074e-01 3.79261285e-01 -2.87101626e-01 -7.79869974e-01 1.10197505e-02 -7.28230655e-01 3.01938534e-01 5.48486829e-01 7.38840640e-01 -5.19550622e-01 -4.76657420e-01 7.92213202e-01 -6.87764704e-01 -4.86704469e-01 3.38301361e-01 6.33148849e-01 4.64688540e-01 3.51493172e-02 8.07016015e-01 -4.89716619e-01 8.08134317e-01 -5.01546681e-01 -1.86519980e-01 3.82479757e-01 -3.40369642e-01 -4.94209349e-01 3.52357447e-01 -5.38261652e-01 -1.14901900e+00 1.43670067e-01 -2.66755432e-01 -3.20421100e-01 -5.38883090e-01 1.17009111e-01 -6.49255037e-01 -2.04544872e-01 6.52974844e-01 -1.59497578e-02 4.09538865e-01 3.34254615e-02 5.73054925e-02 1.04583156e+00 2.71952480e-01 -4.64090109e-01 2.22403005e-01 3.89654189e-01 -2.69837817e-03 -9.27236676e-01 -3.33238721e-01 -2.49424487e-01 -5.57490885e-01 -7.09210098e-01 8.39001477e-01 -7.39914954e-01 -1.10650229e+00 6.04234219e-01 -1.04872620e+00 1.44753709e-01 4.60129790e-02 3.50843519e-01 -3.35734636e-01 2.00613216e-01 -3.58736813e-01 -1.14189208e+00 -3.97454351e-01 -1.28180754e+00 9.63089705e-01 5.27080834e-01 -5.30305266e-01 -6.51798308e-01 -2.78789490e-01 1.41525954e-01 4.22328025e-01 4.71703917e-01 6.36141360e-01 -4.37770426e-01 2.92064786e-01 -3.05859268e-01 -1.34464607e-01 4.88577187e-01 1.86047882e-01 5.23739219e-01 -1.37661886e+00 1.32424742e-01 2.83329129e-01 -5.74149728e-01 4.48452771e-01 2.61245579e-01 1.16027319e+00 -7.86958635e-02 -1.64522678e-01 4.57010835e-01 1.10048425e+00 6.52770042e-01 1.05376065e+00 -1.17960349e-01 2.35502031e-02 1.36334336e+00 7.79838562e-01 5.04015565e-01 -1.26566082e-01 7.15761304e-01 4.63742495e-01 -2.87190765e-01 2.78470516e-01 8.66429135e-02 5.14326930e-01 3.17140102e-01 -2.20613688e-01 -1.67684808e-01 -4.33168471e-01 2.00683221e-01 -1.35667944e+00 -1.00596666e+00 3.76299769e-02 2.02826691e+00 5.15042365e-01 -5.08089662e-01 1.14250042e-01 2.31651276e-01 8.39115262e-01 -7.85005763e-02 -3.60028863e-01 -1.02252400e+00 -3.56573969e-01 2.54752100e-01 -3.80149662e-01 2.27280736e-01 -9.62774396e-01 8.16976428e-01 6.06573820e+00 6.62125885e-01 -1.65360141e+00 -2.34727085e-01 1.10638714e+00 -1.47591159e-01 1.94022417e-01 -3.38263214e-01 -6.52293026e-01 4.43573177e-01 6.79045975e-01 6.59947991e-02 4.13134009e-01 9.47265148e-01 1.80457741e-01 -2.95380563e-01 -9.55835700e-01 1.41652620e+00 3.16208422e-01 -3.77403051e-01 -1.90491915e-01 -1.05917908e-01 2.30802566e-01 -5.48758388e-01 9.59329680e-02 3.56715769e-01 -4.53981966e-01 -1.30762672e+00 3.24840546e-01 6.27401829e-01 8.31567168e-01 -6.69465601e-01 8.78762424e-01 8.10972229e-02 -9.11500692e-01 3.48694623e-02 -3.28374445e-01 -2.12849542e-01 2.37336401e-02 4.03483123e-01 -6.02062166e-01 2.55968809e-01 8.71618390e-01 3.81333113e-01 -2.96501219e-01 5.57123184e-01 -1.84021801e-01 2.55100846e-01 -2.06509635e-01 -2.19554022e-01 -2.19109386e-01 -3.64214808e-01 1.22842759e-01 1.06773067e+00 4.05039072e-01 4.63638157e-01 -4.17974859e-01 1.00298250e+00 6.08459935e-02 5.81562221e-01 -7.42732584e-01 -3.09257209e-03 1.46148875e-01 1.72538161e+00 -5.73976576e-01 -4.62261252e-02 -2.87486970e-01 1.07740009e+00 7.92634487e-02 4.93975967e-01 -8.39847207e-01 -4.51318830e-01 8.83751869e-01 -6.80918843e-02 -2.59167492e-01 3.31378043e-01 -1.87748149e-01 -8.69315445e-01 9.12709385e-02 -9.26561356e-01 2.50789493e-01 -1.03084922e+00 -1.25282097e+00 8.72369468e-01 -1.78764045e-01 -9.99467492e-01 6.45338371e-03 -8.74382436e-01 -6.69735670e-01 9.38357770e-01 -1.27132583e+00 -8.26237023e-01 -5.91297626e-01 7.59835660e-01 1.67344883e-01 -2.05931157e-01 1.03484905e+00 3.50046009e-01 -8.22886944e-01 8.62566829e-01 -4.23277080e-01 1.92110106e-01 8.40620875e-01 -5.35784423e-01 -4.88195896e-01 3.29308420e-01 -1.53858438e-01 5.10460556e-01 4.93637145e-01 -7.97944888e-02 -1.06831098e+00 -7.02049732e-01 9.51977909e-01 -2.18623891e-01 2.97828615e-01 -3.87409210e-01 -7.70516217e-01 3.07934135e-01 1.10616960e-01 2.45078221e-01 1.03160930e+00 1.16900600e-01 -1.94611073e-01 -3.10346156e-01 -1.53568089e+00 6.37485921e-01 6.26929164e-01 -4.24163431e-01 -1.96449116e-01 -2.48569790e-02 -6.51876926e-02 -3.77117167e-03 -9.19265330e-01 6.77637160e-01 1.01570535e+00 -1.26750040e+00 5.97219050e-01 -7.02555001e-01 5.75512528e-01 -1.83750868e-01 -2.19305769e-01 -1.19680345e+00 1.25101078e-02 -3.28745991e-01 3.75781208e-01 1.43686199e+00 3.36117357e-01 -8.83264720e-01 4.19884235e-01 8.96462977e-01 2.00138837e-01 -1.13719976e+00 -9.80037212e-01 -4.18000579e-01 -1.58775866e-01 -3.35976362e-01 6.38733685e-01 1.08598387e+00 3.54012549e-01 2.50123441e-01 -3.41954350e-01 -1.23734623e-01 -8.15993100e-02 1.70219485e-02 8.01816046e-01 -1.23939681e+00 2.07018852e-01 -4.63230878e-01 -6.62650883e-01 -2.80851811e-01 4.12301868e-01 -3.71685654e-01 -5.86590581e-02 -9.35337842e-01 1.68264970e-01 -1.16749570e-01 -3.00786734e-01 7.00533450e-01 5.73407747e-02 4.22792554e-01 2.07222059e-01 -1.62015289e-01 -3.37593406e-01 6.15829110e-01 1.19945705e+00 1.81975707e-01 -9.68056545e-02 6.27008732e-03 -7.35894382e-01 9.12443757e-01 8.50226283e-01 -2.85529029e-02 -1.85990840e-01 1.79558828e-01 4.89778109e-02 1.11677103e-01 2.27003157e-01 -7.17960954e-01 -8.09525773e-02 -1.73535421e-01 6.78820789e-01 -1.05475113e-01 6.82401240e-01 -9.89911914e-01 1.20390050e-01 1.82798669e-01 6.66785166e-02 6.26194924e-02 4.24992442e-01 9.69930217e-02 -5.75979948e-01 -6.14228360e-02 1.07551181e+00 -8.49456117e-02 -8.00653756e-01 1.11552738e-01 -4.98937517e-01 -3.41412842e-01 1.51035058e+00 -6.33298218e-01 -2.07724229e-01 -5.69848478e-01 -9.18149292e-01 -2.68722773e-02 2.94720024e-01 4.84153956e-01 5.89590788e-01 -1.30841005e+00 -4.66949880e-01 6.75517857e-01 2.67183036e-01 -6.24645412e-01 4.50201601e-01 1.26295698e+00 -1.59393743e-01 1.80875421e-01 -6.98099554e-01 -3.48770648e-01 -1.70537293e+00 4.87886906e-01 5.98031580e-01 2.89151967e-01 3.12940687e-01 8.67358804e-01 4.37808990e-01 -3.02010030e-01 1.78271428e-01 6.54022070e-03 -6.24738514e-01 4.44697887e-01 5.66909909e-01 2.37424225e-01 -3.33511941e-02 -1.14534569e+00 -5.63053846e-01 8.49804103e-01 2.32533529e-01 1.46981478e-01 9.34078097e-01 9.17770118e-02 -4.00141180e-01 2.98627943e-01 1.29969835e+00 8.12180713e-02 -6.44795835e-01 2.79696226e-01 -2.81376630e-01 -4.18777198e-01 -2.49673560e-01 -9.43617344e-01 -1.33385456e+00 1.14843416e+00 1.02617836e+00 -9.30036828e-02 1.68236446e+00 -1.21784031e-01 2.42378354e-01 6.63006604e-02 2.80167848e-01 -1.28159273e+00 1.27690360e-02 2.24214643e-01 1.23405588e+00 -1.36936808e+00 -3.01780522e-01 -5.92626870e-01 -9.67698038e-01 1.18960202e+00 8.60280752e-01 2.55579054e-01 8.99745643e-01 3.06317329e-01 3.90579045e-01 -2.83078015e-01 -6.19558692e-01 -9.34248138e-03 3.13023120e-01 5.61252713e-01 6.79434180e-01 -7.26812705e-02 -4.70207810e-01 1.12075794e+00 -2.45770782e-01 1.82747379e-01 1.71650887e-01 5.25888324e-01 -2.91492671e-01 -8.39092553e-01 -6.24649405e-01 2.86782593e-01 -5.89699686e-01 2.43475288e-01 -8.63141179e-01 7.93722749e-01 3.11560333e-01 1.12993956e+00 6.44748732e-02 -5.87375402e-01 3.56505692e-01 3.90877366e-01 3.75605762e-01 -1.31685451e-01 -5.21114588e-01 -6.49036840e-02 -4.10188399e-02 -7.55370855e-01 -5.53528845e-01 -5.75369060e-01 -1.26883733e+00 -2.32880712e-01 -3.94125044e-01 1.67569861e-01 8.39872479e-01 7.55980074e-01 5.07496536e-01 1.31184697e-01 7.13497341e-01 -7.54834414e-01 -1.43547982e-01 -1.10769391e+00 -9.76902425e-01 6.10954344e-01 -3.10381502e-02 -8.32813978e-01 -6.02917194e-01 -8.70369468e-03]
[13.524161338806152, 1.9101639986038208]
4fc595d8-9694-470a-a93a-448c16a23d42
sicilian-translator-a-recipe-for-low-resource
2110.01938
null
https://arxiv.org/abs/2110.01938v1
https://arxiv.org/pdf/2110.01938v1.pdf
Sicilian Translator: A Recipe for Low-Resource NMT
With 17,000 pairs of Sicilian-English translated sentences, Arba Sicula developed the first neural machine translator for the Sicilian language. Using small subword vocabularies, we trained small Transformer models with high dropout parameters and achieved BLEU scores in the upper 20s. Then we supplemented our dataset with backtranslation and multilingual translation and pushed our scores into the mid 30s. We also attribute our success to incorporating theoretical information in our dataset. Prior to training, we biased the subword vocabulary towards the desinences one finds in a textbook. And we included textbook exercises in our dataset.
['Eryk Wdowiak']
2021-10-05
null
null
null
null
['low-resource-neural-machine-translation']
['natural-language-processing']
[-7.57161900e-02 4.09650713e-01 -3.91774267e-01 -3.46910596e-01 -1.24381566e+00 -9.10202801e-01 6.05090737e-01 -1.36332273e-01 -7.40240276e-01 1.29558969e+00 5.56197524e-01 -1.04127729e+00 1.12386957e-01 -6.85098350e-01 -1.02520895e+00 -6.10786788e-02 5.39206326e-01 9.93699551e-01 -2.50454128e-01 -7.23300576e-01 -4.06651162e-02 3.26016955e-02 -6.09412551e-01 2.72317290e-01 1.32918823e+00 1.85896829e-01 2.80750632e-01 6.08870864e-01 6.18779659e-02 6.37316942e-01 -5.79329312e-01 -1.15968788e+00 3.83574128e-01 -6.44593120e-01 -9.87430573e-01 -5.78621149e-01 6.76078975e-01 -2.30857641e-01 -7.34302282e-01 9.31383491e-01 4.11520422e-01 -1.08643204e-01 5.03596485e-01 -6.05412364e-01 -1.17246985e+00 1.31466651e+00 9.59591866e-02 2.13756114e-01 2.10096434e-01 2.13962421e-01 1.19926596e+00 -8.50065887e-01 1.17155433e+00 1.04041195e+00 6.14000082e-01 6.72658622e-01 -1.23038077e+00 -7.83553004e-01 -1.70291662e-01 2.57700861e-01 -1.16222382e+00 -6.42905593e-01 6.28980458e-01 -4.77091447e-02 1.30646861e+00 8.35878216e-03 6.90918446e-01 1.48888516e+00 5.98138809e-01 7.69152045e-01 1.20347762e+00 -5.80392420e-01 -2.56501704e-01 3.68407577e-01 -2.76905060e-01 5.98381281e-01 1.34205490e-01 4.45043072e-02 -3.84478569e-01 2.44942576e-01 7.13061571e-01 -5.02957344e-01 -1.92611367e-01 2.41750002e-01 -1.39198065e+00 9.24109638e-01 3.62112939e-01 4.59230185e-01 -1.20806724e-01 8.99357870e-02 5.05871654e-01 1.16224623e+00 2.53235400e-01 8.37000012e-01 -6.94526136e-01 -3.23442131e-01 -7.77474403e-01 -1.28388777e-01 8.00448656e-01 1.24968326e+00 2.89919436e-01 -7.27078831e-03 9.01535749e-02 9.17087853e-01 -9.86469239e-02 6.58065438e-01 7.23425627e-01 -9.88812089e-01 1.02813947e+00 2.87149549e-01 -1.19891807e-01 -3.90877366e-01 -1.47001565e-01 -8.83797348e-01 -6.83122754e-01 -4.83157218e-01 5.25140524e-01 -4.14836794e-01 -8.67885470e-01 1.80186486e+00 -3.14797252e-01 -7.25524902e-01 2.35069126e-01 7.81914473e-01 5.86577058e-01 6.31920993e-01 -4.09419164e-02 -2.76415557e-01 1.06001210e+00 -9.70159829e-01 -6.52061582e-01 -3.37395906e-01 9.42812204e-01 -1.12867749e+00 1.50402009e+00 3.11050683e-01 -1.37963820e+00 -4.83832657e-01 -1.05639219e+00 -4.44397897e-01 -2.59232789e-01 -5.83933927e-02 4.13693696e-01 4.64580595e-01 -1.28047466e+00 8.22065175e-01 -6.69351041e-01 -2.50502765e-01 1.54893160e-01 5.08565664e-01 -4.90316868e-01 -1.27483100e-01 -1.70190716e+00 1.45869184e+00 1.96458533e-01 -2.16628283e-01 -6.44578576e-01 -5.47500253e-01 -9.11216497e-01 -1.65376425e-01 3.08612082e-02 -8.11858535e-01 1.42001367e+00 -1.08143210e+00 -1.61278367e+00 1.06179142e+00 1.33358873e-02 -5.09474516e-01 6.26795948e-01 6.01458028e-02 -4.35148120e-01 -1.60869017e-01 6.59778994e-03 6.44312978e-01 2.07384408e-01 -6.22581005e-01 -4.14067537e-01 -1.66410550e-01 1.43981889e-01 3.55367422e-01 -2.48070404e-01 1.56259239e-01 -3.18475366e-01 -8.21110487e-01 -1.03438616e-01 -8.05661500e-01 -1.01608872e-01 -4.65003550e-01 1.89609155e-02 -2.01287866e-01 -4.60743811e-03 -1.24965119e+00 1.05263627e+00 -1.79903913e+00 4.58933443e-01 1.71929032e-01 9.92456824e-02 -3.72986570e-02 -5.16621470e-01 5.22746980e-01 1.48467362e-01 4.98573780e-02 5.62911965e-02 -3.53191972e-01 1.39449149e-01 2.99183249e-01 -2.92879701e-01 4.56051677e-01 -1.01030208e-02 1.29971266e+00 -1.03080249e+00 -5.50020576e-01 -6.00548610e-02 3.43063325e-01 -6.51706576e-01 5.90199977e-03 -8.95855427e-02 3.04593027e-01 5.70560843e-02 5.31058729e-01 1.08549155e-01 9.08011198e-02 2.54295528e-01 1.78983450e-01 6.82540983e-02 1.12887764e+00 -2.94791996e-01 1.89037967e+00 -1.17713475e+00 1.00870204e+00 -1.68759078e-01 -7.01786518e-01 9.09892678e-01 4.14564550e-01 1.47248447e-01 -1.14589179e+00 3.10632825e-01 7.56669760e-01 4.87408340e-01 -1.65992528e-01 7.11517453e-01 -4.51574266e-01 -1.91547796e-01 3.77315253e-01 4.54896182e-01 -5.55100679e-01 3.47858608e-01 1.46566451e-01 7.64207602e-01 1.92013636e-01 1.37696937e-01 -5.24293482e-01 3.23827744e-01 1.38146117e-01 3.82949769e-01 5.36121666e-01 -2.28485782e-02 6.10493660e-01 3.76154482e-01 -4.52949047e-01 -1.52201056e+00 -1.11776006e+00 -1.74800724e-01 1.07255256e+00 -4.74405199e-01 -5.91110945e-01 -7.40856469e-01 -6.73898518e-01 -5.24986863e-01 1.06789231e+00 -4.45885181e-01 -2.17186660e-01 -1.01762486e+00 -2.43363023e-01 7.58510113e-01 4.11602795e-01 1.87700093e-01 -1.05187440e+00 3.56353372e-02 3.67536634e-01 -3.82670075e-01 -9.99761641e-01 -9.37439442e-01 4.35379982e-01 -8.72062683e-01 -7.59362936e-01 -8.46073151e-01 -1.27759624e+00 2.78924912e-01 -4.69776899e-01 1.57753181e+00 -2.33445123e-01 1.78147674e-01 -2.31714696e-01 -1.10696338e-01 -3.15225065e-01 -9.94364560e-01 7.68938303e-01 1.19891167e-01 -8.82527709e-01 5.71604371e-01 -6.47351742e-01 -1.42437428e-01 6.34521544e-02 -3.39588761e-01 2.00384647e-01 6.79994702e-01 9.96487498e-01 2.40516007e-01 -7.83478081e-01 5.02238512e-01 -9.25912619e-01 8.09627712e-01 -1.53530627e-01 -7.18355536e-01 1.60017818e-01 -6.47514462e-01 3.03193331e-01 1.25481033e+00 -5.66975474e-01 -6.53013945e-01 -2.98777223e-01 -3.94028991e-01 3.87798622e-02 2.91295916e-01 4.05099601e-01 -1.43417511e-02 1.25725091e-01 8.82780313e-01 3.06317538e-01 -3.56957316e-01 -2.53451884e-01 4.91800249e-01 9.16906178e-01 5.67257404e-01 -7.09753633e-01 7.18343139e-01 -3.35102081e-01 -3.54141444e-01 -5.58518708e-01 -5.73293030e-01 1.76918447e-01 -6.04994893e-01 4.37113233e-02 5.01993358e-01 -1.15668416e+00 -5.80625713e-01 -1.14091169e-02 -1.32012355e+00 -6.22323632e-01 -4.75916564e-01 7.48896241e-01 -6.04941487e-01 8.35316926e-02 -1.13874435e+00 -2.31802344e-01 -4.85791981e-01 -1.18247330e+00 6.18119121e-01 -3.69746424e-03 -7.29283452e-01 -1.25767171e+00 3.30505967e-01 4.55473661e-01 6.18247390e-01 -4.84808475e-01 1.07358778e+00 -6.53112173e-01 -1.88932002e-01 -9.73805338e-02 -8.93064737e-02 5.04798770e-01 7.85912275e-02 -5.41502178e-01 -5.00521779e-01 -3.81835788e-01 -6.62696511e-02 -5.96548200e-01 5.65159082e-01 9.16167349e-02 5.32449126e-01 -5.01944780e-01 -2.11474113e-02 8.01624656e-01 1.21177018e+00 7.53416121e-02 5.54864645e-01 5.43630004e-01 4.40078139e-01 2.95319676e-01 1.25202760e-01 -4.12022203e-01 5.80287218e-01 4.86778378e-01 -2.13166401e-01 -3.91381085e-02 -3.08988690e-01 -7.16717601e-01 7.54952550e-01 1.77413130e+00 -1.17229216e-01 -1.11566670e-01 -9.35108960e-01 9.53469574e-01 -1.28382325e+00 -6.00655138e-01 -3.54946069e-02 1.90688241e+00 1.55537057e+00 5.22433043e-01 -2.00681239e-01 -4.35950637e-01 3.30492914e-01 -1.89159513e-01 -3.04934114e-01 -8.46589267e-01 -3.91460001e-01 7.17865944e-01 8.96103501e-01 1.08556938e+00 -6.42693400e-01 1.55378306e+00 7.53961563e+00 7.85642207e-01 -9.73950446e-01 2.88161129e-01 4.96641815e-01 -4.05961722e-01 -6.66840911e-01 3.79059613e-02 -7.83130169e-01 4.12212014e-01 1.59445953e+00 -3.34914863e-01 9.59153771e-01 4.10080403e-01 3.01025584e-02 3.38048339e-01 -1.16882908e+00 7.26442814e-01 4.75351512e-02 -1.24650860e+00 2.84020938e-02 9.02060717e-02 1.18041980e+00 6.56276047e-01 -8.08978640e-03 6.14924908e-01 8.10207844e-01 -1.39055026e+00 6.99889958e-01 1.56833515e-01 1.02025366e+00 -7.97069192e-01 8.75102699e-01 2.67315865e-01 -5.73928893e-01 9.97721925e-02 -5.39440632e-01 -3.33969951e-01 -3.98885533e-02 1.53732568e-01 -9.84182477e-01 2.94289529e-01 -6.71828538e-02 5.75146675e-01 -5.66015422e-01 5.45619428e-01 -5.47532678e-01 8.20111930e-01 -3.92048329e-01 -2.87340462e-01 4.42859381e-01 -5.41248858e-01 3.08312416e-01 1.19719672e+00 2.48086095e-01 -2.68357635e-01 -1.21745139e-01 6.03892326e-01 -8.57787609e-01 5.17705679e-01 -7.01447725e-01 1.70780215e-02 3.94919395e-01 7.09229231e-01 -1.85864151e-01 -4.16715711e-01 -5.21841168e-01 1.08858144e+00 6.33820891e-01 2.94311911e-01 -6.44972503e-01 -7.12721050e-01 5.16243815e-01 1.25300273e-01 -6.99893944e-03 -3.14300150e-01 -6.04003251e-01 -1.46602190e+00 1.23803362e-01 -1.36460638e+00 7.65883550e-02 -6.75502062e-01 -9.73445654e-01 8.30828369e-01 -3.13329816e-01 -8.26959908e-01 -4.19264793e-01 -7.31255949e-01 -3.07056874e-01 1.16760802e+00 -1.25960159e+00 -1.12933111e+00 6.29855275e-01 3.19696784e-01 8.76900077e-01 -4.57318187e-01 8.76573205e-01 6.03137970e-01 -9.43039805e-02 1.07859635e+00 4.83032197e-01 4.36807245e-01 1.05006170e+00 -1.18044615e+00 8.88894737e-01 6.66729510e-01 3.22855532e-01 1.06449032e+00 9.24536705e-01 -5.37216425e-01 -1.21834862e+00 -9.06090617e-01 1.90482545e+00 -6.79926395e-01 1.23968399e+00 -5.41707873e-01 -5.46805203e-01 1.09988463e+00 7.92775989e-01 -7.14497209e-01 4.54944730e-01 1.28621280e-01 -1.70866057e-01 7.44977742e-02 -8.23389471e-01 9.59060073e-01 1.18438148e+00 -9.15948212e-01 -9.53846693e-01 5.65659404e-01 1.03576708e+00 -5.35340071e-01 -1.01038349e+00 -9.58993286e-02 5.58185875e-01 -1.27933890e-01 6.62382364e-01 -1.06230330e+00 8.59888077e-01 3.32103848e-01 -2.35049218e-01 -1.83020627e+00 -2.64990509e-01 -8.18254709e-01 2.26581216e-01 9.24094498e-01 1.00140643e+00 -5.56867540e-01 6.15775943e-01 1.61175281e-01 -2.68027842e-01 -6.10835552e-01 -1.11723638e+00 -7.70109773e-01 1.13328004e+00 -3.72145414e-01 4.00001913e-01 9.55109298e-01 3.70748103e-01 6.70395672e-01 -3.98580730e-01 -5.21903694e-01 4.05180514e-01 -1.95199866e-02 5.09138525e-01 -8.98466229e-01 -4.15840477e-01 -6.33927345e-01 4.91421781e-02 -1.23290133e+00 4.55516219e-01 -1.51951647e+00 -1.11548036e-01 -1.50719190e+00 1.49777576e-01 -5.84819168e-02 -1.19188666e-01 4.43611830e-01 -3.86072248e-02 6.49940372e-01 2.11403653e-01 -9.68750566e-02 -2.66144335e-01 4.57362711e-01 1.51498902e+00 -4.24920201e-01 -1.04820371e-01 -2.09434971e-01 -9.06781554e-01 3.75605285e-01 9.53648448e-01 -5.77597976e-01 -1.94884852e-01 -7.83341765e-01 5.68992853e-01 9.20918137e-02 -1.06607154e-01 -6.26809418e-01 1.44773200e-01 -1.81407064e-01 3.06672364e-01 -2.07867280e-01 1.39274612e-01 -6.47161722e-01 -2.17843009e-03 5.60656190e-01 -6.65053487e-01 2.98745275e-01 2.87433773e-01 -3.92651975e-01 -2.43891813e-02 -1.78525701e-01 7.80453682e-01 -2.05954641e-01 5.77233732e-02 -1.56422362e-01 -5.09982586e-01 3.42410922e-01 3.79760504e-01 1.70461550e-01 -9.87435281e-02 -4.73745644e-01 -6.46408916e-01 2.23212525e-01 7.72702157e-01 5.36433578e-01 1.92180425e-01 -1.38116503e+00 -1.06060362e+00 2.91524827e-01 -1.14654064e-01 -3.87375563e-01 -4.25322950e-01 7.35218108e-01 -6.71118855e-01 8.38254035e-01 -4.07870054e-01 -8.50589424e-02 -9.83614266e-01 2.02990264e-01 4.13751602e-01 -4.26483095e-01 -5.20819843e-01 5.83202839e-01 -2.88194716e-01 -9.93923903e-01 1.02773383e-01 -4.48810101e-01 2.32870921e-01 -2.48383120e-01 1.77530140e-01 3.28333303e-02 2.36889780e-01 -5.02424240e-01 -1.70907021e-01 4.97238040e-01 -7.91897774e-02 -3.56839538e-01 1.25935626e+00 -8.42223093e-02 5.61731085e-02 4.83890563e-01 1.43305683e+00 3.51160258e-01 -7.46243238e-01 -1.88599944e-01 -1.00098804e-01 -3.66306417e-02 8.62645637e-03 -1.03963268e+00 -7.21427321e-01 7.25859284e-01 3.68882939e-02 -4.32321787e-01 8.10251176e-01 -5.30004986e-02 1.11025023e+00 8.10752511e-01 4.84983921e-01 -1.30394173e+00 -5.83185792e-01 1.18151212e+00 5.80894113e-01 -1.15945375e+00 -2.03771651e-01 2.31779397e-01 -5.79188347e-01 1.10994601e+00 3.26730162e-01 -3.04927230e-01 7.78353810e-02 2.53546476e-01 3.38761121e-01 2.26163417e-01 -9.63684380e-01 1.58516496e-01 3.89310837e-01 4.60918039e-01 9.78401661e-01 1.43127963e-01 -8.62662375e-01 6.97179317e-01 -1.04064178e+00 9.74538475e-02 7.29705095e-01 4.50421154e-01 -3.07485372e-01 -1.37888670e+00 -2.14132160e-01 3.27305675e-01 -6.25021577e-01 -6.88275516e-01 -4.81032670e-01 8.19784641e-01 -1.03275932e-01 7.59099185e-01 -3.86538878e-02 -3.77091348e-01 3.92695427e-01 3.96839678e-01 7.92920828e-01 -3.34697962e-01 -9.28950608e-01 -2.63506826e-02 2.23643005e-01 -2.08723828e-01 3.40923429e-01 -5.00313520e-01 -9.54804361e-01 -7.67138958e-01 -1.29622474e-01 5.68646550e-01 6.76948547e-01 1.07880712e+00 -3.41716409e-02 5.51391006e-01 4.22395110e-01 -3.03637505e-01 -8.10999334e-01 -1.40835035e+00 -3.90819684e-02 1.41629592e-01 2.84890652e-01 1.48405120e-01 -2.19977766e-01 7.51852468e-02]
[11.552206039428711, 10.309807777404785]
d49ca235-b81c-438c-bf54-822cd723cf85
an-online-algorithm-for-nonparametric
1712.01521
null
http://arxiv.org/abs/1712.01521v1
http://arxiv.org/pdf/1712.01521v1.pdf
An Online Algorithm for Nonparametric Correlations
Nonparametric correlations such as Spearman's rank correlation and Kendall's tau correlation are widely applied in scientific and engineering fields. This paper investigates the problem of computing nonparametric correlations on the fly for streaming data. Standard batch algorithms are generally too slow to handle real-world big data applications. They also require too much memory because all the data need to be stored in the memory before processing. This paper proposes a novel online algorithm for computing nonparametric correlations. The algorithm has O(1) time complexity and O(1) memory cost and is quite suitable for edge devices, where only limited memory and processing power are available. You can seek a balance between speed and accuracy by changing the number of cutpoints specified in the algorithm. The online algorithm can compute the nonparametric correlations 10 to 1,000 times faster than the corresponding batch algorithm, and it can compute them based either on all past observations or on fixed-size sliding windows.
['Wei Xiao']
2017-12-05
null
null
null
null
['sequential-correlation-estimation', 'data-summarization']
['miscellaneous', 'miscellaneous']
[-3.86751890e-01 -5.88659108e-01 -3.97320449e-01 -6.74237311e-01 -4.80700165e-01 -4.31737989e-01 1.59541249e-01 5.89382231e-01 -6.20941937e-01 8.92866254e-01 -3.78790975e-01 -4.45327193e-01 -4.57169533e-01 -1.05492508e+00 -1.98441252e-01 -6.65624380e-01 -7.60041416e-01 7.60130167e-01 7.83625126e-01 -2.92771626e-02 4.19521153e-01 4.27546024e-01 -1.53913748e+00 -6.66611865e-02 8.57797861e-01 1.47356129e+00 -1.73422992e-01 1.05255044e+00 -3.41467500e-01 3.55334818e-01 -7.24225581e-01 -2.08458230e-01 2.16913775e-01 -5.60813963e-01 -1.98913455e-01 -3.93815964e-01 -3.11942875e-01 -3.37490976e-01 4.30760272e-02 8.58946919e-01 3.70198250e-01 3.91291082e-01 4.22902763e-01 -1.40242875e+00 -2.34599203e-01 5.34639120e-01 -1.00273466e+00 5.57260931e-01 4.28303212e-01 -5.63404799e-01 9.05447602e-01 -7.41871893e-01 6.78193793e-02 9.26055253e-01 5.93688965e-01 -3.52801867e-02 -9.19585168e-01 -7.07211077e-01 -1.42758578e-01 3.65884721e-01 -1.31941569e+00 -6.18637390e-02 1.98825881e-01 -1.03919186e-01 6.02019846e-01 7.54579008e-01 9.05887306e-01 1.97839782e-01 3.83263797e-01 2.24816993e-01 9.69603658e-01 -2.95666754e-01 6.66587234e-01 -1.52092859e-01 2.54518121e-01 2.01561466e-01 7.71275520e-01 4.77488898e-02 -7.98446715e-01 -5.96701860e-01 9.23807859e-01 3.03785741e-01 2.13268891e-01 -1.25599578e-01 -1.14022768e+00 7.85796106e-01 -2.19409227e-01 2.73912400e-01 -2.58696556e-01 8.08690265e-02 6.21315777e-01 8.18441272e-01 7.09749460e-01 -5.58084296e-03 -6.35724366e-01 -5.93726933e-01 -1.19496512e+00 2.55272001e-01 1.19021261e+00 1.00243151e+00 3.07249695e-01 -4.60238636e-01 2.02375516e-01 8.50003541e-01 -1.12638943e-01 5.02031207e-01 3.33075643e-01 -6.74458385e-01 2.74514437e-01 2.41428107e-01 3.53319556e-01 -8.44716311e-01 -6.56328678e-01 1.38653189e-01 -9.18820560e-01 8.86212289e-02 6.40591860e-01 -2.11412981e-01 -2.94542730e-01 7.99407959e-01 5.30759633e-01 1.77927434e-01 -3.63231122e-01 7.47351706e-01 4.34917659e-01 6.51457131e-01 -3.33881468e-01 -7.58646786e-01 1.34520078e+00 -7.24388301e-01 -9.29169714e-01 2.97630429e-01 6.04784787e-01 -7.99754322e-01 8.21355045e-01 7.08688378e-01 -1.20211172e+00 -2.74205565e-01 -1.08901775e+00 1.44315302e-01 -4.56256360e-01 -2.09292755e-01 1.04356349e+00 1.02002883e+00 -9.56041574e-01 7.94146657e-01 -1.11366999e+00 -2.81294227e-01 -2.87975110e-02 5.19493759e-01 -1.22036964e-01 3.08001377e-02 -9.35902894e-01 3.90495926e-01 6.78155124e-02 3.33623081e-01 1.28514990e-01 -5.13133824e-01 -5.09773374e-01 8.68293196e-02 2.22183600e-01 -4.30788934e-01 1.04349768e+00 -6.37085199e-01 -1.58289230e+00 3.93811464e-01 -4.52639788e-01 -3.63801390e-01 6.60704494e-01 -2.54776001e-01 -6.82648003e-01 5.68562783e-02 -5.92872389e-02 -3.19058806e-01 4.58901525e-01 -3.78770798e-01 -8.49063694e-01 -5.08500516e-01 -4.50995743e-01 -1.56302646e-01 -3.90097171e-01 4.16840553e-01 -3.22732031e-01 -4.57541227e-01 4.99892980e-01 -8.62383842e-01 -4.12370920e-01 4.49202023e-03 7.28203282e-02 -3.19469064e-01 6.30477130e-01 -3.53768289e-01 1.43973839e+00 -1.95547664e+00 -7.29097426e-01 8.35588992e-01 1.19427387e-02 -6.94166645e-02 2.95619994e-01 6.25058353e-01 2.09979072e-01 -7.49338558e-03 1.17342956e-01 -2.65363231e-02 -3.66050631e-01 3.52915645e-01 -1.18001200e-01 8.94104838e-01 -2.82469481e-01 4.71638680e-01 -9.63507473e-01 -4.89490330e-01 -4.78187539e-02 -8.28332976e-02 -3.21059585e-01 3.06040972e-01 2.05789626e-01 1.79200783e-01 -3.39789391e-01 4.69906867e-01 1.07337391e+00 -2.29716808e-01 4.24583524e-01 3.33160758e-01 -4.89996701e-01 2.30239443e-02 -1.34486723e+00 9.29020941e-01 -3.65416825e-01 6.38378739e-01 -1.41409963e-01 -1.27350056e+00 1.44253337e+00 2.36902341e-01 7.11110771e-01 -6.15264475e-01 8.40510875e-02 5.54981530e-01 -1.35145858e-01 -3.20785701e-01 6.30463362e-01 -4.06446308e-01 -7.28576630e-02 6.58141732e-01 -5.34306288e-01 -1.46643398e-02 6.36634350e-01 -4.29190509e-02 1.27196014e+00 -5.39663196e-01 4.54478532e-01 -3.86558324e-01 5.68526387e-01 -5.06332107e-02 5.93275487e-01 7.18857169e-01 3.91740426e-02 2.59880334e-01 9.74548697e-01 -7.58389056e-01 -9.52800453e-01 -1.30839205e+00 -1.91457048e-01 1.32857311e+00 2.33327851e-01 -4.21632171e-01 -2.10093278e-02 -6.28149658e-02 6.97833672e-02 2.17453614e-01 -6.58015668e-01 2.13460907e-01 -3.34205031e-01 -7.77542591e-01 9.28310677e-03 7.90896535e-01 3.38771313e-01 -6.94538951e-01 -6.78128123e-01 1.95863992e-01 3.14216852e-01 -9.55928683e-01 -3.74954462e-01 1.26447067e-01 -1.53166378e+00 -9.59896684e-01 -3.53872716e-01 -3.90757233e-01 7.09348917e-01 3.43563437e-01 1.04098785e+00 2.29926243e-01 -5.87020628e-03 -1.40956610e-01 -6.23602271e-01 -4.57981884e-01 1.89895183e-01 -2.01170295e-01 9.05273631e-02 -2.38245785e-01 5.44580877e-01 -6.19253278e-01 -4.98428464e-01 7.07442760e-01 -7.32727468e-01 -6.86623037e-01 3.59932899e-01 9.37074065e-01 7.54483759e-01 6.74219370e-01 7.18796194e-01 -1.14476180e+00 8.69785130e-01 -6.21896565e-01 -1.07041860e+00 -3.50272208e-02 -7.24422812e-01 -1.09616369e-01 9.22097743e-01 -5.05490005e-01 -8.22356999e-01 -2.12814912e-01 2.31725425e-01 -8.99687633e-02 5.06337464e-01 6.31297350e-01 2.81964332e-01 7.89766684e-02 3.01971018e-01 -4.68746051e-02 -3.74491746e-03 -3.79551023e-01 8.77158269e-02 5.73978961e-01 4.67425108e-01 -5.09144187e-01 5.42928398e-01 7.42605865e-01 3.32474679e-01 -8.06793571e-01 -5.28985500e-01 -9.61624801e-01 -6.55625045e-01 3.42347543e-03 4.11458135e-01 -5.53962648e-01 -1.09489775e+00 1.59205228e-01 -8.75355721e-01 -1.91486865e-01 -1.49551287e-01 8.82374763e-01 -5.44678628e-01 2.66454875e-01 -4.85156804e-01 -1.13643920e+00 -2.88805485e-01 -2.83525884e-01 5.59889257e-01 3.37476224e-01 -2.85546184e-01 -1.12673569e+00 2.53788233e-01 -1.86253995e-01 3.70741606e-01 2.93024391e-01 4.61969793e-01 -7.45820582e-01 -9.84944478e-02 -8.04166079e-01 -4.21182275e-01 -7.27647468e-02 2.55661947e-03 4.81617212e-01 -3.40031594e-01 -1.59958318e-01 4.60988991e-02 2.21378237e-01 5.14598250e-01 5.12050271e-01 1.51532149e+00 -2.67473787e-01 -1.83255211e-01 4.69667345e-01 1.12268031e+00 3.06704253e-01 6.19511127e-01 2.38989472e-01 1.38897479e-01 4.58736598e-01 1.18739545e+00 1.18681848e+00 2.68264681e-01 1.06466815e-01 -6.22558370e-02 1.10390857e-01 8.19331646e-01 1.06200427e-01 1.92438185e-01 1.21393132e+00 -4.68820274e-01 1.06835701e-02 -8.10816109e-01 5.39045215e-01 -2.09466648e+00 -1.07774830e+00 -6.29886627e-01 2.82178164e+00 6.85585380e-01 4.21500087e-01 4.62874949e-01 3.73111486e-01 6.86385274e-01 -3.84749286e-02 -3.16075861e-01 -1.10817826e+00 2.08685920e-01 4.51185405e-01 8.72761905e-01 8.02418217e-02 -6.70463920e-01 2.71334827e-01 7.25232315e+00 5.87199450e-01 -8.73317778e-01 -5.89841269e-02 5.43582737e-01 -6.03910387e-01 -5.83502762e-02 8.38395581e-02 -4.77470487e-01 7.10645139e-01 1.58329546e+00 -5.73434889e-01 1.73525333e-01 1.06262302e+00 5.16757488e-01 -9.14715528e-01 -1.04886746e+00 1.21472728e+00 -4.34109509e-01 -1.08850873e+00 -6.19271576e-01 1.02980375e-01 7.80331850e-01 -2.17565283e-01 -2.15425506e-01 -1.06770158e-01 1.11758888e-01 -9.03722882e-01 2.20612228e-01 6.23266220e-01 6.60184979e-01 -9.76701498e-01 1.30118597e+00 3.13512564e-01 -1.46090555e+00 -1.93763413e-02 -7.68785059e-01 -5.99629343e-01 2.39580169e-01 1.44492173e+00 -4.25804317e-01 2.98149258e-01 1.05818272e+00 4.41279054e-01 -1.66329488e-01 1.31098545e+00 9.63943228e-02 4.68018413e-01 -7.88571358e-01 -7.23115325e-01 -6.44783676e-02 -6.13819003e-01 -3.88377197e-02 1.01499760e+00 8.06923807e-01 3.68281454e-01 -3.72815207e-02 1.07548311e-01 2.41775125e-01 3.63326371e-01 -2.35112831e-01 -8.02928880e-02 7.83741295e-01 9.84354138e-01 -1.11935282e+00 -6.29136026e-01 -7.72877753e-01 5.65900743e-01 -1.21122733e-01 -1.30600901e-02 -7.84654498e-01 -8.19169462e-01 4.50081170e-01 4.79995519e-01 3.12763572e-01 -8.23078334e-01 -9.18565214e-01 -6.68654859e-01 1.50426880e-01 -2.22675413e-01 8.70165586e-01 -3.95005047e-01 -1.43865371e+00 2.89828449e-01 1.44882366e-01 -1.18879998e+00 -2.40788445e-01 -5.19748747e-01 -8.25081885e-01 7.11964905e-01 -9.11310434e-01 -1.36898488e-01 -5.09334624e-01 6.13030851e-01 1.39706999e-01 1.78414628e-01 7.30129421e-01 2.32238188e-01 -3.43189031e-01 5.98511696e-01 6.75132573e-01 -2.74796069e-01 8.02076221e-01 -1.19182003e+00 4.57063437e-01 4.87422168e-01 -2.32144922e-01 6.21919215e-01 9.93452489e-01 -6.64582849e-01 -1.47347641e+00 -3.37022424e-01 1.12387884e+00 -1.10018618e-01 1.20678806e+00 -2.71518111e-01 -9.13395047e-01 6.33773506e-02 -3.39047223e-01 6.37994528e-01 1.10662818e+00 6.88795984e-01 -1.98322445e-01 -5.61943352e-01 -1.14185166e+00 3.85865122e-01 8.08628917e-01 -2.98862427e-01 -6.27107397e-02 2.25295931e-01 2.15112582e-01 -2.02864721e-01 -1.31400907e+00 5.36824130e-02 1.07151568e+00 -1.24993098e+00 4.79466558e-01 -4.15540129e-01 8.19398761e-02 -1.40770823e-01 1.36349723e-01 -8.89991581e-01 5.42119257e-02 -8.84812653e-01 -2.80475289e-01 7.23114491e-01 3.23271126e-01 -1.16782355e+00 9.02839422e-01 8.56577396e-01 4.99496579e-01 -1.05627787e+00 -1.03523898e+00 -1.27444208e+00 -3.60093981e-01 -5.70637941e-01 7.61842370e-01 6.91283345e-01 5.49792707e-01 -6.71580434e-02 -3.01384479e-01 2.43311077e-02 6.88065886e-01 5.17207980e-01 9.42007601e-01 -1.37180626e+00 -2.18043536e-01 -1.84811711e-01 -5.90569615e-01 -9.83494401e-01 -5.29437840e-01 -9.41733196e-02 -1.27689585e-01 -1.18214321e+00 -3.95768089e-03 -8.27630818e-01 -4.33578789e-01 3.52458507e-02 -4.29505631e-02 2.48752207e-01 -3.51033449e-01 1.57993764e-01 -6.40546322e-01 2.38572031e-01 9.32363749e-01 5.33951402e-01 -4.76356387e-01 4.54020590e-01 -2.35893801e-01 5.37017345e-01 1.07366407e+00 -6.53416157e-01 -3.29343230e-01 1.15596745e-02 6.37866497e-01 4.99874771e-01 -2.32197970e-01 -8.05732012e-01 5.36870718e-01 -3.85530055e-01 5.30743182e-01 -1.01684523e+00 6.68600425e-02 -6.79036796e-01 5.29319122e-02 3.60232294e-01 1.01850800e-01 5.34333706e-01 5.50347613e-04 5.35240829e-01 -4.00961339e-01 -2.55824983e-01 6.97695851e-01 1.28334031e-01 -3.72648716e-01 3.70831639e-01 -3.39070231e-01 -1.30202800e-01 1.06733322e+00 -3.49319637e-01 -2.09352717e-01 -4.66805249e-01 -6.98972046e-01 3.83526951e-01 3.10449451e-01 1.04076177e-01 5.87932587e-01 -1.26591432e+00 -3.77930701e-01 2.41841719e-01 -9.98804048e-02 -1.23712629e-01 3.03279757e-01 1.20686865e+00 -8.37947249e-01 2.45314851e-01 -2.66503602e-01 -3.64593923e-01 -1.21373773e+00 6.02617979e-01 -3.74028355e-01 -3.86945903e-01 -4.71518248e-01 7.85601318e-01 -3.09724450e-01 1.30557358e-01 -1.83534890e-01 -4.46778327e-01 2.03844413e-01 2.12313607e-01 1.06186450e+00 1.02542782e+00 1.72394719e-02 7.86916465e-02 -4.95196074e-01 4.11939889e-01 3.44087631e-02 -2.54681796e-01 1.44464636e+00 -9.24426243e-02 -4.71348494e-01 9.91955519e-01 1.20626831e+00 2.77322590e-01 -9.11943018e-01 1.00086048e-01 5.08620739e-01 -8.73134673e-01 -2.60267347e-01 -5.61561994e-02 -8.13414872e-01 5.90542793e-01 3.13597202e-01 7.36635447e-01 1.23357773e+00 1.49194114e-02 9.00704801e-01 3.27651769e-01 5.88916063e-01 -1.58568847e+00 -9.21189114e-02 3.96876335e-01 4.74893600e-01 -1.04953802e+00 6.30049825e-01 -5.29201627e-01 -4.18299288e-01 1.47172904e+00 1.49468094e-01 -5.17787397e-01 1.00082171e+00 5.55805087e-01 -1.50532424e-01 7.88284242e-02 -1.04580939e+00 8.31781924e-02 7.38368407e-02 4.53428626e-01 6.63949132e-01 3.68809283e-01 -1.16254747e+00 7.89692402e-01 -4.83890235e-01 6.68105036e-02 6.31223381e-01 9.12852108e-01 -6.53305769e-01 -1.19777250e+00 -5.16797781e-01 1.08894169e+00 -4.92026716e-01 1.45184919e-01 1.46066546e-01 5.28657258e-01 -3.08693469e-01 1.15879297e+00 6.11951828e-01 -2.93644965e-01 2.06394568e-01 -3.88235718e-01 2.55483538e-01 -3.33395958e-01 -1.96505249e-01 5.00758961e-02 1.47337735e-01 -8.18651438e-01 -2.53574967e-01 -8.42716336e-01 -1.45026553e+00 -9.68538344e-01 -4.39021707e-01 5.65156758e-01 8.51715684e-01 6.66755497e-01 -1.10089354e-01 1.93267167e-01 8.62544715e-01 -4.75906134e-01 -5.55601656e-01 -7.94165552e-01 -1.21536160e+00 -1.47083580e-01 -2.36844607e-02 -7.37330616e-01 -3.91045183e-01 -1.85687974e-01]
[7.165168285369873, 3.9698615074157715]
5ac3348f-63a9-4a15-bc87-8a8fc2655cec
a-neural-attention-model-for-abstractive
1509.00685
null
http://arxiv.org/abs/1509.00685v2
http://arxiv.org/pdf/1509.00685v2.pdf
A Neural Attention Model for Abstractive Sentence Summarization
Summarization based on text extraction is inherently limited, but generation-style abstractive methods have proven challenging to build. In this work, we propose a fully data-driven approach to abstractive sentence summarization. Our method utilizes a local attention-based model that generates each word of the summary conditioned on the input sentence. While the model is structurally simple, it can easily be trained end-to-end and scales to a large amount of training data. The model shows significant performance gains on the DUC-2004 shared task compared with several strong baselines.
['Alexander M. Rush', 'Jason Weston', 'Sumit Chopra']
2015-09-02
a-neural-attention-model-for-abstractive-1
https://aclanthology.org/D15-1044
https://aclanthology.org/D15-1044.pdf
emnlp-2015-9
['abstractive-sentence-summarization', 'extractive-document-summarization']
['natural-language-processing', 'natural-language-processing']
[ 3.97609025e-01 5.70030689e-01 -2.74218529e-01 -3.75368506e-01 -1.50123632e+00 -4.60454911e-01 7.24481344e-01 3.82373720e-01 -4.04014438e-01 1.11679602e+00 1.09333038e+00 -3.22101116e-01 4.12727207e-01 -4.69897121e-01 -6.31481469e-01 -8.61407071e-02 2.81138062e-01 6.05201542e-01 -1.25637606e-01 -4.15218771e-01 6.64708793e-01 3.08719147e-02 -1.00200987e+00 5.68712771e-01 1.08476067e+00 5.05079687e-01 2.01367438e-01 1.20559013e+00 -1.86931208e-01 9.93849277e-01 -1.14620078e+00 -4.45955843e-01 -1.68766811e-01 -8.84545565e-01 -1.22213745e+00 4.96883057e-02 8.55898678e-01 -5.74658036e-01 -2.86928087e-01 7.12487936e-01 6.86573744e-01 1.86307222e-01 7.02984571e-01 -5.73174238e-01 -8.20067942e-01 8.32489610e-01 -2.67422110e-01 3.40232104e-01 6.07035697e-01 1.26519978e-01 1.34256709e+00 -9.01840150e-01 6.96372688e-01 1.16316152e+00 4.93450373e-01 7.88747728e-01 -1.21095693e+00 -1.42579302e-01 2.60684162e-01 -8.16490278e-02 -6.59510374e-01 -7.33326077e-01 6.21179223e-01 -1.07145526e-01 1.46971703e+00 4.15899247e-01 6.42341614e-01 1.07615232e+00 3.63657922e-01 1.21206284e+00 5.72428346e-01 -4.90687579e-01 2.09908843e-01 -3.24970156e-01 3.77094299e-01 4.94572997e-01 3.93440396e-01 -6.98532104e-01 -8.02714944e-01 -1.60274908e-01 1.00925200e-01 -4.38416004e-01 -2.45899498e-01 2.73052394e-01 -1.03192282e+00 8.78727853e-01 2.33157039e-01 1.05055027e-01 -5.89480221e-01 2.50859141e-01 6.95654511e-01 1.81653619e-01 1.07491934e+00 9.57366645e-01 -2.61958510e-01 -3.97347569e-01 -1.58375537e+00 7.29153931e-01 9.76035058e-01 1.08329439e+00 2.38231793e-01 2.74603754e-01 -7.42172360e-01 6.40083373e-01 -1.14633709e-01 3.51124644e-01 5.67218959e-01 -9.20626760e-01 9.84142423e-01 3.38613659e-01 2.47209951e-01 -7.21063197e-01 -2.83596754e-01 -4.23092455e-01 -7.46878266e-01 -1.98745638e-01 -2.22609993e-02 -5.26126921e-01 -8.28357458e-01 1.37889278e+00 -1.29566863e-01 -2.97793418e-01 3.09479624e-01 5.46348035e-01 1.05785024e+00 1.05196631e+00 -1.32077768e-01 -4.06814635e-01 9.80005443e-01 -1.45385528e+00 -8.94757450e-01 -5.80349743e-01 5.58643460e-01 -6.04610622e-01 9.23791051e-01 2.89184630e-01 -1.73640561e+00 -2.44798198e-01 -1.25888646e+00 -7.15170264e-01 6.09083995e-02 1.26548439e-01 3.60852599e-01 3.20919961e-01 -1.29310679e+00 6.53891146e-01 -9.85487700e-01 -5.42344034e-01 4.79346901e-01 1.32310778e-01 -1.30648583e-01 2.21051648e-01 -7.34432518e-01 9.33382094e-01 5.69402874e-01 -7.80774429e-02 -7.02737868e-01 -5.12727261e-01 -7.65076160e-01 2.91228950e-01 3.32933962e-01 -1.31623340e+00 1.87589192e+00 -7.88453281e-01 -1.86518657e+00 4.25573647e-01 -6.54506028e-01 -9.36128139e-01 3.71411890e-01 -7.22395122e-01 7.97811076e-02 4.19135988e-01 2.05964312e-01 5.49160779e-01 6.90917134e-01 -9.04595852e-01 -5.21954536e-01 1.39040651e-03 2.58351304e-02 4.19123322e-01 -4.42738086e-01 2.84471035e-01 -3.73926871e-02 -8.89569461e-01 -3.48284215e-01 -6.34538352e-01 -4.16900903e-01 -5.98203480e-01 -7.62096465e-01 -2.55343676e-01 5.08442342e-01 -1.04694355e+00 1.50000191e+00 -1.52806067e+00 3.78225476e-01 -5.95453322e-01 4.16668393e-02 5.39590061e-01 -3.82822752e-01 1.01151788e+00 1.84364542e-01 4.09661770e-01 -5.98848939e-01 -6.89320385e-01 4.32853140e-02 -2.97242463e-01 -6.35562420e-01 -4.37154137e-02 5.88820875e-01 1.25249350e+00 -1.15561080e+00 -5.42532265e-01 -1.03279248e-01 -5.23330905e-02 -8.19730937e-01 3.70506018e-01 -5.36093295e-01 5.18822558e-02 -4.30240393e-01 1.87037945e-01 3.81901771e-01 -6.81865439e-02 1.05048411e-01 9.34874490e-02 -1.19391210e-01 9.02550459e-01 -4.10586894e-01 2.03522563e+00 -4.53628123e-01 9.30909812e-01 -1.28261611e-01 -8.30504715e-01 7.11803317e-01 4.00546730e-01 -6.50987104e-02 -2.03663141e-01 -1.44307064e-02 1.95866942e-01 -2.37029895e-01 -4.28670287e-01 1.39251506e+00 -4.56509925e-02 -3.74234021e-01 8.04985523e-01 3.15645158e-01 -6.41989768e-01 7.05519557e-01 8.25269461e-01 1.31883276e+00 1.42441332e-01 6.29492819e-01 -2.62765288e-01 3.02051783e-01 2.37697259e-01 3.83255154e-01 8.66917014e-01 2.27303192e-01 1.04510069e+00 6.15175009e-01 -2.16936097e-01 -1.21307492e+00 -9.21158850e-01 3.63758266e-01 9.75135624e-01 -2.58111656e-01 -9.85796213e-01 -1.07999647e+00 -9.50013578e-01 -3.43922913e-01 1.33031082e+00 -4.18533057e-01 -1.21363044e-01 -7.85084128e-01 -4.14858460e-01 6.68425679e-01 6.66006804e-01 4.27774876e-01 -1.16968369e+00 -6.08522832e-01 5.22400379e-01 -5.45002639e-01 -9.50885057e-01 -8.48206878e-01 -2.23291665e-01 -1.01741254e+00 -5.86793661e-01 -6.45688236e-01 -6.30630493e-01 4.83008385e-01 2.46408030e-01 1.40314865e+00 -5.06911986e-02 4.90136519e-02 1.08173542e-01 -3.56092930e-01 -9.25141394e-01 -6.15956008e-01 7.33109176e-01 -2.47202992e-01 -3.08737755e-01 2.19602004e-01 -2.79421926e-01 -4.67193455e-01 -5.42818308e-01 -9.35096800e-01 2.40594000e-01 6.47363305e-01 8.33711445e-01 1.74576655e-01 -6.48933232e-01 1.16814971e+00 -1.08306515e+00 1.28960180e+00 -2.91067779e-01 4.39294688e-02 2.21350953e-01 -4.10543263e-01 2.61929512e-01 8.55872750e-01 1.63160279e-01 -1.16475213e+00 -2.57716179e-01 -3.64907295e-01 2.67740488e-01 -1.10602900e-01 8.38337660e-01 6.40186220e-02 7.66660154e-01 7.05649972e-01 2.62263387e-01 -1.52380824e-01 -5.67523777e-01 5.61962843e-01 9.43568885e-01 6.38367057e-01 -2.79111087e-01 5.98538935e-01 1.85188264e-01 -2.95531213e-01 -1.12781394e+00 -1.22567642e+00 -1.98677301e-01 -5.68760812e-01 1.92402169e-01 7.99904585e-01 -1.00378942e+00 1.01012446e-01 1.68711007e-01 -1.65708256e+00 -3.34604472e-01 -5.93892574e-01 1.52034178e-01 -6.40982270e-01 6.28108144e-01 -7.36375093e-01 -6.45333230e-01 -1.24949872e+00 -6.16635799e-01 1.14471114e+00 1.89779401e-01 -8.24070394e-01 -9.00780737e-01 2.39400715e-01 2.93056935e-01 5.08153915e-01 1.86861515e-01 6.35969520e-01 -8.47599149e-01 -3.55603576e-01 -4.29938674e-01 -9.01910961e-02 5.37643611e-01 9.97020081e-02 2.05504939e-01 -5.81192672e-01 -2.43000925e-01 -1.44716159e-01 -6.60998046e-01 1.32124412e+00 3.35796267e-01 9.94328916e-01 -8.78956199e-01 -2.50752773e-02 8.18696171e-02 1.08549142e+00 -3.63826215e-01 6.19313180e-01 1.32086203e-01 6.14831626e-01 4.12139654e-01 5.08121431e-01 3.44756216e-01 6.10526562e-01 3.07237267e-01 5.17656207e-02 1.10773392e-01 -3.97986263e-01 -4.30237979e-01 5.77317297e-01 1.26605463e+00 -2.52338853e-02 -7.33708620e-01 -5.70491374e-01 8.02900612e-01 -1.96500313e+00 -1.26204097e+00 -3.07226658e-01 1.79644907e+00 1.02423847e+00 2.34195054e-01 1.07412294e-01 -2.32388437e-01 3.72120649e-01 6.54374063e-01 -3.77637625e-01 -1.15618110e+00 -1.42963663e-01 5.05838513e-01 9.50432345e-02 7.28895903e-01 -8.55287313e-01 1.13988686e+00 7.45146561e+00 5.61394811e-01 -9.38876390e-01 -2.27608514e-04 2.86657035e-01 -6.47423685e-01 -4.68448728e-01 -2.67214086e-02 -7.55827367e-01 2.58044511e-01 1.29680455e+00 -7.10954130e-01 -9.39669311e-02 6.08568132e-01 4.51653779e-01 -2.33143389e-01 -1.27140057e+00 5.60794234e-01 6.07893169e-01 -1.62675035e+00 4.81765270e-01 -1.02937594e-01 8.89167964e-01 7.25883022e-02 -3.31346422e-01 3.48381728e-01 2.40455970e-01 -9.43979323e-01 8.02932799e-01 4.84386891e-01 6.73307359e-01 -8.37304115e-01 6.45469844e-01 4.87010241e-01 -6.85952902e-01 1.76921591e-01 -4.17155147e-01 -3.75469506e-01 5.41609526e-01 4.90656614e-01 -8.66620600e-01 8.68316770e-01 4.83013764e-02 8.45099866e-01 -7.90554821e-01 7.86161482e-01 -5.84122658e-01 1.00414050e+00 6.62932470e-02 -4.26517576e-01 3.31363767e-01 2.85231564e-02 8.33056211e-01 1.72796297e+00 2.30658129e-01 1.43407494e-01 1.57510251e-01 6.18289590e-01 -4.54705536e-01 2.17109784e-01 -7.28036463e-01 -3.50420862e-01 1.93052068e-01 1.16682446e+00 -2.31672883e-01 -7.42328048e-01 -2.82619417e-01 1.27085459e+00 7.09315479e-01 4.16617244e-01 -5.02034545e-01 -8.20197999e-01 1.25415057e-01 -1.27382576e-01 4.63629514e-01 -3.57984483e-01 -5.56283414e-01 -1.49617612e+00 1.37208879e-01 -9.72670138e-01 1.73126459e-01 -6.49556816e-01 -1.08945322e+00 7.67771542e-01 1.54044442e-02 -8.76691282e-01 -6.93192899e-01 -1.45859674e-01 -1.05763483e+00 8.49287450e-01 -1.33178532e+00 -9.96972144e-01 1.72910150e-02 -9.25473422e-02 1.23038065e+00 -1.12614721e-01 9.36143696e-01 -2.50802249e-01 -7.28986919e-01 5.66623032e-01 1.65022656e-01 1.04529085e-02 7.27144659e-01 -1.56453419e+00 1.05670345e+00 1.20155847e+00 5.82194366e-02 6.41423166e-01 1.02040505e+00 -6.63307369e-01 -1.22613227e+00 -1.18369472e+00 1.51769352e+00 -5.10408103e-01 4.86893922e-01 -3.71383607e-01 -6.92939818e-01 8.63946557e-01 1.13851511e+00 -6.85704887e-01 6.47597849e-01 2.10615247e-01 -1.64435461e-01 1.71536103e-01 -7.36893713e-01 7.75400996e-01 8.35755110e-01 -3.71199131e-01 -1.19475079e+00 6.32659256e-01 8.31051171e-01 -4.04562086e-01 -5.06227911e-01 2.35600378e-02 3.52174819e-01 -6.44004464e-01 5.15804708e-01 -7.85851538e-01 1.16683722e+00 1.13674020e-02 1.91651776e-01 -1.70328379e+00 -3.72269750e-01 -9.46078360e-01 -5.35336494e-01 1.33874893e+00 7.46922433e-01 -3.88556898e-01 6.75465763e-01 5.74845552e-01 -6.40289843e-01 -7.17998147e-01 -6.66117847e-01 -7.96517849e-01 4.45646197e-01 9.66378301e-02 4.66071278e-01 3.13445240e-01 4.88897920e-01 1.15155816e+00 -3.76789004e-01 -3.96893501e-01 4.09523159e-01 2.20808521e-01 1.04723442e+00 -8.89739692e-01 -2.90476233e-01 -6.45934403e-01 8.85660276e-02 -1.38907492e+00 2.39678264e-01 -1.00988662e+00 2.96197981e-01 -2.32006097e+00 3.41856748e-01 5.04322112e-01 2.04529673e-01 3.36617082e-01 -6.67207718e-01 6.61119148e-02 3.60416859e-01 -5.11070676e-02 -8.86350513e-01 9.52124238e-01 1.00911427e+00 -4.03353930e-01 -2.68014610e-01 -1.73068643e-01 -1.08640420e+00 5.05757809e-01 1.05792582e+00 -4.48920876e-01 -3.30353498e-01 -8.20399344e-01 7.01972246e-02 7.07403421e-02 -4.06257510e-02 -8.06096792e-01 3.09465617e-01 -2.02114917e-02 7.29078427e-02 -1.03709686e+00 1.94764853e-01 2.46846616e-01 -5.11318624e-01 2.74797112e-01 -7.93588161e-01 2.45000228e-01 1.52940854e-01 5.25271177e-01 -2.87057459e-01 -5.06684482e-01 4.83219594e-01 -2.72570878e-01 4.72275466e-02 2.94049829e-02 -6.17912233e-01 5.49333751e-01 6.12917483e-01 8.75846595e-02 -3.88181359e-01 -7.08718836e-01 -1.23662934e-01 2.58797735e-01 4.00264442e-01 1.93469122e-01 6.21291041e-01 -9.37663615e-01 -1.37704408e+00 -3.45149040e-01 -9.78854224e-02 3.54008913e-01 -2.75941268e-02 4.61918533e-01 -7.33837605e-01 6.44585788e-01 8.74760449e-02 -1.57165706e-01 -1.12605953e+00 4.29199070e-01 -1.47348553e-01 -6.43686354e-01 -7.85229385e-01 6.75510108e-01 -6.21075705e-02 -2.56310180e-02 -5.15938736e-02 -4.07883704e-01 -4.56292070e-02 9.92271155e-02 9.68790472e-01 4.03182298e-01 1.78547785e-01 -3.24374229e-01 -6.86214492e-02 4.15854193e-02 -5.37587583e-01 -5.22434056e-01 1.45626855e+00 -1.69566229e-01 -1.89979181e-01 5.72436869e-01 1.05958140e+00 1.98070869e-01 -9.83078420e-01 -7.14983419e-02 1.77207291e-01 -9.81231928e-02 -5.00747636e-02 -7.86467969e-01 -4.19450849e-01 9.71785307e-01 -5.44969916e-01 2.73281574e-01 9.35328662e-01 -1.46922633e-01 1.24241543e+00 6.35094404e-01 -1.19896777e-01 -1.28541255e+00 2.50888228e-01 7.50597954e-01 1.30884027e+00 -9.56400931e-01 5.40748835e-01 -8.92480314e-02 -7.06333816e-01 1.18674040e+00 3.89473110e-01 -4.72537965e-01 -4.45830673e-02 4.96583730e-02 -1.83360279e-01 -9.27038118e-02 -1.21248126e+00 -5.27096586e-03 3.11093360e-01 3.45368236e-01 7.80330181e-01 -1.27257451e-01 -8.07076097e-01 5.86408496e-01 -5.37926912e-01 4.77786660e-02 1.13458228e+00 1.07704866e+00 -6.30812764e-01 -1.09224474e+00 1.12022966e-01 6.26089334e-01 -6.82049096e-01 -4.15709645e-01 -8.63332033e-01 3.38943511e-01 -7.91518033e-01 1.31571138e+00 -2.90984325e-02 4.59647551e-02 4.33627039e-01 2.27476507e-01 6.22168064e-01 -1.12927425e+00 -8.91279280e-01 -1.31882086e-01 7.28056729e-01 -1.74446702e-01 -2.94341564e-01 -8.62017632e-01 -1.05563724e+00 -3.26174319e-01 -1.56975403e-01 3.23695600e-01 5.33032358e-01 9.23676491e-01 7.05487251e-01 5.77004492e-01 6.18973672e-01 -1.10034895e+00 -7.48466551e-01 -1.38960254e+00 -1.88343510e-01 2.19379127e-01 5.40022492e-01 3.08521181e-01 -1.59815118e-01 1.92780390e-01]
[12.445212364196777, 9.436189651489258]
3e2ddb7e-3032-455d-a072-ff1dc7b40a05
semi-supervised-anomaly-detection-on
2002.12011
null
https://arxiv.org/abs/2002.12011v1
https://arxiv.org/pdf/2002.12011v1.pdf
Semi-supervised Anomaly Detection on Attributed Graphs
We propose a simple yet effective method for detecting anomalous instances on an attribute graph with label information of a small number of instances. Although with standard anomaly detection methods it is usually assumed that instances are independent and identically distributed, in many real-world applications, instances are often explicitly connected with each other, resulting in so-called attributed graphs. The proposed method embeds nodes (instances) on the attributed graph in the latent space by taking into account their attributes as well as the graph structure based on graph convolutional networks (GCNs). To learn node embeddings specialized for anomaly detection, in which there is a class imbalance due to the rarity of anomalies, the parameters of a GCN are trained to minimize the volume of a hypersphere that encloses the node embeddings of normal instances while embedding anomalous ones outside the hypersphere. This enables us to detect anomalies by simply calculating the distances between the node embeddings and hypersphere center. The proposed method can effectively propagate label information on a small amount of nodes to unlabeled ones by taking into account the node's attributes, graph structure, and class imbalance. In experiments with five real-world attributed graph datasets, we demonstrate that the proposed method achieves better performance than various existing anomaly detection methods.
['Tomoharu Iwata', 'Yasuhiro Fujiwara', 'Atsutoshi Kumagai']
2020-02-27
null
null
null
null
['supervised-anomaly-detection', 'semi-supervised-anomaly-detection']
['computer-vision', 'computer-vision']
[-1.20548457e-02 4.45364058e-01 1.46022037e-01 -3.37196738e-01 4.34651792e-01 -3.22181553e-01 3.69613945e-01 7.50048637e-01 2.28337590e-02 2.10427523e-01 -3.59788507e-01 -2.23843515e-01 -1.05039753e-01 -1.34454906e+00 -4.01777208e-01 -9.26661432e-01 -4.83873367e-01 6.89602852e-01 1.18356161e-01 -9.12495255e-02 2.14181021e-01 6.56039357e-01 -1.29011762e+00 -4.40794826e-01 1.04879606e+00 9.48451400e-01 -6.74490035e-01 3.40943962e-01 -5.40338337e-01 5.85071683e-01 -8.16085875e-01 -2.80424327e-01 3.05576771e-01 -2.95755506e-01 -4.10999864e-01 4.32984531e-01 2.98021138e-01 3.94312330e-02 -5.23998678e-01 1.46846509e+00 4.42365445e-02 1.87736183e-01 8.68416488e-01 -1.88892543e+00 -7.72964299e-01 3.44586432e-01 -8.33138406e-01 4.16179270e-01 -5.00349794e-03 -9.84878391e-02 1.01439273e+00 -4.55965608e-01 2.88547397e-01 1.04303324e+00 4.15946573e-01 2.67645419e-01 -9.11572516e-01 -6.19833827e-01 5.82788587e-01 2.33346447e-01 -1.43781424e+00 1.06648862e-01 1.15851927e+00 -4.24100786e-01 5.11825979e-01 2.57206410e-01 6.26548886e-01 5.80255806e-01 7.31664300e-02 2.96270013e-01 2.33471841e-01 -1.96405262e-01 3.49475801e-01 9.25171226e-02 4.02466983e-01 8.83516848e-01 5.50758779e-01 -2.91086495e-01 -1.69520285e-02 -5.38421988e-01 2.81575650e-01 5.56296170e-01 -2.66156256e-01 -6.26023352e-01 -9.73451436e-01 9.11283195e-01 6.96698725e-01 2.30845436e-01 -2.78265804e-01 -2.04223484e-01 4.81747448e-01 3.88001710e-01 9.11553144e-01 1.76649362e-01 -7.46179968e-02 5.65469801e-01 -1.85040385e-01 -2.76463002e-01 6.64857030e-01 8.44054341e-01 9.25999105e-01 1.69595122e-01 1.01503134e-01 6.42762661e-01 5.30013263e-01 3.23071271e-01 4.59575832e-01 -6.44488484e-02 5.52681029e-01 1.66874540e+00 -2.21422791e-01 -1.70967746e+00 -3.66925985e-01 -6.03119731e-01 -1.18493629e+00 1.95339620e-02 3.45012099e-01 5.57468273e-02 -9.26414967e-01 1.50572586e+00 7.61582315e-01 6.93238616e-01 -4.11828160e-02 8.06955040e-01 6.65341675e-01 5.32600284e-01 1.89219732e-02 -1.32612213e-01 9.49928224e-01 -6.87072039e-01 -5.65191507e-01 -2.15643108e-01 1.00864744e+00 -1.19665973e-01 8.98749650e-01 -3.76881729e-03 -3.73652667e-01 -1.20438421e-02 -1.09557307e+00 3.89042467e-01 -6.64258778e-01 -2.30045781e-01 4.74862427e-01 4.73838419e-01 -7.26161897e-01 6.02518618e-01 -6.55803084e-01 -3.35650146e-01 4.29658145e-01 2.55785227e-01 -6.34282947e-01 -1.92504805e-02 -1.07196140e+00 2.79082447e-01 4.07422870e-01 2.90667832e-01 -5.26572108e-01 -2.25909933e-01 -1.08623183e+00 3.14789176e-01 4.15943682e-01 -1.20026864e-01 3.29302609e-01 -9.61114883e-01 -6.79941177e-01 7.36532986e-01 1.45169944e-01 -1.60782501e-01 1.89089343e-01 2.56395757e-01 -8.97332430e-01 9.74496175e-03 9.74861458e-02 -1.72252879e-01 8.62637818e-01 -7.91541457e-01 -4.31987375e-01 -9.60243523e-01 -1.11475773e-01 1.64388955e-01 -1.00235748e+00 -4.98392463e-01 -6.57725856e-02 -4.80386913e-01 7.27393270e-01 -8.72957289e-01 -1.32121846e-01 3.92352045e-03 -7.89394796e-01 -6.43432856e-01 1.39155245e+00 -2.43227169e-01 1.25088906e+00 -2.43460155e+00 6.37664124e-02 7.66063690e-01 8.75329733e-01 8.77565295e-02 -9.33057517e-02 3.78301084e-01 -3.25470269e-01 1.23800181e-01 -6.06302142e-01 -2.61883792e-02 -2.61741906e-01 3.84366572e-01 -2.97513336e-01 1.01635468e+00 3.24290060e-02 4.23121959e-01 -1.07929802e+00 -2.27279380e-01 2.09091365e-01 2.00517967e-01 -1.84124544e-01 3.30876112e-01 -6.74127117e-02 3.61652225e-01 -7.54633307e-01 5.86823702e-01 7.99445868e-01 -4.02536333e-01 -1.76415835e-02 2.60356843e-01 4.16472733e-01 -5.35682263e-03 -1.27366984e+00 9.10253704e-01 3.51251028e-02 5.26143968e-01 -1.93712175e-01 -1.47402918e+00 1.26949167e+00 1.68723181e-01 4.52110797e-01 -2.55545765e-01 -1.44557923e-01 6.52537122e-02 1.26303494e-01 -3.32592249e-01 -1.29909813e-01 2.22278550e-01 1.63447648e-01 6.05535746e-01 -1.99182183e-01 5.18800318e-01 1.81557477e-01 6.14658713e-01 1.34733534e+00 -6.51222408e-01 8.04522485e-02 -2.04783618e-01 9.27255511e-01 -3.45070302e-01 7.30798483e-01 2.68907100e-01 -2.44587958e-01 3.65038872e-01 9.64374125e-01 -8.32907736e-01 -8.11096549e-01 -9.20016944e-01 -1.91545501e-01 8.56087685e-01 3.94984692e-01 -4.56755549e-01 -5.61645269e-01 -1.34913325e+00 3.56009424e-01 6.06722653e-01 -7.96258330e-01 -6.80700302e-01 -2.29173332e-01 -9.78979349e-01 2.46473148e-01 2.77117252e-01 3.43745172e-01 -1.13372028e+00 1.82609156e-01 -8.88971612e-02 1.05540514e-01 -9.21323121e-01 -3.28976452e-01 -1.75874293e-01 -9.02710438e-01 -1.47746682e+00 -1.55298486e-01 -5.88878989e-01 1.47191906e+00 1.29461527e-01 8.15495968e-01 7.87476301e-01 -4.39557523e-01 3.13946724e-01 -3.84130120e-01 -3.75571877e-01 -3.92162234e-01 -2.04567965e-02 2.23776460e-01 6.05907440e-01 7.07720280e-01 -7.58457065e-01 -5.96999347e-01 3.72704804e-01 -1.06781292e+00 -4.90575016e-01 1.76055089e-01 4.45313215e-01 5.32523215e-01 4.54827845e-01 5.52225590e-01 -1.22330534e+00 5.01721621e-01 -9.94120836e-01 -7.08178282e-01 1.72152817e-01 -1.02709961e+00 3.59657966e-02 8.47651899e-01 -1.37616694e-01 -4.15595382e-01 -3.41334224e-01 3.15977544e-01 -4.72820371e-01 -2.42631510e-01 4.05140072e-01 -2.63951212e-01 -9.57192928e-02 5.31831205e-01 1.68190226e-01 1.89701900e-01 -2.09670812e-01 1.28703177e-01 5.90667307e-01 1.20181851e-01 -8.14127773e-02 1.09155452e+00 4.44129765e-01 4.08753306e-01 -7.92526841e-01 -6.35290980e-01 -5.56077003e-01 -7.24392831e-01 -1.18170530e-01 4.73857433e-01 -4.73314494e-01 -6.19479060e-01 4.91417408e-01 -6.41672671e-01 2.38092184e-01 -2.74182796e-01 3.52081805e-01 2.01768845e-01 6.61857128e-01 -5.09414911e-01 -6.20011985e-01 -3.15068841e-01 -6.79563105e-01 7.50774801e-01 2.17884958e-01 1.59018964e-01 -1.35918164e+00 1.67190298e-01 -1.87890619e-01 1.68712303e-01 5.65609515e-01 1.40647244e+00 -1.35366619e+00 -3.92742425e-01 -7.89468825e-01 -1.40234873e-01 2.79323637e-01 5.17872453e-01 1.30554914e-01 -7.48050749e-01 -5.19508541e-01 1.43520953e-03 2.05462143e-01 6.64807856e-01 6.17713621e-03 1.31947303e+00 -4.08874810e-01 -7.05991328e-01 7.56954849e-01 1.16776037e+00 -5.64926565e-02 4.84763354e-01 2.11151391e-01 1.17274332e+00 6.41714871e-01 3.21221828e-01 4.11812842e-01 1.10767014e-01 1.84133008e-01 9.56114531e-01 5.66599378e-03 4.50929314e-01 -1.16791375e-01 6.56584874e-02 8.47872615e-01 8.85563567e-02 -4.39728349e-01 -9.61016595e-01 4.15216565e-01 -1.74130774e+00 -5.18829584e-01 -4.45688158e-01 2.46676612e+00 1.18585318e-01 1.70490950e-01 -4.93898392e-02 2.93205827e-01 1.15380764e+00 3.74517888e-01 -8.76734018e-01 -2.49536946e-01 -5.94435818e-03 -3.97729963e-01 2.78207839e-01 2.22265497e-01 -1.12115133e+00 4.59389359e-01 4.68403816e+00 3.12860638e-01 -9.54602718e-01 -2.34451920e-01 6.76003158e-01 1.46754265e-01 -4.33912456e-01 1.87077504e-02 -4.99803036e-01 6.69436276e-01 6.38348103e-01 -2.91473567e-01 3.11414868e-01 9.88829911e-01 -3.45029324e-01 3.51667106e-01 -1.04090858e+00 6.32374406e-01 1.12987228e-01 -7.45453358e-01 1.06765866e-01 2.62520283e-01 4.91979897e-01 -2.83170454e-02 3.10237091e-02 3.42199922e-01 1.97492633e-02 -9.61597681e-01 -1.88800588e-01 1.71899199e-01 5.86843908e-01 -8.92048061e-01 8.43089700e-01 3.67909580e-01 -1.26817477e+00 -1.96480259e-01 -6.82118475e-01 -1.16826277e-02 -3.78450722e-01 1.01951802e+00 -8.95847857e-01 3.98199052e-01 5.42111456e-01 1.00576115e+00 -7.62502968e-01 8.99296880e-01 -3.78450930e-01 5.71868718e-01 -3.56319785e-01 2.54948556e-01 1.98871017e-01 -9.16422844e-01 9.29427207e-01 4.74953204e-01 3.08472216e-01 9.46411639e-02 2.05520526e-01 8.81169915e-01 -2.42168739e-01 4.48974967e-01 -1.13620996e+00 -2.11476937e-01 5.78396678e-01 1.54423785e+00 -1.03805828e+00 -2.58173287e-01 -5.06949008e-01 8.12279880e-01 5.85764825e-01 4.54299718e-01 -4.97794598e-01 -6.82883978e-01 6.20046079e-01 2.95889705e-01 -1.63981721e-01 1.41524404e-01 9.63717625e-02 -1.09104419e+00 2.74445474e-01 -3.88798147e-01 8.85259807e-01 -2.35129088e-01 -1.59357917e+00 7.22343564e-01 -4.27164704e-01 -1.38113523e+00 -4.63466085e-02 -5.89303195e-01 -1.16206849e+00 7.45439708e-01 -1.16725850e+00 -6.70781314e-01 -6.60240293e-01 7.87921309e-01 -1.41985953e-01 -4.02482688e-01 9.07821834e-01 1.05068646e-01 -9.37046111e-01 5.42550266e-01 1.31863281e-01 5.69781303e-01 5.07533610e-01 -1.35713434e+00 5.43113410e-01 8.32767189e-01 2.51986802e-01 4.35246915e-01 3.53142291e-01 -9.62450564e-01 -1.01471007e+00 -1.25330412e+00 5.27468979e-01 -1.48129612e-01 7.21075296e-01 -4.06828672e-01 -1.38946903e+00 1.02124345e+00 -1.48374185e-01 8.83459866e-01 6.36876464e-01 2.88694173e-01 -3.96150261e-01 -2.09093168e-01 -1.33089316e+00 3.83087486e-01 9.48673189e-01 -3.01577628e-01 -2.87323475e-01 7.78332531e-01 6.27248824e-01 -1.89849183e-01 -6.26650691e-01 6.03597164e-01 -2.08992928e-01 -8.46740246e-01 6.76269293e-01 -9.04301643e-01 -1.93357300e-02 -4.05593336e-01 2.89041758e-01 -1.54487705e+00 -2.54967123e-01 -1.11670941e-01 -3.90402138e-01 1.11642754e+00 1.82005301e-01 -1.22376490e+00 8.97015333e-01 5.74927092e-01 -6.73165247e-02 -6.98033690e-01 -9.00873184e-01 -4.87943769e-01 -4.20458734e-01 1.00866616e-01 8.56153011e-01 1.43727827e+00 -8.16730261e-02 3.06225985e-01 8.71030763e-02 8.38740885e-01 9.00495946e-01 1.33489922e-01 5.01003683e-01 -1.93103683e+00 3.52050841e-01 -2.92403936e-01 -1.18799067e+00 -2.33199418e-01 3.60126168e-01 -1.23932111e+00 -4.89148885e-01 -1.29781318e+00 -7.13200793e-02 -6.91392660e-01 -6.09875143e-01 5.37866950e-01 -3.54092747e-01 3.86297889e-02 -4.92516577e-01 3.11599910e-01 -4.35580194e-01 7.16304362e-01 8.57886910e-01 -2.31262237e-01 -2.36222431e-01 1.87066630e-01 -3.47045004e-01 9.65730190e-01 7.39676416e-01 -6.14607871e-01 -5.95830560e-01 -1.92661002e-01 1.63975209e-01 -2.67958552e-01 3.09905946e-01 -8.70403826e-01 1.90256834e-01 5.05499281e-02 3.75601113e-01 -3.60977709e-01 -1.99574471e-01 -1.02061236e+00 -1.87707603e-01 2.46451452e-01 -2.01503411e-01 2.27190286e-01 -8.64746422e-02 1.14623892e+00 -2.66514540e-01 -1.50379866e-01 5.25237322e-01 1.15592673e-01 -5.01620889e-01 8.68171632e-01 7.27653876e-02 1.67254105e-01 1.44897389e+00 -4.53967005e-02 -4.83155698e-01 -3.41749668e-01 -7.08933413e-01 4.81388658e-01 4.80965674e-01 4.33942288e-01 7.41522491e-01 -1.45533121e+00 -6.32568955e-01 6.91540658e-01 5.54697990e-01 2.82618314e-01 3.07064980e-01 6.36424363e-01 -4.97529238e-01 -3.49150002e-02 -1.17016308e-01 -7.82680690e-01 -1.08867240e+00 7.54297078e-01 3.67992252e-01 -3.12642828e-02 -1.05403626e+00 7.44017959e-01 4.22376484e-01 -7.39302337e-01 2.36338839e-01 2.71969855e-01 -5.04254460e-01 -1.46984484e-03 3.05656552e-01 4.15556699e-01 9.99804959e-02 -6.95695877e-01 -4.11827296e-01 3.60472471e-01 -3.29078615e-01 6.06556177e-01 1.00996542e+00 2.23126933e-02 -6.80790067e-01 4.33797657e-01 1.16281462e+00 1.07323460e-01 -7.58899152e-01 -3.37707192e-01 2.39415988e-02 -6.53365016e-01 -8.37625787e-02 -5.14876191e-03 -1.64646316e+00 9.11849499e-01 5.76278389e-01 7.60046542e-01 9.79828835e-01 -1.24048721e-02 5.13069630e-01 3.59695494e-01 1.15052812e-01 -8.42045963e-01 2.67944522e-02 1.18232086e-01 4.81310755e-01 -1.17124224e+00 -1.21849589e-01 -5.99162340e-01 -4.15618241e-01 1.17448378e+00 1.08300841e+00 -3.73498440e-01 8.34690154e-01 -2.93276846e-01 1.59731489e-02 -6.00628436e-01 -2.98531175e-01 3.06323707e-01 2.45316043e-01 4.52243090e-01 1.41847776e-02 1.69534802e-01 8.09351262e-03 1.14169791e-01 1.49911165e-01 -8.26333225e-01 5.71768224e-01 5.63829303e-01 -5.12600899e-01 -7.91917801e-01 -2.50757545e-01 9.11215842e-01 -3.23462099e-01 3.82426754e-02 -3.64753157e-01 6.68334484e-01 -1.15207344e-01 7.79933751e-01 6.12369955e-01 -2.31671825e-01 2.54932672e-01 1.85443223e-01 -1.90417409e-01 -7.76165724e-01 2.33730599e-02 -5.88551462e-01 -4.12978262e-01 -3.82557184e-01 9.74601209e-02 -2.31307998e-01 -1.44033730e+00 -2.82219261e-01 -4.60432142e-01 5.57404280e-01 3.11480999e-01 9.75822806e-01 5.49188614e-01 4.81227309e-01 9.77230430e-01 -8.94512385e-02 -3.57793450e-01 -9.85570192e-01 -1.22094560e+00 8.86274219e-01 4.12767172e-01 -7.43971407e-01 -1.08290637e+00 -5.69199979e-01]
[6.6659016609191895, 5.79884672164917]
912afabe-6b5c-4e86-88ff-a7646d196f02
show-me-how-to-revise-improving-lexically
2109.05797
null
https://arxiv.org/abs/2109.05797v1
https://arxiv.org/pdf/2109.05797v1.pdf
Show Me How To Revise: Improving Lexically Constrained Sentence Generation with XLNet
Lexically constrained sentence generation allows the incorporation of prior knowledge such as lexical constraints into the output. This technique has been applied to machine translation, and dialog response generation. Previous work usually used Markov Chain Monte Carlo (MCMC) sampling to generate lexically constrained sentences, but they randomly determined the position to be edited and the action to be taken, resulting in many invalid refinements. To overcome this challenge, we used a classifier to instruct the MCMC-based models where and how to refine the candidate sentences. First, we developed two methods to create synthetic data on which the pre-trained model is fine-tuned to obtain a reliable classifier. Next, we proposed a two-step approach, "Predict and Revise", for constrained sentence generation. During the predict step, we leveraged the classifier to compute the learned prior for the candidate sentence. During the revise step, we resorted to MCMC sampling to revise the candidate sentence by conducting a sampled action at a sampled position drawn from the learned prior. We compared our proposed models with many strong baselines on two tasks, generating sentences with lexical constraints and text infilling. Experimental results have demonstrated that our proposed model performs much better than the previous work in terms of sentence fluency and diversity. Our code and pre-trained models are available at https://github.com/NLPCode/MCMCXLNet.
['Victor O. K. Li', 'Xingwei He']
2021-09-13
null
null
null
null
['text-infilling']
['natural-language-processing']
[ 6.70403302e-01 2.74912685e-01 -2.15761051e-01 -5.75798631e-01 -1.16469932e+00 -6.27667844e-01 9.91249740e-01 1.02429770e-01 -4.77710038e-01 1.22244990e+00 4.65497553e-01 -5.58626950e-01 5.81267059e-01 -6.40943885e-01 -5.57305336e-01 -3.84996116e-01 5.85289240e-01 8.36987257e-01 2.12476924e-01 -1.88343823e-01 6.30215764e-01 4.62549180e-02 -1.20676732e+00 6.98951960e-01 1.03489149e+00 2.04386517e-01 6.81822836e-01 8.16941679e-01 -2.54773587e-01 6.47440374e-01 -8.17261577e-01 -4.45628881e-01 -1.59560412e-01 -1.01008880e+00 -1.03664172e+00 9.84205380e-02 -2.52447069e-01 -1.82742998e-01 4.92871791e-01 8.82927597e-01 5.73592722e-01 3.84844691e-01 8.25736463e-01 -8.07418585e-01 -4.46375698e-01 1.05031526e+00 -1.76032826e-01 6.80309236e-02 8.13194692e-01 1.53869152e-01 7.26676524e-01 -1.14055037e+00 7.78123617e-01 1.42375159e+00 3.68332297e-01 1.02464294e+00 -1.22868657e+00 -4.91791308e-01 1.47593766e-01 3.72444056e-02 -1.13888931e+00 -5.91055095e-01 5.79202533e-01 -4.99949932e-01 9.17265892e-01 3.39173168e-01 6.46677792e-01 1.33366454e+00 2.14144304e-01 6.95446849e-01 1.24750245e+00 -1.05289900e+00 5.57835042e-01 3.83643687e-01 -2.13653788e-01 6.41665101e-01 -3.35371047e-01 -3.08816005e-02 -4.42893863e-01 -4.44670796e-01 4.07602936e-01 -5.19623518e-01 -1.45549268e-01 4.05189097e-01 -1.18366182e+00 1.21152616e+00 -2.19949663e-01 1.46285102e-01 -4.76367950e-01 -6.78563043e-02 3.07457507e-01 7.84844309e-02 5.19064546e-01 6.43474460e-01 -2.68477231e-01 -3.43002170e-01 -1.15364659e+00 3.29587668e-01 9.54434872e-01 9.84870672e-01 6.21359706e-01 -2.94431001e-01 -7.63605833e-01 8.13100457e-01 2.92376250e-01 2.66549498e-01 5.46277046e-01 -9.71178591e-01 7.34617472e-01 3.11483800e-01 4.68337804e-01 -4.95032519e-01 1.07663997e-01 1.89867783e-02 -5.32712817e-01 -1.71328798e-01 2.88841128e-01 -4.76561308e-01 -1.02159464e+00 1.63604999e+00 3.89335215e-01 2.10652035e-02 9.68301967e-02 7.98040986e-01 2.81653583e-01 8.36811006e-01 1.51199177e-01 -4.64473575e-01 1.25409174e+00 -9.72685277e-01 -7.60591984e-01 -1.80623502e-01 6.93223417e-01 -1.26381528e+00 1.33646786e+00 2.07221106e-01 -1.13195074e+00 -5.28251648e-01 -8.90280485e-01 2.33427718e-01 9.32090729e-02 2.85580665e-01 3.71697783e-01 5.62120259e-01 -9.98483121e-01 6.13853455e-01 -8.56824338e-01 -3.07603091e-01 5.56836873e-02 4.99899387e-02 1.97568730e-01 1.48575768e-01 -1.45760405e+00 9.67172742e-01 4.58692372e-01 1.60533965e-01 -9.19039249e-01 -1.49500325e-01 -6.83563292e-01 -1.91168860e-01 2.24839896e-01 -9.36759531e-01 1.84331763e+00 -8.90289724e-01 -2.17010760e+00 5.16148984e-01 -5.87143540e-01 -4.47242409e-01 5.30636370e-01 -8.87386203e-02 -4.22003642e-02 4.69188206e-02 2.30055898e-01 9.62986588e-01 7.84948170e-01 -1.09095657e+00 -6.09510839e-01 3.42206806e-01 7.22638369e-02 5.01751542e-01 1.70762554e-01 1.75410494e-01 -4.55606163e-01 -5.04392028e-01 -4.91392650e-02 -1.09313035e+00 -4.38974082e-01 -8.86518240e-01 -7.10043669e-01 -3.54213387e-01 2.74464488e-01 -5.79684913e-01 1.46707106e+00 -1.54010391e+00 1.97915077e-01 1.30248023e-02 -4.19287145e-01 3.38206828e-01 -1.58283144e-01 8.47010195e-01 3.12750965e-01 3.48579019e-01 -2.55428910e-01 -6.32192910e-01 -4.51992713e-02 7.09607452e-02 -4.87457424e-01 -2.82811858e-02 2.53830105e-01 7.07785726e-01 -9.60259676e-01 -6.43977225e-01 1.18837222e-01 2.70444661e-01 -8.55653346e-01 4.43773240e-01 -7.53547728e-01 8.01752627e-01 -5.87041855e-01 2.44637311e-01 3.99475515e-01 -2.24359572e-01 3.68131667e-01 1.83864310e-01 -8.32429975e-02 7.02298224e-01 -1.05370879e+00 1.64717555e+00 -6.62495792e-01 3.18712205e-01 -3.97630513e-01 -5.36610663e-01 9.15461123e-01 4.91209209e-01 -2.59668738e-01 -2.20606402e-01 1.65418833e-01 2.44775534e-01 -9.63065028e-02 -4.68011349e-01 7.44756341e-01 -2.55689174e-01 -1.80737272e-01 6.84786379e-01 -2.74246603e-01 -5.22779167e-01 3.78837019e-01 3.85629386e-01 7.69426286e-01 5.39730370e-01 3.83675903e-01 6.78230971e-02 8.00006032e-01 2.62909263e-01 5.24719894e-01 9.16728377e-01 3.07070334e-02 6.43555403e-01 4.39135641e-01 -1.36384293e-01 -1.03216660e+00 -7.06129909e-01 1.82934582e-01 1.06234586e+00 -1.99050039e-01 -4.72407281e-01 -1.21243501e+00 -7.05868125e-01 -5.73322356e-01 1.36554992e+00 -2.83307403e-01 -1.08106267e-02 -8.15498471e-01 -6.16829157e-01 3.75019819e-01 2.13285968e-01 3.18539679e-01 -1.42057693e+00 -5.62394917e-01 4.58537847e-01 -6.62983418e-01 -8.29210818e-01 -8.85486960e-01 -9.61276442e-02 -7.52669454e-01 -6.67601764e-01 -4.99934465e-01 -8.17138433e-01 9.03104782e-01 -2.51732826e-01 9.75126266e-01 1.64125890e-01 2.79132426e-01 -9.97023210e-02 -5.79954028e-01 -2.23061025e-01 -1.17672062e+00 3.54040504e-01 1.20588960e-02 -1.37690291e-01 2.96743423e-01 -2.62159795e-01 -5.45387805e-01 2.76002288e-01 -7.28726804e-01 6.34190440e-01 6.30271494e-01 1.03320563e+00 5.96266329e-01 -4.52977985e-01 7.56517649e-01 -1.28384125e+00 1.13002872e+00 -2.52874851e-01 -4.55421448e-01 2.75251210e-01 -6.23418391e-01 3.54798347e-01 8.15981448e-01 -5.62261283e-01 -1.39282870e+00 2.49173373e-01 -3.82457614e-01 1.08226441e-01 -2.30053589e-01 6.95841968e-01 5.35029778e-03 7.46651828e-01 5.88053286e-01 2.41007879e-01 -1.94035009e-01 -3.73890162e-01 4.02633518e-01 1.04688823e+00 1.19733758e-01 -8.45290899e-01 4.77771878e-01 -1.35379612e-01 -4.75137204e-01 -3.00164729e-01 -7.94067919e-01 2.77906545e-02 -5.31164169e-01 -1.37756586e-01 9.42936659e-01 -7.41205096e-01 -2.23233178e-01 1.75345197e-01 -1.54600525e+00 -5.48421323e-01 -4.29673977e-02 6.43219411e-01 -6.40799701e-01 2.64798969e-01 -6.46820903e-01 -1.02696955e+00 -5.13941586e-01 -1.25516677e+00 1.02775598e+00 2.79140472e-01 -8.14759731e-01 -9.45963979e-01 1.86412096e-01 5.42716980e-01 3.05442244e-01 -1.39194755e-02 9.91427004e-01 -7.50635266e-01 -5.10795295e-01 -1.36423647e-01 3.75127256e-01 5.33273518e-02 7.65629858e-02 3.55822854e-02 -7.16935396e-01 -2.34968051e-01 -4.05895859e-02 -3.62994820e-01 4.63789701e-01 1.38999641e-01 6.90929055e-01 -5.19761384e-01 -4.18750703e-01 1.13583751e-01 9.64781761e-01 3.61497790e-01 5.71925402e-01 1.18733272e-01 2.12244347e-01 4.10740048e-01 8.70375633e-01 4.85434681e-01 3.45631897e-01 8.02726209e-01 -1.60786003e-01 2.30484277e-01 -1.09172739e-01 -7.24896491e-01 5.53050101e-01 1.09772205e+00 1.76122323e-01 -5.03123522e-01 -7.04191625e-01 3.56928766e-01 -1.80257821e+00 -9.52971101e-01 1.21870406e-01 2.05243134e+00 1.56148779e+00 3.94393116e-01 -1.01399221e-01 -1.31790444e-01 1.02037990e+00 -1.01763137e-01 -2.09892243e-01 -7.99257100e-01 2.30914026e-01 3.07968259e-01 2.84257531e-02 1.17501247e+00 -6.93157911e-01 1.43735135e+00 5.72583342e+00 8.72128129e-01 -9.99385118e-01 2.26649910e-01 5.88236034e-01 -1.18938476e-01 -5.54353178e-01 4.89710540e-01 -1.24410713e+00 6.33279383e-01 1.23093903e+00 -1.02613792e-01 4.41762269e-01 3.99220586e-01 8.11286807e-01 -3.85288835e-01 -1.11814117e+00 2.89776832e-01 1.32357525e-02 -1.41483021e+00 3.11742544e-01 -2.43710741e-01 8.78643870e-01 -4.17017579e-01 -3.77871066e-01 5.33830166e-01 5.91265917e-01 -8.34307313e-01 7.85804749e-01 5.48321068e-01 7.23474562e-01 -6.00029826e-01 6.20693862e-01 8.01230252e-01 -8.10613096e-01 3.26400548e-01 -2.32888937e-01 -2.43359685e-01 4.18448865e-01 5.05919635e-01 -1.65486264e+00 3.45278472e-01 -4.96854186e-02 8.41636211e-02 -3.26084614e-01 4.85298693e-01 -8.43413889e-01 9.36507642e-01 -3.25658880e-02 -5.34304440e-01 1.19742543e-01 -2.33295411e-01 5.30657053e-01 1.29697907e+00 2.95384735e-01 1.98496971e-02 4.12719995e-01 9.75839853e-01 1.22332677e-01 3.10729593e-01 -1.98241159e-01 4.09285538e-02 9.47290361e-01 1.09880757e+00 -8.90463829e-01 -5.70121109e-01 1.64539516e-02 1.22225702e+00 4.13727432e-01 3.67644191e-01 -9.02506888e-01 -4.14332688e-01 -7.44118020e-02 6.71201348e-02 3.63616258e-01 -1.07834257e-01 -2.19060972e-01 -1.12082994e+00 -1.12436175e-01 -1.10372257e+00 1.10554375e-01 -6.46723866e-01 -8.86242509e-01 9.15988982e-01 1.42696112e-01 -9.70088065e-01 -9.78105366e-01 1.24298502e-02 -8.37346375e-01 1.44868028e+00 -1.23058355e+00 -8.57559085e-01 2.47205034e-01 1.26237035e-01 1.03727698e+00 -1.28974974e-01 1.00675666e+00 -1.36501789e-01 -6.20981693e-01 5.71966290e-01 -3.38297039e-01 -6.92451075e-02 7.21083581e-01 -1.14163268e+00 5.24647057e-01 9.00700450e-01 -3.01703643e-02 8.00537646e-01 9.79259431e-01 -1.04259479e+00 -9.57532346e-01 -1.03519404e+00 1.38917291e+00 -5.11284709e-01 3.01962465e-01 -6.09624922e-01 -6.15551054e-01 4.93668258e-01 4.43952739e-01 -7.49229550e-01 6.41769707e-01 -2.12773472e-01 3.02289665e-01 4.09646899e-01 -1.02905560e+00 8.99461865e-01 6.45558298e-01 -3.08064312e-01 -6.90833330e-01 4.99456048e-01 8.26955140e-01 -5.92122912e-01 -4.83279049e-01 4.75727879e-02 3.24744970e-01 -6.37850881e-01 3.65023255e-01 -5.10603189e-01 5.56221426e-01 -3.84524494e-01 -8.15726444e-02 -1.62482250e+00 -1.25373006e-01 -1.10008693e+00 -7.32210139e-03 1.34708512e+00 1.03771842e+00 -4.65631455e-01 7.22127616e-01 6.21526718e-01 -2.68729776e-01 -1.04280055e+00 -6.35257781e-01 -4.01418209e-01 1.73685014e-01 -1.81860447e-01 6.79663420e-01 5.22798121e-01 9.02422145e-02 7.22299099e-01 -4.00451660e-01 2.70847864e-02 1.84400082e-01 2.28322521e-01 7.39773393e-01 -6.55262530e-01 -5.66412151e-01 -1.15966477e-01 4.58595604e-01 -1.24234331e+00 1.66109532e-01 -1.00656223e+00 4.66410637e-01 -1.66937172e+00 2.23203510e-01 -2.77047932e-01 2.92677432e-01 3.19668770e-01 -6.86371088e-01 3.38844676e-03 2.01211601e-01 3.08815926e-01 -2.85562426e-01 5.07774889e-01 1.39450192e+00 1.55867130e-01 -6.05360031e-01 3.16673428e-01 -6.80752933e-01 4.67585057e-01 1.10529077e+00 -6.48765266e-01 -3.93320978e-01 -2.93173075e-01 1.97898284e-01 3.36679220e-01 -6.90783262e-02 -6.65980041e-01 2.80498922e-01 -4.21970040e-01 2.94165134e-01 -6.87671840e-01 3.42236429e-01 -9.22330394e-02 1.66564569e-01 5.44752240e-01 -8.21859658e-01 1.17821760e-01 -6.59900308e-02 3.18265766e-01 2.01171301e-02 -7.88353860e-01 7.33291149e-01 -3.20098162e-01 -1.77279562e-01 -1.42407119e-01 -8.51597726e-01 9.86160114e-02 8.19234312e-01 -7.51874745e-02 2.36967509e-03 -5.31116307e-01 -6.99555337e-01 2.05848768e-01 3.50651711e-01 3.28044742e-01 6.69773817e-01 -1.07497215e+00 -8.32045555e-01 1.22473769e-01 -2.53714234e-01 -6.01350367e-02 -2.05163702e-01 6.74990773e-01 -4.14361417e-01 4.71319616e-01 2.46766314e-01 -4.84832883e-01 -1.02240431e+00 3.29538822e-01 1.44571334e-01 -4.82575327e-01 -3.54649842e-01 7.10319102e-01 -4.13492829e-01 -5.06662428e-01 1.26426533e-01 -2.77200133e-01 -2.01737598e-01 -1.32756382e-02 3.87953639e-01 1.12129953e-02 -2.79315282e-02 -2.60754317e-01 -1.87046424e-01 2.02007726e-01 -4.54301327e-01 -7.96445847e-01 8.49473953e-01 -3.80334646e-01 7.29420856e-02 4.67134178e-01 7.42565691e-01 1.98284984e-01 -1.13542259e+00 -1.09579064e-01 1.71199933e-01 -2.76742280e-01 -3.07704002e-01 -1.06975782e+00 -1.95639133e-01 6.13931477e-01 1.27737656e-01 1.43583701e-03 8.42488408e-01 -2.23799452e-01 8.28064084e-01 3.51713568e-01 3.10869992e-01 -1.31212151e+00 1.03515103e-01 8.41063023e-01 8.14196467e-01 -1.08003116e+00 -1.14017084e-01 -4.54506040e-01 -1.02654386e+00 9.33643639e-01 6.43092811e-01 1.65083170e-01 2.42601335e-01 1.47838429e-01 2.57654786e-01 2.28248790e-01 -1.15530109e+00 1.77194059e-01 8.00663754e-02 2.44899943e-01 7.46695638e-01 2.35890567e-01 -8.53601635e-01 5.20371675e-01 -5.66413939e-01 2.03622043e-01 7.21418440e-01 9.82872963e-01 -5.64279795e-01 -1.60143876e+00 -3.01772624e-01 3.34970951e-01 -3.32467467e-01 -4.18132991e-01 -5.57593942e-01 2.88074911e-01 1.24086356e-02 1.33727753e+00 -1.49756297e-01 -2.82440811e-01 1.71455324e-01 5.03484488e-01 3.94959152e-01 -1.19020927e+00 -7.92019784e-01 2.23008871e-01 5.43818116e-01 -1.40987024e-01 -1.78617030e-01 -8.59803379e-01 -1.38006091e+00 6.00471646e-02 -5.70642650e-01 6.44819736e-01 6.32692575e-01 1.01156878e+00 4.22707349e-01 3.79217416e-01 8.87846589e-01 -8.64899874e-01 -9.48171258e-01 -1.44811523e+00 1.58875942e-01 2.48881117e-01 -1.08226202e-01 -3.01209718e-01 -3.31214666e-01 1.62945867e-01]
[11.879053115844727, 8.957694053649902]
ba8dd14d-7af1-4227-b050-4f501f9c5592
to-wake-up-or-not-to-wake-up-reducing-keyword
2304.03416
null
https://arxiv.org/abs/2304.03416v1
https://arxiv.org/pdf/2304.03416v1.pdf
To Wake-up or Not to Wake-up: Reducing Keyword False Alarm by Successive Refinement
Keyword spotting systems continuously process audio streams to detect keywords. One of the most challenging tasks in designing such systems is to reduce False Alarm (FA) which happens when the system falsely registers a keyword despite the keyword not being uttered. In this paper, we propose a simple yet elegant solution to this problem that follows from the law of total probability. We show that existing deep keyword spotting mechanisms can be improved by Successive Refinement, where the system first classifies whether the input audio is speech or not, followed by whether the input is keyword-like or not, and finally classifies which keyword was uttered. We show across multiple models with size ranging from 13K parameters to 2.41M parameters, the successive refinement technique reduces FA by up to a factor of 8 on in-domain held-out FA data, and up to a factor of 7 on out-of-domain (OOD) FA data. Further, our proposed approach is "plug-and-play" and can be applied to any deep keyword spotting model.
['Hongxia Jin', 'Yilin Shen', 'Chouchang Yang', 'Ching-Hua Lee', 'Rakshith Sharma Srinivasa', 'Yashas Malur Saidutta']
2023-04-06
null
null
null
null
['keyword-spotting']
['speech']
[ 2.58627206e-01 -3.45738344e-02 1.37517676e-01 -1.75795987e-01 -1.09962308e+00 -7.34559298e-01 2.92039186e-01 4.66477066e-01 -5.68658650e-01 4.54630166e-01 -1.25617862e-01 -6.72145963e-01 -1.51535928e-01 -4.14320946e-01 -7.65430689e-01 -5.32550395e-01 -1.26074642e-01 3.74548405e-01 7.02192783e-01 2.12209076e-02 1.21259235e-01 4.62000787e-01 -1.58274126e+00 3.06483090e-01 1.36401922e-01 1.32229292e+00 2.02792168e-01 1.32579219e+00 -4.31588143e-02 7.25273371e-01 -9.92156267e-01 -5.84097430e-02 1.56653762e-01 -3.17002088e-01 -8.61935794e-01 5.14657460e-02 2.54144043e-01 -5.37697077e-01 -4.66006398e-01 8.86453032e-01 5.60652673e-01 -9.59176719e-02 2.44286180e-01 -1.31206465e+00 1.39913678e-01 7.11725116e-01 -1.97753191e-01 5.05502522e-01 4.20694858e-01 -4.11048084e-02 1.16350698e+00 -9.95332539e-01 4.94815819e-02 1.20434654e+00 3.08528841e-01 3.85228395e-01 -1.30803788e+00 -9.04700279e-01 2.41862446e-01 1.41592892e-02 -1.69167483e+00 -7.44627774e-01 3.80230039e-01 -2.07414091e-01 1.13620377e+00 4.81418550e-01 2.73765028e-01 7.67741263e-01 -8.71496797e-02 8.39014649e-01 6.24762058e-01 -5.11549950e-01 2.30739519e-01 3.01058944e-02 1.25351220e-01 4.19527054e-01 -1.22587919e-01 -1.42303988e-01 -8.68460000e-01 -5.10583818e-01 2.91893333e-01 -3.98074329e-01 -3.58482152e-01 2.35827088e-01 -1.09407365e+00 5.64144194e-01 -4.42347467e-01 1.82848468e-01 -2.59977072e-01 2.06853464e-01 2.97072202e-01 5.79835296e-01 2.56774038e-01 4.15796757e-01 -5.21269500e-01 -5.39480031e-01 -1.41892385e+00 3.58435452e-01 8.62125576e-01 7.65450299e-01 5.19077301e-01 1.29910365e-01 -8.24777931e-02 8.80986154e-01 1.47809863e-01 5.27353466e-01 5.93081295e-01 -8.72276843e-01 2.13755831e-01 9.06204954e-02 3.07926983e-01 -7.11784422e-01 -2.18934163e-01 -4.08970147e-01 -4.45696741e-01 -4.85363677e-02 3.75728190e-01 -2.12432042e-01 -7.89435804e-01 1.65992308e+00 1.93173632e-01 1.83234915e-01 -8.46011490e-02 6.69146538e-01 2.82672554e-01 7.25389719e-01 -3.77705067e-01 -3.64697188e-01 1.53865516e+00 -5.04073381e-01 -8.11497331e-01 -1.35553047e-01 4.87377107e-01 -1.10410190e+00 1.01603305e+00 9.51353252e-01 -9.94663298e-01 -4.48409170e-01 -1.10881162e+00 2.51905829e-01 -3.30833048e-02 1.33846216e-02 1.17854096e-01 7.17010617e-01 -1.09666467e+00 2.74973541e-01 -6.50830209e-01 -3.47778238e-02 -2.71027852e-02 5.46021521e-01 -2.11836249e-01 2.03411818e-01 -1.27461481e+00 1.28017902e-01 3.69431049e-01 -1.17560826e-01 -1.19764721e+00 -5.59002101e-01 -3.84831339e-01 3.04745346e-01 7.55736530e-01 -8.18803832e-02 1.84468937e+00 -5.35717487e-01 -1.37398231e+00 5.52536368e-01 -6.70742095e-01 -7.08685935e-01 4.89840418e-01 -5.16617477e-01 -7.56348312e-01 3.63560855e-01 -4.19174097e-02 2.09497645e-01 1.15745878e+00 -8.76956165e-01 -1.16166782e+00 -1.10589474e-01 2.56266668e-02 -1.05779685e-01 -4.57679629e-01 2.54581183e-01 -7.10736215e-01 -6.52014375e-01 1.39312893e-01 -7.18826234e-01 1.43379271e-01 -8.40533301e-02 -6.92965865e-01 -3.03752869e-01 9.64908779e-01 -5.11144817e-01 1.66095424e+00 -2.47921848e+00 -4.64206666e-01 4.95285481e-01 2.85614431e-01 3.93316925e-01 8.75770599e-02 4.55819786e-01 1.22119248e-01 2.21883625e-01 1.51837125e-01 -2.96653897e-01 -2.31642593e-02 1.91314861e-01 -6.62741601e-01 3.86056304e-01 2.33737633e-01 2.26604804e-01 -7.70180047e-01 -3.40295821e-01 -6.49614260e-02 1.71586484e-01 -4.69565898e-01 4.71383572e-01 -1.56405136e-01 -1.70774519e-01 -6.57013506e-02 3.18174571e-01 5.02791703e-01 -1.92698464e-01 1.66092768e-01 -2.57055610e-02 -1.90225527e-01 7.85735071e-01 -1.69159651e+00 1.06425822e+00 -3.26251537e-01 7.92861521e-01 3.18911970e-01 -8.40964675e-01 9.04117405e-01 6.28427029e-01 1.14856802e-01 -5.30120611e-01 2.92569343e-02 3.37187082e-01 -1.10430427e-01 -1.63562551e-01 7.25231230e-01 -4.84250374e-02 -9.44886431e-02 6.42348886e-01 6.07606843e-02 1.91403449e-01 1.61027133e-01 3.91848505e-01 1.39309669e+00 -6.65202379e-01 8.57858136e-02 -3.02259736e-02 4.78168875e-01 -3.64566952e-01 3.19701016e-01 1.23638594e+00 -4.26281318e-02 5.22242904e-01 7.53208935e-01 2.80238967e-03 -6.68343067e-01 -9.92752492e-01 -3.12385596e-02 1.11485851e+00 -2.18395591e-02 -7.40344107e-01 -6.91055059e-01 -4.44562554e-01 3.72650027e-02 5.58393598e-01 -1.86592907e-01 -3.24773163e-01 -3.07516545e-01 -1.50253728e-01 1.18095517e+00 3.30601752e-01 2.98520833e-01 -6.54404283e-01 -6.77602947e-01 4.51724231e-01 -1.64817661e-01 -1.22582185e+00 -5.23189306e-01 6.51571631e-01 -4.09566373e-01 -8.59520018e-01 -6.05299532e-01 -6.34223163e-01 4.15263362e-02 2.94219196e-01 7.80608833e-01 1.27801940e-01 -7.03272689e-03 1.92621067e-01 -3.65317196e-01 -2.34848276e-01 -6.49424732e-01 2.96311709e-03 4.18172866e-01 2.20399722e-01 2.77170628e-01 -2.56336719e-01 -3.03139776e-01 4.66929942e-01 -9.93108928e-01 -4.07842726e-01 3.68742585e-01 6.59998119e-01 4.19699103e-01 5.58280408e-01 5.11772454e-01 -6.84093893e-01 7.56846607e-01 -2.64046222e-01 -6.21410966e-01 1.15464695e-01 -4.54319596e-01 1.18122190e-01 6.22135758e-01 -8.14042330e-01 -4.25170839e-01 5.91281801e-02 -4.98341769e-01 -5.41737497e-01 -3.18934768e-01 2.83588111e-01 -1.79094613e-01 2.24305436e-01 3.63618135e-01 3.98433387e-01 -2.74145663e-01 -5.90877950e-01 1.66991521e-02 1.07808506e+00 7.21878469e-01 -3.58700931e-01 6.20324016e-01 2.66462117e-01 -4.46244806e-01 -1.22223270e+00 -4.09865558e-01 -6.76797748e-01 -1.65670943e-02 -1.32027984e-01 4.39636946e-01 -9.62130427e-01 -1.00526714e+00 5.90851605e-01 -9.99188662e-01 -2.54900455e-01 -3.24012302e-02 3.98800015e-01 -2.31812000e-01 4.53742623e-01 -6.01988673e-01 -1.22629106e+00 -2.31323168e-01 -1.02671492e+00 1.17208576e+00 -1.54668763e-01 -5.33662856e-01 -3.09394658e-01 -3.14112186e-01 -1.16416708e-01 3.41316968e-01 -5.53282499e-01 9.50806677e-01 -1.12525618e+00 -1.76491812e-01 -4.11823034e-01 -3.63261476e-02 5.07514536e-01 2.30267227e-01 -4.99253459e-02 -1.06963837e+00 -3.28093916e-01 1.18538827e-01 -4.95017394e-02 6.00514948e-01 9.72063467e-03 1.35620856e+00 -5.57354033e-01 -2.48516679e-01 1.19493425e-01 8.84364069e-01 4.42545205e-01 4.73681390e-01 3.68764512e-02 3.09255838e-01 2.78159410e-01 6.80459082e-01 6.41330004e-01 6.26888424e-02 9.22826886e-01 2.11559504e-01 2.91166771e-02 1.32396892e-01 -2.51412630e-01 3.61138999e-01 7.02070773e-01 7.69513905e-01 -7.38384843e-01 -8.00233781e-01 7.22607851e-01 -1.50910592e+00 -8.68833363e-01 -5.40400222e-02 2.30772305e+00 1.05080163e+00 6.73502207e-01 3.00101846e-01 9.90745962e-01 6.67751908e-01 -3.22808810e-02 -3.52772564e-01 -4.51494187e-01 -7.87822343e-03 4.41177517e-01 3.93668741e-01 7.02745140e-01 -9.79897201e-01 8.73459220e-01 6.01666927e+00 1.11570382e+00 -1.41262436e+00 -1.22645229e-01 5.94435275e-01 -1.53634578e-01 -6.61828965e-02 -8.34233463e-02 -9.99486268e-01 6.10455394e-01 1.40430057e+00 -9.34856310e-02 2.52382576e-01 6.47824049e-01 3.23982239e-01 -3.41175109e-01 -1.22471952e+00 9.52879608e-01 -1.30554006e-01 -9.90005791e-01 4.08097468e-02 5.84342554e-02 -1.49666727e-01 -3.26624423e-01 -9.78128389e-02 1.94613367e-01 -2.35662777e-02 -8.33531022e-01 1.20130062e+00 1.58538133e-01 8.59250724e-01 -8.40453029e-01 4.96545553e-01 6.02966666e-01 -1.13000512e+00 3.93334366e-02 -6.44403473e-02 -1.37202203e-01 1.74807236e-01 8.07337999e-01 -1.41867268e+00 2.41096467e-01 9.70709562e-01 -1.19001381e-01 -3.02764148e-01 9.80966747e-01 -9.80984885e-03 1.03367972e+00 -8.85948300e-01 -1.27419069e-01 1.86227020e-02 7.16501892e-01 6.65377438e-01 1.43962216e+00 3.71260405e-01 -9.99595970e-03 1.44849226e-01 4.29945976e-01 2.66636000e-03 -1.16619244e-01 -1.75507545e-01 -2.57004380e-01 8.41231108e-01 7.39477694e-01 -7.45237827e-01 -4.95336831e-01 -6.61329404e-02 1.08876085e+00 -1.77043825e-01 3.14691484e-01 -6.74132884e-01 -7.11907506e-01 7.19970345e-01 3.09900254e-01 7.48400211e-01 -2.35820770e-01 3.30052227e-01 -6.48007035e-01 2.99493000e-02 -1.10351741e+00 2.11588725e-01 -6.04851067e-01 -7.94523180e-01 6.49782419e-01 -1.42309546e-01 -9.21549201e-01 -4.75707889e-01 -4.21164155e-01 -1.78363517e-01 8.03242385e-01 -1.04353416e+00 -2.49134928e-01 8.78915638e-02 6.64300978e-01 6.16932988e-01 1.35658532e-02 8.63710225e-01 6.60374105e-01 -2.84268320e-01 9.74567771e-01 -1.62964791e-01 2.05778629e-01 5.81040800e-01 -9.84476328e-01 6.07053757e-01 1.02849948e+00 5.07622659e-01 5.80846369e-01 9.54082191e-01 -4.06755030e-01 -1.23858869e+00 -8.17280352e-01 1.29882896e+00 3.96448262e-02 7.09838867e-01 -8.24583113e-01 -9.96230662e-01 1.79920748e-01 -2.23672971e-01 -7.39625916e-02 5.33707678e-01 -6.40949309e-02 -4.32097822e-01 -1.60229996e-01 -9.65700805e-01 3.88066858e-01 5.37044108e-01 -9.15685594e-01 -4.08879191e-01 1.16667137e-01 1.02070737e+00 -4.29997504e-01 -4.65263754e-01 2.24018171e-01 5.85694313e-01 -8.55718851e-01 6.56542301e-01 -5.19743800e-01 -2.32680529e-01 -5.46043634e-01 -4.72712696e-01 -8.23729753e-01 2.72836946e-02 -1.10103154e+00 -4.25334066e-01 1.29866505e+00 4.88319427e-01 -4.36363250e-01 5.23695529e-01 2.11567834e-01 2.93076448e-02 -4.57473457e-01 -1.35822499e+00 -7.68265665e-01 -4.83059704e-01 -1.02659655e+00 5.10235190e-01 5.57450831e-01 -1.94410384e-01 2.66197294e-01 -5.52976489e-01 5.64177573e-01 1.61288112e-01 -3.74241590e-01 7.92168558e-01 -9.98846710e-01 -5.42311251e-01 -2.72295505e-01 -3.24219316e-01 -1.62484729e+00 -3.00600320e-01 -3.20713073e-01 3.46728116e-01 -8.60239685e-01 -3.93886894e-01 -3.36193711e-01 -3.99286717e-01 5.30408204e-01 5.62502295e-02 8.87966156e-02 1.50925606e-01 1.03800045e-02 -5.43310881e-01 4.46308441e-02 4.43899393e-01 -2.79542040e-02 -1.58513486e-01 2.91735262e-01 -4.65226144e-01 3.41195881e-01 6.65781498e-01 -7.44120061e-01 -1.80685923e-01 -1.99139729e-01 3.23796570e-01 2.38250658e-01 4.50988054e-01 -1.13460612e+00 2.51726091e-01 2.72091955e-01 6.99602515e-02 -6.67407870e-01 4.70914394e-01 -9.34033930e-01 -2.02332828e-02 4.02457893e-01 -5.42245865e-01 2.00167179e-01 4.85964894e-01 6.87284589e-01 -2.15253517e-01 -2.12697417e-01 4.80012357e-01 2.03213125e-01 -5.75463891e-01 -1.10965446e-01 -1.03174579e+00 1.04486205e-01 7.12324083e-01 -8.92607570e-02 -2.20469628e-02 -7.30719566e-01 -6.19717836e-01 1.97134152e-01 -8.79331976e-02 4.71747398e-01 6.89153016e-01 -9.44720328e-01 -4.02043760e-01 5.61064422e-01 -3.68822329e-02 -1.60957307e-01 3.41468416e-02 6.85488820e-01 -2.40423962e-01 6.35161579e-01 5.74368238e-01 -9.25304234e-01 -1.60625172e+00 2.48865291e-01 3.22851956e-01 6.62370771e-02 -3.96653533e-01 1.20409608e+00 -1.37665942e-01 1.77575305e-01 8.48670125e-01 -6.00241542e-01 3.04303974e-01 8.15673769e-02 8.96530092e-01 1.70381308e-01 4.82672602e-01 -4.87669408e-01 -5.96261501e-01 9.74296629e-02 -3.76311898e-01 -5.24365544e-01 8.85829508e-01 -1.02110691e-01 1.37302473e-01 7.28777230e-01 1.23950255e+00 3.12537551e-01 -9.25526738e-01 -3.90547067e-01 7.75991604e-02 -3.10108840e-01 1.32530510e-01 -7.13845730e-01 -7.15584874e-01 8.43048692e-01 7.20612288e-01 5.92257082e-01 1.19874632e+00 1.76047504e-01 9.32651401e-01 4.67633635e-01 3.47065210e-01 -9.85440731e-01 -6.03095675e-03 5.49523056e-01 4.70651537e-01 -5.94078064e-01 -2.93265164e-01 -1.14131331e-01 -2.34299526e-01 1.03352416e+00 1.87999561e-01 7.30512142e-02 7.85086393e-01 6.48715854e-01 -1.96318477e-02 -4.53483947e-02 -9.85917509e-01 -4.45301682e-02 8.86474699e-02 1.66883066e-01 1.69881001e-01 9.91677307e-03 5.93191870e-02 8.14524829e-01 -4.40648526e-01 -1.46612629e-01 4.34341788e-01 9.08767939e-01 -7.20331073e-01 -1.04213643e+00 -4.90868419e-01 5.47990859e-01 -8.88713896e-01 -1.79013491e-01 -4.46976632e-01 5.34416974e-01 -1.29380614e-01 1.39534712e+00 3.00924748e-01 -7.77194917e-01 4.41008031e-01 2.53221542e-01 -2.73402408e-02 -6.26654148e-01 -4.73572940e-01 4.40809250e-01 1.97812065e-01 -4.68634605e-01 -2.70113386e-02 -4.44828004e-01 -1.23773646e+00 -1.81311086e-01 -5.89578092e-01 4.70269561e-01 4.37876523e-01 1.05967724e+00 3.19458514e-01 5.26548564e-01 6.43758893e-01 -3.67209941e-01 -5.97308636e-01 -8.26584339e-01 -5.75009882e-01 -1.60545185e-01 8.77190113e-01 -4.62956488e-01 -7.50583231e-01 -6.80401325e-02]
[14.433695793151855, 6.075394153594971]
fbc5c14c-92ab-4b34-9559-fdd59d78fd97
deconvolutional-time-series-regression-a
null
null
https://aclanthology.org/D18-1288
https://aclanthology.org/D18-1288.pdf
Deconvolutional Time Series Regression: A Technique for Modeling Temporally Diffuse Effects
Researchers in computational psycholinguistics frequently use linear models to study time series data generated by human subjects. However, time series may violate the assumptions of these models through temporal diffusion, where stimulus presentation has a lingering influence on the response as the rest of the experiment unfolds. This paper proposes a new statistical model that borrows from digital signal processing by recasting the predictors and response as convolutionally-related signals, using recent advances in machine learning to fit latent impulse response functions (IRFs) of arbitrary shape. A synthetic experiment shows successful recovery of true latent IRFs, and psycholinguistic experiments reveal plausible, replicable, and fine-grained estimates of latent temporal dynamics, with comparable or improved prediction quality to widely-used alternatives.
['William Schuler', 'Cory Shain']
2018-10-01
null
null
null
emnlp-2018-10
['time-series-regression']
['time-series']
[ 2.01456577e-01 -4.18975800e-01 7.98387453e-02 -4.45279360e-01 -2.81559706e-01 -7.53923178e-01 8.68038774e-01 -1.47951429e-03 -4.77992922e-01 4.76800263e-01 5.55459499e-01 -5.88538826e-01 -5.24472445e-02 -4.34424907e-01 -6.77746415e-01 -4.10287887e-01 -5.73596537e-01 2.65479535e-02 -5.45694083e-02 -1.84790164e-01 3.74186754e-01 4.04693007e-01 -1.18936121e+00 5.48358440e-01 4.92833495e-01 6.82600319e-01 -6.83151782e-02 8.02740812e-01 2.54409492e-01 9.64357257e-01 -5.52672207e-01 -6.34799289e-05 4.30276021e-02 -3.99261981e-01 -5.78660607e-01 -3.08082640e-01 2.12409440e-02 -3.03317457e-01 -6.98480129e-01 5.46831846e-01 8.90906230e-02 3.35244983e-01 8.88363361e-01 -1.03253412e+00 -1.24250162e+00 8.76342595e-01 -3.48582923e-01 5.16517162e-01 4.47023600e-01 4.15712148e-01 9.37516272e-01 -5.14183462e-01 3.97683650e-01 1.41587853e+00 8.12968194e-01 2.31463835e-01 -1.87961984e+00 -6.32822454e-01 1.99574426e-01 2.15307668e-01 -9.75343823e-01 -5.19152284e-01 8.28898668e-01 -7.33133972e-01 1.12227333e+00 -3.99684124e-02 7.45774925e-01 1.79130614e+00 8.82483959e-01 5.34314990e-01 1.52201259e+00 -4.15873766e-01 1.56429008e-01 -1.97455585e-01 3.24682981e-01 1.82994053e-01 -3.08326125e-01 7.76879132e-01 -8.10210347e-01 -1.84904128e-01 9.97828245e-01 -2.60055989e-01 3.55608277e-02 2.38117471e-01 -1.33593011e+00 7.01034546e-01 1.53982639e-01 5.08452415e-01 -8.81503761e-01 2.24798709e-01 4.34495866e-01 7.59732962e-01 9.08407509e-01 2.38612533e-01 -5.50860047e-01 -3.17975253e-01 -1.09765148e+00 3.74276161e-01 5.41972399e-01 4.88583714e-01 2.23968312e-01 5.16918957e-01 -1.54087350e-01 6.64465070e-01 6.26651687e-04 2.71148115e-01 8.03879082e-01 -8.28815937e-01 -3.96399982e-02 -6.91106692e-02 3.23173881e-01 -1.06818569e+00 -6.23290598e-01 -2.29941070e-01 -5.82865715e-01 -1.61358550e-01 8.91526163e-01 -7.53613859e-02 -8.98249984e-01 1.96764958e+00 -2.90644944e-01 2.75179923e-01 -1.60993487e-01 6.96181536e-01 1.16204739e-01 9.90520775e-01 3.55820894e-01 -6.88450754e-01 1.07334495e+00 -1.63377881e-01 -8.88587952e-01 -2.95134753e-01 7.37818703e-02 -6.40151381e-01 1.20062518e+00 4.62057918e-01 -1.15494096e+00 -8.05720687e-01 -8.35290194e-01 -4.46568057e-03 -1.09410562e-01 -2.23164484e-01 1.10193634e+00 3.38080406e-01 -1.12850773e+00 8.03859472e-01 -1.00374293e+00 -8.54828656e-02 4.75100838e-02 1.86380163e-01 -2.70229697e-01 3.58160675e-01 -1.31550658e+00 9.26307857e-01 1.74154878e-01 1.30714193e-01 -1.11393642e+00 -8.02623153e-01 -5.03375828e-01 -8.48977789e-02 -4.65410203e-02 -1.80710793e-01 1.52583408e+00 -1.14187014e+00 -1.68782282e+00 6.86086476e-01 -3.94397229e-01 -5.63478887e-01 3.04860830e-01 -2.30573833e-01 -7.49481916e-01 -2.05287918e-01 -2.76874118e-02 2.17111290e-01 1.17519355e+00 -6.46222591e-01 5.09230979e-02 -1.63681865e-01 -3.53953570e-01 -4.04306650e-01 -1.57371629e-02 1.71754837e-01 2.25598916e-01 -8.81321967e-01 3.51645321e-01 -7.51583517e-01 -8.33626166e-02 -4.13689673e-01 -1.63081661e-02 -3.29177290e-01 8.43231678e-02 -8.52434278e-01 1.33647811e+00 -2.27432656e+00 2.35028006e-03 3.78250703e-02 2.68466145e-01 -2.82378197e-01 -4.33631390e-01 7.43514061e-01 -6.23009861e-01 -2.91570053e-02 -1.06111523e-02 -6.19511604e-02 1.01922173e-03 -7.57586956e-02 -7.39284217e-01 7.52757907e-01 1.15698159e-01 1.24741471e+00 -9.45541382e-01 1.01843834e-01 -1.03766024e-01 2.20982596e-01 -4.36951071e-01 1.38039291e-01 -1.21786818e-01 4.66331273e-01 2.58291718e-02 2.23772153e-01 3.21671039e-01 -1.13630712e-01 1.41091630e-01 -4.73291762e-02 -4.82537836e-01 5.76660454e-01 -4.75468695e-01 1.48081768e+00 -2.70565569e-01 1.12991571e+00 -3.50435823e-01 -1.19599354e+00 8.29792917e-01 4.78939563e-01 5.87943614e-01 -1.14211214e+00 1.86897323e-01 -9.59765911e-03 3.85049045e-01 -5.47623158e-01 4.56369698e-01 -5.74503958e-01 -1.77175507e-01 8.99552524e-01 1.90367922e-01 1.41604230e-01 -1.07023165e-01 1.38136465e-03 1.17734039e+00 3.54233354e-01 2.69495517e-01 -1.56472474e-01 -2.38307729e-01 -9.89960879e-02 3.64316404e-01 9.31510031e-01 -1.98940933e-01 4.81575262e-03 6.31344497e-01 -6.22325718e-01 -1.47271121e+00 -1.30922139e+00 -8.39627981e-02 1.34695339e+00 -6.12266362e-01 7.08865002e-03 -4.97049332e-01 4.66848388e-02 -6.45193532e-02 1.14227927e+00 -1.00119162e+00 -4.13844377e-01 -6.05288982e-01 -7.34641612e-01 5.89529514e-01 6.47230327e-01 -3.49115133e-01 -1.49819338e+00 -6.01531267e-01 7.13711619e-01 -4.52235281e-01 -9.84354913e-01 -4.66002703e-01 2.92877644e-01 -8.81396830e-01 -4.91005450e-01 -4.56851304e-01 -3.31209004e-01 2.89247602e-01 1.41046524e-01 9.74741399e-01 -5.57973564e-01 -3.26470286e-01 2.68607140e-01 -1.40328512e-01 -4.53768373e-01 -5.90517581e-01 -5.63422680e-01 4.46110338e-01 7.16848597e-02 6.04645967e-01 -8.47404480e-01 -2.47092322e-01 2.70750299e-02 -7.40260780e-01 5.00899851e-02 2.93619901e-01 8.71504307e-01 1.01717822e-01 5.82563877e-02 7.17989683e-01 -5.84155202e-01 1.20382309e+00 -7.99162090e-01 -5.37212491e-01 -3.87132317e-02 -5.79750419e-01 1.18077695e-01 7.29859412e-01 -1.26178598e+00 -1.10156798e+00 -4.24147815e-01 3.70134383e-01 -4.10509855e-01 -2.45809987e-01 8.21321964e-01 6.01340830e-01 4.76260334e-01 1.03119135e+00 5.15010536e-01 4.29886393e-02 -2.23598629e-01 6.97309315e-01 2.98365384e-01 5.96648335e-01 -6.29279733e-01 5.53120553e-01 3.96620601e-01 -3.20632726e-01 -8.53763402e-01 -4.81909096e-01 -4.34432179e-02 -9.44334447e-01 -4.17971045e-01 4.39551741e-01 -7.58138239e-01 -9.63529289e-01 6.22531295e-01 -1.35620379e+00 -6.14594758e-01 -3.52231383e-01 8.29352975e-01 -1.00334561e+00 -9.02232155e-02 -9.69562948e-01 -1.03749168e+00 2.42203414e-01 -5.37063360e-01 5.37534595e-01 -1.93855941e-01 -7.98318088e-01 -1.14204299e+00 2.33679771e-01 -2.29720548e-01 4.29310858e-01 1.53618321e-01 1.14509284e+00 -4.44514036e-01 -1.09085910e-01 -5.36798298e-01 -2.02303962e-03 1.28335625e-01 1.70382053e-01 1.38297230e-01 -9.98367488e-01 -7.85972551e-02 3.91164631e-01 -1.03567064e-01 6.24750137e-01 1.01883614e+00 7.89607406e-01 -3.37016165e-01 -4.11484130e-02 3.95216197e-01 1.02694082e+00 5.22264123e-01 4.32264030e-01 -1.51852921e-01 1.43345401e-01 1.05460584e+00 1.18208192e-01 3.95045370e-01 2.12654069e-01 3.27645302e-01 -2.45284662e-01 2.57432461e-01 4.42857534e-01 -6.29388630e-01 7.73747504e-01 1.04035854e+00 -1.48488373e-01 -1.25464767e-01 -8.05120468e-01 6.48698628e-01 -1.88593590e+00 -1.44095063e+00 -2.72688478e-01 2.13466358e+00 8.63019764e-01 4.00032908e-01 3.01482409e-01 9.40806884e-03 3.14583331e-01 1.17495917e-01 -7.11706817e-01 -5.65496624e-01 -2.50287622e-01 1.71913743e-01 3.12442064e-01 3.81861836e-01 -6.72292531e-01 1.07271099e+00 8.54137897e+00 3.94636184e-01 -1.32137764e+00 1.30282477e-01 4.31374699e-01 -1.10224798e-01 -3.81321460e-01 5.77314198e-02 -1.03241697e-01 2.53614187e-01 1.67263126e+00 -7.17241466e-01 8.10834169e-01 2.01610461e-01 1.02987754e+00 -1.02774844e-01 -1.20806789e+00 5.35771608e-01 -3.14048767e-01 -1.08017671e+00 -2.83445895e-01 -4.41253670e-02 3.39763761e-01 -1.81934342e-01 3.77780199e-01 1.52106658e-01 5.13908207e-01 -1.19172871e+00 1.15090334e+00 1.14349890e+00 7.47546017e-01 -2.29237452e-01 5.59083521e-02 5.97842097e-01 -1.02303374e+00 -2.60058612e-01 -5.51311910e-01 -6.64124250e-01 2.57003069e-01 2.41490170e-01 -7.10219085e-01 -1.92347974e-01 4.52126443e-01 7.90351987e-01 -2.60261834e-01 4.78011996e-01 -2.13854343e-01 1.27751970e+00 -8.40243921e-02 -9.96070132e-02 1.58914685e-01 1.56431925e-03 4.60908502e-01 1.12781060e+00 2.10779339e-01 3.06905448e-01 -3.01807553e-01 1.29151642e+00 4.30702418e-01 8.93306285e-02 -9.44355786e-01 -6.16504252e-01 4.09756273e-01 7.28837788e-01 -7.67199457e-01 -1.45608306e-01 -4.42081064e-01 7.48327315e-01 3.19645852e-01 8.40084672e-01 -8.63534033e-01 2.49531463e-01 3.17372739e-01 2.41355777e-01 -3.29114087e-02 -7.21500874e-01 -3.36573422e-01 -1.14448142e+00 -1.67526364e-01 -7.29489744e-01 -5.30056953e-02 -9.69622970e-01 -1.76743340e+00 4.36982006e-01 2.92543411e-01 -1.00227702e+00 -5.83215594e-01 -4.85362560e-01 -6.32600605e-01 1.32789648e+00 -7.35430241e-01 -1.07330799e+00 4.20202613e-01 5.81708789e-01 6.05600774e-01 -2.31392328e-02 7.86079705e-01 -2.06883952e-01 -1.45591304e-01 3.10677618e-01 7.74154541e-05 -1.10076204e-01 6.19392395e-01 -1.11849189e+00 7.96199501e-01 7.88122118e-01 1.42841965e-01 1.08905828e+00 9.91534591e-01 -7.18529463e-01 -1.24104834e+00 -5.17975152e-01 9.44474876e-01 -5.63263714e-01 1.17777777e+00 -5.79366386e-01 -1.19511282e+00 8.02182019e-01 2.83730179e-01 -3.96369547e-01 6.86516643e-01 4.60506797e-01 -5.18313348e-01 2.30956033e-01 -5.09034753e-01 7.98752010e-01 8.73634934e-01 -1.14837420e+00 -7.46764719e-01 1.82850540e-01 7.94526935e-01 1.46035999e-01 -6.62545145e-01 2.18998995e-02 1.08411705e+00 -6.88240647e-01 9.17299271e-01 -1.08850765e+00 3.95853639e-01 2.83031404e-01 -3.24470811e-02 -1.52336764e+00 -9.60581601e-01 -8.31510961e-01 -1.19560532e-01 7.51396894e-01 2.68235713e-01 -5.10518193e-01 2.81591356e-01 5.66482365e-01 1.62275240e-01 -1.17390655e-01 -7.07180023e-01 -7.02355683e-01 3.32721502e-01 -1.01591921e+00 2.95107454e-01 1.05815732e+00 3.78828704e-01 3.53216797e-01 -5.71801364e-01 -6.06025532e-02 6.15781069e-01 -2.74500065e-02 3.48768950e-01 -1.26805210e+00 -2.96637744e-01 -6.32832825e-01 -2.32414491e-02 -1.05581677e+00 3.32707644e-01 -6.40222132e-01 2.02893689e-01 -8.63537610e-01 -1.46335050e-01 -1.12717278e-01 -5.64146578e-01 3.83929253e-01 2.15568468e-02 -1.32631183e-01 2.45273113e-04 2.99213827e-01 1.52802065e-01 2.47704506e-01 1.05159664e+00 7.68251419e-02 -3.69811326e-01 1.16518170e-01 -5.54993451e-01 6.77368402e-01 6.40593290e-01 -4.61286545e-01 -5.01582623e-01 -2.35989168e-01 3.62617075e-01 8.59144390e-01 5.76980531e-01 -5.56389391e-01 3.11460823e-01 -4.69288170e-01 4.92806375e-01 -4.16602969e-01 3.09324831e-01 -4.60363686e-01 2.93111414e-01 2.96205223e-01 -8.40468645e-01 5.43577194e-01 5.88551581e-01 6.38840497e-01 -1.01895981e-01 1.88448653e-01 6.69886112e-01 -1.08315244e-01 -5.22948563e-01 2.24289864e-01 -1.15974224e+00 -2.26575479e-01 6.42733335e-01 -2.64560759e-01 -1.09633356e-01 -5.90239048e-01 -1.13447821e+00 -3.99455547e-01 -2.04799417e-02 7.27774143e-01 6.98728144e-01 -1.09217739e+00 -9.14241970e-01 3.86256874e-01 -1.67652756e-01 -1.10248446e+00 4.04889464e-01 9.13231909e-01 -1.47581836e-02 7.22608089e-01 -4.62756544e-01 -4.10482943e-01 -6.70866489e-01 7.14989305e-01 7.63187483e-02 4.03496511e-02 -5.35895824e-01 7.80089557e-01 2.39917651e-01 -4.00183946e-02 -1.18294351e-01 -5.04574299e-01 -1.52281806e-01 3.88089299e-01 4.74190056e-01 1.27587512e-01 -3.92080307e-01 -5.87028623e-01 1.03206793e-02 8.88955444e-02 -1.32014468e-01 -6.57635331e-01 1.49277687e+00 -1.67530209e-01 -1.27417162e-01 1.53199506e+00 9.25036609e-01 -1.69939667e-01 -1.44028318e+00 -5.69051385e-01 1.08126953e-01 -1.55012593e-01 1.12939803e-02 -8.32197666e-01 -1.68353274e-01 1.02660418e+00 3.39333147e-01 5.99495709e-01 1.07865322e+00 -1.99379116e-01 3.02426159e-01 -7.04757795e-02 4.12372530e-01 -9.72860336e-01 2.48975560e-01 5.86381733e-01 9.73684311e-01 -6.75811291e-01 -2.45127022e-01 2.80304581e-01 -5.19470394e-01 1.24480665e+00 8.14733729e-02 -3.06100994e-01 8.78278852e-01 2.45720088e-01 -6.32371381e-02 -1.53769970e-01 -1.49913442e+00 2.48380363e-01 3.67681891e-01 4.86197919e-01 8.52103472e-01 4.34219062e-01 -4.49843854e-01 7.52210796e-01 -4.02488500e-01 1.54268891e-01 4.87187743e-01 6.01008177e-01 -2.15961888e-01 -6.96056306e-01 -2.71299273e-01 4.71049666e-01 -4.18176264e-01 -3.97684753e-01 9.96792805e-04 5.69180906e-01 -2.03617766e-01 9.52031672e-01 3.21469218e-01 -3.76927227e-01 2.86148489e-01 3.94818813e-01 6.10749424e-01 -4.34600025e-01 -4.05759275e-01 4.72847044e-01 -2.25070521e-01 -6.49549365e-01 -3.59399796e-01 -1.05806220e+00 -9.10119534e-01 -4.31544751e-01 -1.45209670e-01 -3.02655194e-02 5.65123737e-01 1.04824078e+00 -7.36094490e-02 7.51555860e-01 6.01519346e-01 -6.83469594e-01 -5.13347805e-01 -1.33414769e+00 -6.03649974e-01 3.38055521e-01 5.25016189e-01 -4.64393616e-01 -3.75070333e-01 5.20369828e-01]
[7.066016674041748, 3.2098586559295654]
09d3a796-b357-4ae9-a683-33476322ca1a
an-effortless-way-to-create-large-scale
null
null
https://aclanthology.org/L14-1283
https://aclanthology.org/L14-1283.pdf
An Effortless Way To Create Large-Scale Datasets For Famous Speakers
The creation of large-scale multimedia datasets has become a scientific matter in itself. Indeed, the fully-manual annotation of hundreds or thousands of hours of video and/or audio turns out to be practically infeasible. In this paper, we propose an extremly handy approach to automatically construct a database of famous speakers from TV broadcast news material. We then run a user experiment with a correctly designed tool that demonstrates that very reliable results can be obtained with this method. In particular, a thorough error analysis demonstrates the value of the approach and provides hints for the improvement of the quality of the dataset.
["F{\\'e}licien Vallet", 'Fran{\\c{c}}ois Salmon']
2014-05-01
null
null
null
lrec-2014-5
['person-identification']
['computer-vision']
[ 3.56037728e-02 2.67863482e-01 3.26568604e-01 -4.22382414e-01 -9.96885240e-01 -6.74800634e-01 7.11136520e-01 2.68807143e-01 -4.55574304e-01 7.26645172e-01 -9.16267708e-02 -9.15736333e-02 -9.31139886e-02 -6.15715325e-01 -6.20118082e-01 -5.07603645e-01 -1.59746394e-01 6.52666986e-01 6.32741153e-01 -2.67793477e-01 4.14351344e-01 3.70722651e-01 -2.01328301e+00 1.36823788e-01 2.84894615e-01 1.13838089e+00 1.51930735e-01 6.58899367e-01 -1.31292403e-01 6.64987385e-01 -8.73990357e-01 -5.59608579e-01 5.73565811e-02 -3.78939450e-01 -9.96623099e-01 2.57751167e-01 5.73191822e-01 -1.72885284e-01 1.91726461e-01 9.09509122e-01 5.09400666e-01 -8.09768513e-02 4.36117128e-02 -9.87018943e-01 6.05560169e-02 5.51118910e-01 -5.26495837e-02 1.86984539e-01 8.70730221e-01 -1.99751586e-01 9.15194750e-01 -6.31244659e-01 9.36266601e-01 7.75572836e-01 6.00314558e-01 -1.48614198e-02 -9.44611013e-01 -3.62171263e-01 -2.80826062e-01 1.55860931e-01 -1.71531236e+00 -7.33815074e-01 7.37995505e-01 -4.39100921e-01 6.10287786e-01 5.93068123e-01 8.05800080e-01 1.02324831e+00 -3.63425702e-01 5.05595386e-01 9.98429596e-01 -8.03341866e-01 2.56590247e-01 6.14746690e-01 1.41118824e-01 6.82186306e-01 1.74304426e-01 -2.65387118e-01 -6.32121384e-01 -1.31932646e-01 7.10288763e-01 -5.55559933e-01 -8.57159048e-02 -3.12232286e-01 -1.22555161e+00 5.19324720e-01 -1.30553290e-01 8.34538043e-01 -1.26969427e-01 -2.34566145e-02 6.15491450e-01 4.32923377e-01 5.01563787e-01 4.91491318e-01 -1.78836644e-01 -4.95415598e-01 -1.18591261e+00 3.80320758e-01 1.07144856e+00 9.37302887e-01 5.82779169e-01 -2.53316283e-01 3.69185001e-01 6.55059040e-01 -9.64182243e-02 1.29485011e-01 3.46764237e-01 -1.02724528e+00 2.06617653e-01 4.88387495e-01 4.39861745e-01 -1.47720599e+00 -4.74792331e-01 -1.75716385e-01 -4.69021320e-01 -9.91128162e-02 8.23262453e-01 -8.06852877e-02 -3.24723810e-01 1.28883290e+00 4.46516097e-01 -7.44466633e-02 -2.78599352e-01 8.39060843e-01 7.23124385e-01 4.12205875e-01 -2.36836687e-01 -3.92257839e-01 1.27681935e+00 -5.97021461e-01 -9.64087665e-01 1.96722671e-01 3.68272275e-01 -8.80112827e-01 1.15247607e+00 8.83166015e-01 -1.05920362e+00 -6.72243774e-01 -7.03032315e-01 1.23169385e-01 -3.91768038e-01 3.98308896e-02 6.29520476e-01 8.15511465e-01 -8.56293499e-01 5.55807471e-01 -6.59233332e-01 -5.26447535e-01 1.30581930e-01 2.58193880e-01 -5.33840179e-01 5.37612215e-02 -9.78406906e-01 6.08008206e-01 4.75775987e-01 6.77131861e-02 -5.97863257e-01 -2.08089158e-01 -4.97606188e-01 4.62711900e-02 6.46423399e-01 -6.57669529e-02 1.27483928e+00 -1.10969460e+00 -1.44038951e+00 1.16663504e+00 1.44803673e-01 -4.27155852e-01 7.26903081e-01 -2.54431903e-01 -4.30809140e-01 5.51686347e-01 -1.18799210e-01 2.39984468e-01 7.36858726e-01 -1.09311461e+00 -7.38878191e-01 -2.91314065e-01 4.38046396e-01 -1.24656253e-01 -6.03155732e-01 3.66594821e-01 -6.63526475e-01 -8.30910027e-01 -2.74063908e-02 -7.50677109e-01 -6.05334528e-02 -2.45130196e-01 -3.54259282e-01 -1.45469412e-01 3.60859811e-01 -7.77469099e-01 1.46621406e+00 -2.41003108e+00 7.26583973e-02 4.26423639e-01 1.15860671e-01 -6.44856989e-02 2.69989580e-01 4.55326587e-01 1.76140979e-01 2.37958938e-01 1.02904104e-01 -8.85925740e-02 4.38124910e-02 1.07557140e-01 -1.52943581e-01 3.82446796e-01 -2.84516275e-01 4.36235368e-01 -8.68392825e-01 -6.85717225e-01 -1.01808317e-01 3.55712265e-01 -4.22061861e-01 2.35241637e-01 -1.96207970e-01 4.98941362e-01 -4.68729794e-01 6.89370275e-01 1.03773430e-01 -1.37852475e-01 2.67087311e-01 -3.65772806e-02 -3.18397880e-01 1.47186145e-01 -1.29514432e+00 1.79383743e+00 -2.00520322e-01 9.23440695e-01 7.89534971e-02 -8.07664096e-01 7.72643447e-01 6.51398957e-01 5.33297002e-01 -6.00362360e-01 2.59080738e-01 2.99888283e-01 -3.47755700e-01 -8.35029066e-01 7.38114476e-01 -1.77652076e-01 -2.11860567e-01 3.55968982e-01 1.68239787e-01 -2.71789998e-01 5.18963516e-01 -2.71857306e-02 9.90009844e-01 -1.39001729e-02 3.74633759e-01 -2.24903956e-01 6.29538834e-01 1.85538486e-01 1.39889881e-01 5.67019284e-01 -1.69862926e-01 5.90506732e-01 5.71646750e-01 -5.42771220e-01 -1.15303147e+00 -5.07617295e-01 -1.08828321e-01 1.00314081e+00 -1.26035854e-01 -5.94286740e-01 -1.07266152e+00 -4.44916278e-01 -4.70526308e-01 3.50805730e-01 -4.01612282e-01 3.54156464e-01 -4.79035765e-01 -5.05180180e-01 6.76755488e-01 5.25970496e-02 2.56168813e-01 -1.00547516e+00 -7.33004868e-01 1.39846861e-01 -3.55104774e-01 -1.31286299e+00 1.01254098e-01 -1.12044729e-01 -6.61235154e-01 -9.62688565e-01 -5.91594636e-01 -6.91557646e-01 4.59066004e-01 5.51797375e-02 1.18297410e+00 3.65293443e-01 -3.59109104e-01 2.83717602e-01 -8.85417700e-01 -2.34075844e-01 -5.31936228e-01 2.26288646e-01 3.21855955e-02 5.61004654e-02 2.67990798e-01 -7.14896917e-01 -2.55615991e-02 6.09511197e-01 -9.47056413e-01 -2.83956528e-01 3.52058232e-01 4.28123593e-01 4.55427587e-01 4.44431782e-01 6.69986188e-01 -1.08648467e+00 5.37087381e-01 -2.13919729e-01 -8.80454183e-01 1.46149427e-01 -3.10453624e-01 -3.08526039e-01 5.22282302e-01 -2.64761835e-01 -7.69825757e-01 3.79718900e-01 -4.02651131e-01 5.32952556e-03 -4.82535690e-01 6.36854529e-01 -2.20575243e-01 -1.45844966e-01 7.12491989e-01 -5.42090982e-02 -3.04652780e-01 -7.27601051e-01 3.03043455e-01 7.78114974e-01 6.11501396e-01 -4.17552412e-01 6.51456177e-01 4.18810964e-01 -3.79954427e-01 -1.19725192e+00 -6.92849994e-01 -5.14212549e-01 -6.30866766e-01 -5.90498209e-01 6.37847602e-01 -6.13199353e-01 -6.39307618e-01 2.87249863e-01 -1.00349677e+00 -1.19953230e-01 -3.21871817e-01 3.49453539e-01 -5.40639222e-01 4.45535243e-01 -3.10881615e-01 -8.03130269e-01 2.27804244e-01 -8.78807485e-01 8.14990282e-01 -1.35302514e-01 -4.18587685e-01 -7.42690265e-01 1.22886546e-01 6.52139783e-01 1.81883678e-01 2.24834949e-01 2.99617589e-01 -7.86360562e-01 -5.54540336e-01 -6.40064299e-01 5.96685447e-02 2.14621842e-01 -9.27234739e-02 3.07590663e-01 -1.34577513e+00 -9.66624636e-03 4.81543951e-02 -3.46375525e-01 2.99281865e-01 -2.69981064e-02 1.23988450e+00 -1.71641633e-01 -5.17652594e-02 1.19816460e-01 1.27464747e+00 5.89926392e-02 7.32854724e-01 4.24569935e-01 2.61202037e-01 8.66158485e-01 8.64524722e-01 6.74606860e-01 2.47146672e-04 9.59913254e-01 1.71392217e-01 1.52538732e-01 2.50499338e-01 -1.26766087e-02 8.82977247e-02 8.88962567e-01 -6.10374331e-01 -2.98341423e-01 -8.72371495e-01 6.06455386e-01 -1.52212369e+00 -1.15381885e+00 -2.31080785e-01 2.24231005e+00 7.00331092e-01 4.16656852e-01 3.83617610e-01 7.09144235e-01 6.17549002e-01 -1.35965183e-01 3.05201322e-01 -1.99066341e-01 -4.23062855e-04 -9.73386317e-02 1.67941779e-01 3.21757883e-01 -1.09763551e+00 6.82732284e-01 7.60076475e+00 8.31747890e-01 -9.01503503e-01 1.21544406e-01 3.15433681e-01 -6.15435690e-02 -2.14522351e-02 -2.23607555e-01 -5.53393424e-01 3.98791105e-01 1.27861857e+00 -1.23902731e-01 3.70591581e-01 9.07905340e-01 2.41248578e-01 -4.55691427e-01 -1.02292335e+00 1.03405678e+00 1.23837106e-01 -1.32703888e+00 -2.32061148e-01 7.59138390e-02 2.51291841e-01 -3.12220961e-01 -2.81818718e-01 3.21350843e-02 -3.22955608e-01 -7.66315818e-01 1.05478585e+00 4.99402434e-01 6.00289404e-01 -7.95360625e-01 8.24120641e-01 4.61361438e-01 -8.68594110e-01 1.34149883e-02 -1.72151744e-01 -1.84703857e-01 3.73869002e-01 7.75283575e-01 -8.28734815e-01 5.93916953e-01 8.91747534e-01 2.94128597e-01 -8.32117498e-01 1.24672294e+00 -1.62742272e-01 6.83746219e-01 -4.52435493e-01 -4.68018577e-02 -4.70838472e-02 3.88591625e-02 5.23675025e-01 1.38441014e+00 2.65563339e-01 -9.52830389e-02 4.39307205e-02 2.93248624e-01 -4.89331000e-02 4.67486709e-01 -7.25954294e-01 -2.43613198e-01 2.27922589e-01 1.46845627e+00 -1.07401001e+00 -4.35038894e-01 -4.22567159e-01 8.76549840e-01 2.65657395e-01 -6.54792562e-02 -8.42466056e-01 -3.55816156e-01 -3.22345309e-02 3.54352385e-01 3.82010072e-01 -2.02658549e-01 -3.07799689e-02 -1.08289695e+00 3.45995337e-01 -1.01275671e+00 3.60121399e-01 -7.15813518e-01 -9.37232435e-01 9.54934776e-01 -2.99596582e-02 -1.21290970e+00 -3.84266019e-01 -3.98192018e-01 -8.33513215e-02 3.39086115e-01 -1.00611544e+00 -7.31666923e-01 -3.36124927e-01 3.60291660e-01 3.53167027e-01 -1.26190767e-01 9.81755376e-01 9.23290968e-01 -4.05286521e-01 3.37428540e-01 -8.48708823e-02 2.18655709e-02 6.54702723e-01 -1.00516486e+00 2.22973488e-02 6.26954496e-01 3.49713206e-01 3.84686470e-01 1.25304210e+00 -2.79190898e-01 -1.24999690e+00 -5.73984027e-01 1.08653486e+00 -5.93382895e-01 9.46451962e-01 -5.20656347e-01 -8.17492187e-01 5.19216835e-01 2.52603829e-01 -2.81876415e-01 7.30961621e-01 1.43992960e-01 -1.07672609e-01 -1.28565103e-01 -1.02810884e+00 1.20542049e-01 7.95681238e-01 -5.27739942e-01 -7.37074852e-01 5.91689467e-01 5.11299431e-01 -3.05389225e-01 -9.58347857e-01 -4.73954901e-02 6.48048639e-01 -1.31398749e+00 7.98479617e-01 -2.61535734e-01 2.52425075e-01 -1.49622351e-01 -2.37343043e-01 -7.42304742e-01 2.93580353e-01 -8.66505921e-01 1.59178391e-01 1.57555842e+00 3.27570409e-01 -1.43406481e-01 6.77863717e-01 5.74890673e-01 6.66091219e-02 -2.36503094e-01 -9.79166985e-01 -7.87975669e-01 -5.40284574e-01 -7.25764096e-01 3.32701802e-01 9.47026551e-01 4.13856804e-01 2.48050019e-01 -4.42418665e-01 3.19770649e-02 3.31589758e-01 8.80369470e-02 9.88038480e-01 -1.50562465e+00 -3.46027285e-01 -1.65674016e-01 -7.59728014e-01 -6.41711831e-01 -5.67648485e-02 -3.04759651e-01 1.03666864e-01 -1.08913040e+00 -4.82937731e-02 -3.83377612e-01 -8.06508064e-02 1.85672432e-01 2.50322968e-01 6.28738225e-01 -7.33148083e-02 2.15186402e-01 -8.78692508e-01 -7.64893666e-02 8.91589999e-01 1.52189508e-01 -1.29177958e-01 1.50245294e-01 -4.76128578e-01 9.47471380e-01 7.87036061e-01 -5.83036959e-01 -1.74351051e-01 -1.58503905e-01 6.96587563e-01 4.08231169e-02 3.55279237e-01 -1.07742202e+00 -3.11615709e-02 9.80924442e-02 -6.82520345e-02 -5.29752731e-01 3.22872490e-01 -1.05563509e+00 3.69612426e-01 1.73904821e-01 -2.98795074e-01 -8.18273202e-02 9.88544300e-02 4.81960356e-01 -6.16651893e-01 -5.41272581e-01 5.50687015e-01 -2.70216048e-01 -6.98091388e-01 -3.02198499e-01 -4.38259423e-01 -6.51396066e-02 9.24716413e-01 -5.62503375e-02 1.69428915e-01 -4.41712499e-01 -9.27704215e-01 -1.52900189e-01 6.30092621e-01 2.09861219e-01 1.90033630e-01 -1.03492725e+00 -3.58639061e-01 -6.91320514e-03 1.53157294e-01 -2.30436608e-01 1.50552019e-01 8.34395528e-01 -8.70026886e-01 2.42767751e-01 -2.66957402e-01 -4.32555795e-01 -1.51816785e+00 7.69638479e-01 -3.26670008e-03 -8.56117234e-02 -5.93885660e-01 5.37146807e-01 -5.13225913e-01 -4.98702563e-02 4.57871050e-01 -3.44885767e-01 -5.17042339e-01 5.02024651e-01 7.88377106e-01 4.35919851e-01 3.28317821e-01 -7.53772020e-01 -2.25749493e-01 2.72488624e-01 2.77252942e-01 -4.84272212e-01 1.65854466e+00 -4.05957133e-01 -1.56618878e-01 7.58926034e-01 8.82870972e-01 4.06414986e-01 -7.76000321e-01 1.59307495e-01 2.98548877e-01 -6.25422597e-01 -4.86576930e-02 -5.92320502e-01 -7.89448857e-01 6.94283903e-01 3.65949690e-01 9.76513922e-01 1.19122267e+00 7.53438622e-02 5.94489157e-01 6.36617303e-01 6.06571853e-01 -1.38324893e+00 -1.46709859e-01 8.78000259e-02 8.34479690e-01 -1.30232573e+00 1.09978721e-01 -5.31434894e-01 -3.37943524e-01 1.28884459e+00 -6.80487929e-03 1.06780402e-01 5.98327100e-01 2.07389832e-01 1.02211073e-01 -3.87020230e-01 -4.62114066e-01 -6.93764240e-02 1.96090385e-01 3.34511071e-01 5.68913996e-01 -2.30319083e-01 -5.55264175e-01 4.71815556e-01 -4.07652855e-01 3.42961252e-01 7.19309747e-01 1.08975303e+00 -6.06476903e-01 -1.04143071e+00 -5.94482064e-01 4.67074430e-03 -7.33127952e-01 1.96111679e-01 -6.23275518e-01 1.02898788e+00 8.22741538e-02 1.04790127e+00 -2.76595116e-01 -1.53086841e-01 3.90419126e-01 1.39962703e-01 4.91024375e-01 -4.93518770e-01 -6.77146912e-01 4.50295389e-01 4.54385221e-01 -6.46507621e-01 -1.03580499e+00 -6.55320883e-01 -9.95513558e-01 -2.12491080e-01 -2.49407560e-01 4.55657482e-01 8.20916891e-01 8.56186807e-01 9.73752588e-02 2.80993581e-01 5.97168982e-01 -8.40349019e-01 -1.77199408e-01 -9.11960602e-01 -6.81163549e-01 5.08734047e-01 1.73057824e-01 -5.00657141e-01 -4.76714581e-01 5.75662315e-01]
[8.749320983886719, 0.04438069090247154]
c5cf1c02-42b0-4483-b469-a9299d61bd5a
wssl-weighted-self-supervised-learning
2211.13856
null
https://arxiv.org/abs/2211.13856v1
https://arxiv.org/pdf/2211.13856v1.pdf
WSSL: Weighted Self-supervised Learning Framework For Image-inpainting
Image inpainting is the process of regenerating lost parts of the image. Supervised algorithm-based methods have shown excellent results but have two significant drawbacks. They do not perform well when tested with unseen data. They fail to capture the global context of the image, resulting in a visually unappealing result. We propose a novel self-supervised learning framework for image-inpainting: Weighted Self-Supervised Learning (WSSL) to tackle these problems. We designed WSSL to learn features from multiple weighted pretext tasks. These features are then utilized for the downstream task, image-inpainting. To improve the performance of our framework and produce more visually appealing images, we also present a novel loss function for image inpainting. The loss function takes advantage of both reconstruction loss and perceptual loss functions to regenerate the image. Our experimentation shows WSSL outperforms previous methods, and our loss function helps produce better results.
['Natarajan Subramanyam', 'Poojasree Dwarkanath', 'Madhoolika Gangaraju', 'Rahul Kunigal Ravishankar', 'Shubham Gupta']
2022-11-25
null
null
null
null
['image-inpainting']
['computer-vision']
[ 4.72933024e-01 1.00806709e-02 -2.75117099e-01 -3.67156327e-01 -8.17866087e-01 -9.39813480e-02 3.64006490e-01 -4.38845940e-02 -3.21703970e-01 8.58405113e-01 3.62286270e-01 9.79131088e-02 1.21247500e-01 -6.85108542e-01 -7.74049044e-01 -7.01441228e-01 1.55193627e-01 -9.83542427e-02 1.79934338e-01 -2.79265200e-03 2.97222704e-01 3.97070765e-01 -1.63014257e+00 5.25586724e-01 9.49450552e-01 8.85126948e-01 6.14234924e-01 5.30127704e-01 -2.70220429e-01 1.21321428e+00 -5.74642837e-01 -5.88174835e-02 3.50328863e-01 -8.16611171e-01 -7.52824903e-01 4.76141602e-01 4.08917367e-01 -3.14073384e-01 -4.18533236e-01 9.56295848e-01 4.07004178e-01 1.24625377e-01 5.53903937e-01 -1.28860581e+00 -4.59730774e-01 2.36007527e-01 -7.38136709e-01 -1.79338194e-02 4.37991261e-01 5.32072224e-02 9.00445700e-01 -8.85551035e-01 7.75333464e-01 1.27753484e+00 6.57245398e-01 6.46425009e-01 -1.34963191e+00 -5.07062316e-01 -1.10020667e-01 2.20477298e-01 -1.04785049e+00 -4.73167628e-01 1.23332548e+00 -1.14727303e-01 4.54362690e-01 3.83578181e-01 4.67168152e-01 9.26859021e-01 3.59763175e-01 1.06439853e+00 1.41576922e+00 -8.14984143e-01 1.55742720e-01 1.34056777e-01 -4.60032493e-01 8.32325637e-01 -1.56566918e-01 2.51608491e-01 -6.75812066e-01 1.78762544e-02 9.74191189e-01 2.86202669e-01 -3.23850363e-01 -4.81181294e-01 -9.17680442e-01 7.50072896e-01 6.21769786e-01 2.08219022e-01 -3.58037680e-01 1.89004511e-01 2.38728955e-01 5.59442520e-01 6.96156621e-01 4.18538392e-01 -1.09759063e-01 1.17630258e-01 -1.26112986e+00 1.31177276e-01 3.58326852e-01 4.89275843e-01 9.93355215e-01 4.53487737e-03 -1.40225396e-01 1.14085591e+00 1.11549012e-01 1.71427116e-01 6.24151587e-01 -1.22943306e+00 3.91536951e-01 5.88535607e-01 1.38475075e-01 -1.10339630e+00 -1.31395876e-01 -1.15953408e-01 -6.63529098e-01 8.54458511e-01 1.60597384e-01 -1.09047676e-02 -1.03532350e+00 1.65483379e+00 4.02845964e-02 1.04245268e-01 -1.15567232e-02 7.51095235e-01 7.88212240e-01 1.04514754e+00 1.25603870e-01 -2.74784237e-01 7.63241231e-01 -1.42761779e+00 -8.65071535e-01 -4.19617891e-01 1.86513171e-01 -9.83638644e-01 1.28284669e+00 5.00790834e-01 -1.33035290e+00 -7.86099017e-01 -1.12360024e+00 -1.84226364e-01 -7.35403672e-02 5.54203428e-02 4.00253057e-01 3.44548434e-01 -1.03241706e+00 7.59880126e-01 -5.28837502e-01 -1.45454273e-01 5.68850815e-01 1.82956249e-01 -4.34512436e-01 -3.49733382e-01 -6.91312909e-01 9.77386713e-01 3.57654929e-01 -3.46695274e-01 -7.66497791e-01 -4.52037454e-01 -1.03035951e+00 -3.63851041e-02 2.73428679e-01 -5.99315941e-01 1.08956277e+00 -1.39944363e+00 -1.31400311e+00 8.72376442e-01 -1.36422873e-01 -4.50977325e-01 7.44278133e-01 -1.97329298e-01 -3.57569963e-01 4.40063268e-01 2.35723659e-01 1.01680005e+00 1.40003967e+00 -1.80643702e+00 -6.15924776e-01 -9.98417065e-02 -3.68814737e-01 2.18413144e-01 -4.04430509e-01 -3.19629997e-01 -4.84313935e-01 -1.29521358e+00 3.52691747e-02 -5.18958151e-01 -3.25796574e-01 4.86348927e-01 -2.23650441e-01 2.31939092e-01 1.22221088e+00 -7.97827005e-01 1.10433352e+00 -2.35120273e+00 6.05605133e-02 -1.64796084e-01 2.37624392e-01 5.36819756e-01 -4.21563327e-01 6.72164023e-01 -1.57230377e-01 -2.41317481e-01 -5.40570259e-01 -7.85604835e-01 -5.06408334e-01 5.11726618e-01 -4.40058082e-01 1.11247517e-01 3.61520618e-01 7.79644847e-01 -1.01657867e+00 -7.59231150e-01 3.78571749e-01 4.36213732e-01 -3.50419551e-01 5.19755304e-01 -2.97482014e-01 4.05175447e-01 -1.71263337e-01 5.14502585e-01 7.12249100e-01 3.86238610e-03 -2.40280971e-01 -2.27660164e-01 1.67920411e-01 -2.53182977e-01 -9.46701527e-01 1.97004187e+00 -6.30031824e-01 6.86451375e-01 2.47961301e-02 -9.18405473e-01 9.24694180e-01 -9.92983393e-03 5.03583193e-01 -8.44227493e-01 -7.62710050e-02 4.20103893e-02 -5.04191160e-01 -8.02718818e-01 3.71782571e-01 -3.28473508e-01 3.08237523e-01 5.88905811e-01 1.57631725e-01 -2.56265402e-01 1.69964693e-02 2.37644941e-01 1.09309125e+00 3.40943187e-01 1.60311475e-01 1.39562428e-01 4.27864432e-01 -7.19035789e-03 5.72466016e-01 5.15474856e-01 -7.28807077e-02 1.06034219e+00 3.21063727e-01 -5.62430918e-01 -1.17910349e+00 -1.08213794e+00 1.69544399e-01 8.05234909e-01 3.79231870e-01 -3.46866637e-01 -7.71471858e-01 -7.93640792e-01 -1.10845625e-01 6.82066679e-01 -4.92221236e-01 -4.14105833e-01 -4.69386131e-01 -3.84219587e-01 1.13761611e-01 3.78107667e-01 7.08129406e-01 -1.41370213e+00 -5.38438678e-01 2.80809343e-01 -3.46402794e-01 -1.09506929e+00 -6.35068417e-01 8.90023820e-03 -1.10902953e+00 -1.10216963e+00 -9.27304864e-01 -1.17639792e+00 1.04806101e+00 5.23589730e-01 9.81556594e-01 1.88160121e-01 -5.10451972e-01 1.92342818e-01 -4.78864133e-01 -2.04725653e-01 -7.94892788e-01 -3.55558306e-01 -2.95060903e-01 3.16104800e-01 -2.65262634e-01 -5.80429494e-01 -7.75061131e-01 2.53567517e-01 -1.34295321e+00 2.11977482e-01 7.54384100e-01 1.15391850e+00 6.08022571e-01 3.91465992e-01 5.08254230e-01 -1.05656171e+00 7.20155239e-01 -4.21196632e-02 -2.02563614e-01 1.87114999e-01 -7.29821801e-01 3.74370247e-01 8.62072289e-01 -4.78350580e-01 -1.24092579e+00 2.83970654e-01 -2.04919070e-01 -4.72098917e-01 1.26908168e-01 1.38081908e-01 8.74079950e-03 -3.39021385e-01 6.17852211e-01 2.85583615e-01 4.55290377e-01 -6.03733480e-01 2.91841298e-01 4.57975596e-01 6.95633113e-01 -2.33542144e-01 7.94269145e-01 5.83577394e-01 -6.94575235e-02 -6.59836054e-01 -8.18173647e-01 -3.76092762e-01 -1.94507003e-01 -2.62433410e-01 4.61457670e-01 -8.45074117e-01 -1.41603723e-01 3.96490395e-01 -9.96963978e-01 -3.34830582e-01 -6.86748147e-01 1.95767179e-01 -8.39837074e-01 6.83401167e-01 -5.92483580e-01 -8.06374013e-01 -3.70670587e-01 -9.92356479e-01 9.60148871e-01 8.05198774e-02 6.44586533e-02 -9.25159752e-01 6.50673360e-02 4.40358788e-01 3.30761403e-01 5.87447226e-01 9.61002469e-01 1.39419571e-01 -3.63032132e-01 -1.48622140e-01 -3.02172691e-01 9.29407120e-01 3.65162820e-01 -3.27114701e-01 -8.79138827e-01 -3.75755697e-01 1.63987175e-01 -5.92429638e-01 1.34092665e+00 1.98593408e-01 1.33301234e+00 -5.01289427e-01 -2.30680138e-01 5.94228745e-01 1.62572789e+00 6.75704330e-02 1.02282751e+00 2.85216063e-01 5.53876579e-01 6.41833961e-01 8.17672789e-01 2.71818191e-01 3.62369977e-02 4.63185430e-01 6.09527111e-01 -6.24223709e-01 -7.18554676e-01 -7.08298385e-01 4.83280629e-01 6.07591271e-01 7.70699084e-02 -1.22572027e-01 -4.61881727e-01 5.36591291e-01 -2.01757932e+00 -9.34121370e-01 1.52774408e-01 2.22986078e+00 1.01646376e+00 1.87217876e-01 1.63605541e-01 3.90955061e-01 5.70744216e-01 2.95838714e-01 -4.11994696e-01 -2.94342548e-01 -8.97869915e-02 4.00118887e-01 3.64687264e-01 5.69209754e-01 -1.00191402e+00 9.71313238e-01 6.81200123e+00 1.13387954e+00 -1.05645525e+00 1.85547233e-01 7.32202411e-01 9.83191654e-02 -2.66780883e-01 1.20097876e-01 -1.52730435e-01 5.50453424e-01 3.39343011e-01 2.05750510e-01 3.43397915e-01 6.84147537e-01 3.15329611e-01 -4.75965559e-01 -8.24377120e-01 1.20809019e+00 4.07708019e-01 -1.35831833e+00 1.85670272e-01 -2.60839015e-01 8.58141184e-01 -3.51509809e-01 1.14858769e-01 -4.37520184e-02 8.21641162e-02 -9.72413719e-01 6.19962215e-01 4.79106784e-01 7.74680555e-01 -9.00688648e-01 5.36860406e-01 3.97046715e-01 -7.04625010e-01 -2.48792052e-01 -4.15576369e-01 4.40807082e-03 1.82302698e-01 6.24600708e-01 -6.43703759e-01 3.46510559e-01 4.78363425e-01 5.79295695e-01 -6.68675661e-01 1.37808931e+00 -3.24256301e-01 3.24078202e-01 9.34012532e-02 3.56139123e-01 2.13771090e-01 -4.33675461e-02 4.79811966e-01 1.00006390e+00 2.03521132e-01 -2.42622480e-01 2.22889826e-01 8.23238671e-01 -1.12090439e-01 1.02726035e-01 -6.85971558e-01 6.23222217e-02 6.09330833e-02 1.10384607e+00 -6.97827280e-01 -8.68552178e-02 -2.67504275e-01 1.52092195e+00 2.02164099e-01 2.87562400e-01 -6.17183745e-01 -5.71668744e-01 1.65345013e-01 4.22862649e-01 2.55709797e-01 -1.03365667e-01 -1.61049679e-01 -1.16378105e+00 3.81529927e-02 -7.51832247e-01 2.44779199e-01 -1.00603807e+00 -1.26761973e+00 6.85705721e-01 -1.48855507e-01 -1.54267836e+00 -1.07133023e-01 -3.04861546e-01 -6.95094943e-01 4.74099338e-01 -1.72259891e+00 -1.36594367e+00 -4.04487044e-01 5.86099625e-01 9.45577979e-01 -2.48984039e-01 7.83493578e-01 2.14740708e-01 -2.26860777e-01 4.90213811e-01 1.36993214e-01 -1.95098028e-01 1.01347876e+00 -1.15853930e+00 1.13117419e-01 8.63283396e-01 2.36027509e-01 7.41062835e-02 7.91527867e-01 -5.09204805e-01 -1.14923143e+00 -1.21757090e+00 7.35536277e-01 1.10809736e-01 9.88259763e-02 -1.74316749e-01 -7.59611011e-01 2.72308022e-01 3.89667958e-01 2.12736428e-01 4.56692427e-01 -4.92807567e-01 -4.42842901e-01 -3.95362139e-01 -1.47917092e+00 5.86202502e-01 7.99862325e-01 -3.23818922e-01 -5.40742517e-01 3.47287685e-01 6.53853059e-01 -7.56615475e-02 -4.21824872e-01 2.20386222e-01 4.03552115e-01 -1.20759058e+00 1.10534060e+00 -1.80621043e-01 8.15812290e-01 -2.51666903e-01 1.26649171e-01 -1.41993153e+00 -1.24785401e-01 -8.53911161e-01 -1.57695919e-01 1.18823445e+00 2.27185145e-01 -3.96280706e-01 8.45407963e-01 1.37556404e-01 1.15083463e-01 -7.82401323e-01 -5.78878701e-01 -8.54439735e-01 -2.19620615e-01 -1.85995370e-01 1.81201354e-01 7.40051508e-01 -1.40588805e-01 3.03075403e-01 -7.76669800e-01 -2.92974174e-01 9.31620061e-01 2.19037428e-01 6.39877260e-01 -8.48710656e-01 -3.57698053e-01 -1.37609348e-01 -3.30098212e-01 -8.78131449e-01 3.81122902e-02 -5.55122912e-01 1.10390492e-01 -1.56900799e+00 2.99504578e-01 -3.69619548e-01 -2.08376169e-01 6.39628828e-01 -2.01814994e-01 7.40677953e-01 3.38030875e-01 3.92710924e-01 -5.26852429e-01 5.89036763e-01 1.55590725e+00 -2.98597634e-01 -2.64339060e-01 -1.43386386e-02 -6.79849744e-01 6.24766111e-01 8.36091757e-01 -6.45332396e-01 -5.81209183e-01 -4.17469233e-01 -3.16854626e-01 1.48556933e-01 3.29272747e-01 -1.12816727e+00 1.09690227e-01 -8.60497952e-02 6.41244650e-01 -5.14311075e-01 4.66083556e-01 -8.50392699e-01 4.55892971e-03 5.33456922e-01 -3.63630027e-01 -5.15619628e-02 -5.12563549e-02 6.45319700e-01 -5.88010371e-01 -4.77288067e-01 1.18153393e+00 -2.82148391e-01 -7.86542058e-01 1.09903939e-01 -1.44115865e-01 -1.47267252e-01 1.13108158e+00 -4.05995905e-01 2.01386005e-01 -7.13526666e-01 -7.24472821e-01 1.01957083e-01 6.68314099e-01 3.75267118e-01 1.06180418e+00 -1.36302543e+00 -6.47816479e-01 4.62504685e-01 1.42054364e-01 -5.39104827e-02 5.32792881e-02 3.26675981e-01 -9.44072843e-01 -2.30748162e-01 -5.61332643e-01 -2.96828330e-01 -1.23989654e+00 6.75545931e-01 -1.86461434e-01 -3.38615447e-01 -7.90076315e-01 6.87415540e-01 6.20409213e-02 -1.26311451e-01 5.62731624e-01 -1.08419187e-01 -3.72219495e-02 -1.57886341e-01 6.42976046e-01 2.82991081e-01 -5.22568747e-02 -4.99517173e-01 1.25319794e-01 5.71712136e-01 -2.93154866e-01 -1.84278160e-01 1.61142004e+00 -1.60122260e-01 -9.04719830e-02 1.71871915e-01 1.33985889e+00 5.38619533e-02 -1.59725261e+00 -4.04570162e-01 -1.39012933e-01 -9.00876284e-01 2.25596383e-01 -7.98835993e-01 -1.18027782e+00 6.83083355e-01 7.51662910e-01 1.28045350e-01 1.67781150e+00 -2.39691600e-01 1.20652032e+00 -7.48430714e-02 2.60048449e-01 -1.24092686e+00 6.38739347e-01 3.36138834e-03 1.04893279e+00 -1.31723797e+00 2.47638226e-01 -5.28382897e-01 -7.47726440e-01 1.11944497e+00 4.47993398e-01 -3.60380799e-01 4.17382568e-01 2.52564937e-01 2.51641929e-01 5.70743456e-02 -4.42571133e-01 -2.62687743e-01 1.65616572e-01 7.18000412e-01 1.26491621e-01 -3.83250743e-01 -4.98006970e-01 1.16962954e-01 1.77883834e-01 9.68547836e-02 3.54655951e-01 1.18950284e+00 -5.19331217e-01 -1.59026968e+00 -4.61923063e-01 3.01839560e-01 -3.87160003e-01 1.97567835e-01 -3.62041742e-01 4.94075239e-01 2.51627535e-01 9.66210604e-01 -3.26739073e-01 -4.03361261e-01 2.23807082e-01 -1.53073877e-01 7.23236084e-01 -6.55959308e-01 -2.72469968e-01 2.37340227e-01 -1.33659035e-01 -6.41880214e-01 -6.43122196e-01 -1.86397627e-01 -1.03828418e+00 -4.44316268e-02 -2.63931930e-01 -2.70917974e-02 5.06457388e-01 6.14513338e-01 2.91684717e-01 2.71078557e-01 1.10143685e+00 -1.02187419e+00 -3.72711033e-01 -6.49198949e-01 -5.88040888e-01 8.32228541e-01 5.81076026e-01 -5.07109821e-01 -2.82324612e-01 3.89467567e-01]
[11.366942405700684, -1.1124476194381714]
075f5276-75ca-4635-9ae0-fd307ff8a98f
simrod-a-simple-adaptation-method-for-robust
2107.13389
null
https://arxiv.org/abs/2107.13389v1
https://arxiv.org/pdf/2107.13389v1.pdf
SimROD: A Simple Adaptation Method for Robust Object Detection
This paper presents a Simple and effective unsupervised adaptation method for Robust Object Detection (SimROD). To overcome the challenging issues of domain shift and pseudo-label noise, our method integrates a novel domain-centric augmentation method, a gradual self-labeling adaptation procedure, and a teacher-guided fine-tuning mechanism. Using our method, target domain samples can be leveraged to adapt object detection models without changing the model architecture or generating synthetic data. When applied to image corruptions and high-level cross-domain adaptation benchmarks, our method outperforms prior baselines on multiple domain adaptation benchmarks. SimROD achieves new state-of-the-art on standard real-to-synthetic and cross-camera setup benchmarks. On the image corruption benchmark, models adapted with our method achieved a relative robustness improvement of 15-25% AP50 on Pascal-C and 5-6% AP on COCO-C and Cityscapes-C. On the cross-domain benchmark, our method outperformed the best baseline performance by up to 8% AP50 on Comic dataset and up to 4% on Watercolor dataset.
['Yong Zhang', 'Xiaolong Bai', 'Xinyu Kang', 'Amin Banitalebi-Dehkordi', 'Rindra Ramamonjison']
2021-07-28
null
http://openaccess.thecvf.com//content/ICCV2021/html/Ramamonjison_SimROD_A_Simple_Adaptation_Method_for_Robust_Object_Detection_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Ramamonjison_SimROD_A_Simple_Adaptation_Method_for_Robust_Object_Detection_ICCV_2021_paper.pdf
iccv-2021-1
['robust-object-detection']
['computer-vision']
[ 3.13897699e-01 -2.33721837e-01 -7.99357668e-02 -3.54353160e-01 -1.35737717e+00 -7.87848711e-01 7.85498857e-01 -1.12685822e-01 -7.48137593e-01 6.32856727e-01 -7.59728029e-02 1.65817022e-01 5.69464922e-01 -2.41495311e-01 -8.88739705e-01 -5.03354967e-01 3.44689637e-01 4.89021778e-01 8.94485772e-01 -2.16593951e-01 3.27911898e-02 3.31630439e-01 -1.34330487e+00 4.20035899e-01 9.17991579e-01 7.71920085e-01 1.19713955e-01 7.45827436e-01 1.43629223e-01 5.22445083e-01 -7.22034395e-01 -4.43227828e-01 5.19943953e-01 -1.28041387e-01 -8.24434102e-01 4.86240089e-01 1.03013945e+00 -3.12621415e-01 -1.76851422e-01 1.23683286e+00 6.89415991e-01 8.88507664e-02 7.05523610e-01 -1.23815453e+00 -6.72313213e-01 1.35404959e-01 -9.99660134e-01 2.93594152e-01 1.62555292e-01 7.19066560e-01 4.22775596e-01 -1.20888126e+00 8.38110566e-01 1.30457950e+00 7.34568775e-01 7.42672026e-01 -1.82930851e+00 -9.18132842e-01 3.08836788e-01 2.45999411e-01 -1.41344106e+00 -4.66859639e-01 5.46005905e-01 -5.69807887e-01 9.15133119e-01 -2.69421875e-01 1.12437673e-01 1.58347595e+00 -3.27778310e-01 8.08870137e-01 1.43950117e+00 -4.28521514e-01 3.62559974e-01 3.44054371e-01 -4.97278012e-02 3.84521484e-01 2.57494509e-01 2.57807970e-01 -3.24271351e-01 3.00139543e-02 5.50265253e-01 -6.22218013e-01 -2.10822262e-02 -3.80850941e-01 -1.18361342e+00 6.01045787e-01 4.00720239e-01 -2.59786665e-01 -5.25355227e-02 -1.09352600e-02 6.01067483e-01 1.36361316e-01 5.57319939e-01 6.50107205e-01 -6.73479378e-01 6.80545419e-02 -8.26766670e-01 3.43054354e-01 4.29057091e-01 1.24376333e+00 6.02823734e-01 3.07574451e-01 -4.02439028e-01 1.02002466e+00 1.52343034e-03 9.05831277e-01 3.28849345e-01 -1.23243451e+00 5.51652133e-01 4.33063775e-01 4.64298695e-01 -5.29872417e-01 -2.79847056e-01 -5.70219457e-01 -5.87355912e-01 5.07327914e-01 5.47267675e-01 -1.70399994e-01 -1.45563924e+00 1.77208173e+00 4.53287572e-01 2.44452506e-01 9.73201841e-02 8.70246828e-01 7.82263637e-01 4.71291751e-01 4.74547058e-01 1.77620575e-01 1.27542520e+00 -1.28743052e+00 -4.02994722e-01 -5.75487137e-01 3.70436400e-01 -8.37083340e-01 1.45551085e+00 4.27916139e-01 -1.02383363e+00 -8.43468010e-01 -1.17214942e+00 7.11233243e-02 -3.44286680e-01 2.74918258e-01 1.06783137e-02 5.85375369e-01 -7.27173805e-01 1.44872129e-01 -6.62998557e-01 -6.49263799e-01 7.53219008e-01 3.08384560e-02 -5.12448430e-01 -4.64166611e-01 -7.59006619e-01 8.81644607e-01 5.16967833e-01 -6.59251392e-01 -1.42776024e+00 -1.06761134e+00 -6.94101572e-01 -3.94910425e-01 3.86132330e-01 -6.29042089e-01 1.34847546e+00 -1.05169129e+00 -1.48699450e+00 1.22906065e+00 3.07636440e-01 -7.09529042e-01 7.89346933e-01 -4.77670282e-01 -7.83740878e-01 2.07748413e-01 4.09575492e-01 1.16090536e+00 1.04838145e+00 -1.50122154e+00 -8.17309380e-01 -4.08115387e-02 -1.38194159e-01 1.65675402e-01 -1.60386235e-01 1.48621917e-01 -8.76183510e-01 -9.18857396e-01 -2.58784831e-01 -1.15136945e+00 -2.83687770e-01 1.28598318e-01 -2.38108084e-01 1.75080508e-01 1.05208802e+00 -4.52925891e-01 7.21684992e-01 -2.20553946e+00 -1.03813477e-01 -3.20136964e-01 -1.09811582e-01 6.67217433e-01 -6.01805627e-01 -5.04518375e-02 -1.68475628e-01 -1.16858020e-01 -6.37266338e-01 -3.91055971e-01 -6.91318139e-02 1.93953663e-01 -2.19451278e-01 5.52202702e-01 7.95788467e-01 5.92722178e-01 -9.72239614e-01 -4.33991849e-01 2.18837738e-01 3.13787818e-01 -6.98659718e-01 3.22690576e-01 -4.91491526e-01 6.03980601e-01 2.10804477e-01 6.77418947e-01 1.13263309e+00 -3.22975755e-01 -1.20687589e-01 -1.10867627e-01 2.03618452e-01 -3.32848281e-02 -1.27254879e+00 1.86352324e+00 -3.05842370e-01 7.02512324e-01 -1.67029351e-02 -5.99852860e-01 7.42549777e-01 -3.08513213e-02 -4.63228151e-02 -9.80144680e-01 -3.08555275e-01 1.61382779e-01 -1.24308094e-01 -2.52294540e-01 4.56118524e-01 2.96943128e-01 -1.23185277e-01 1.85117438e-01 4.43253517e-01 -3.41032386e-01 2.89209276e-01 4.73151445e-01 1.19008017e+00 3.84985179e-01 1.66892514e-01 -4.20059860e-01 5.48630476e-01 3.11421007e-01 7.07116127e-01 7.95952678e-01 -4.27624524e-01 1.06410730e+00 2.29901001e-01 -4.43365932e-01 -1.23076367e+00 -1.28733349e+00 -1.80394098e-01 1.41806114e+00 2.50036985e-01 -9.28316712e-02 -8.69920135e-01 -1.03626657e+00 8.80993158e-02 8.47885549e-01 -6.71296835e-01 -1.30683154e-01 -2.96399891e-01 -9.66920912e-01 8.30534816e-01 6.55550122e-01 1.01876712e+00 -8.27631712e-01 -1.83054835e-01 1.58292443e-01 -2.66603827e-01 -1.83252978e+00 -6.62675977e-01 2.91601747e-01 -6.80262387e-01 -9.70784128e-01 -7.73610771e-01 -7.18042910e-01 7.39411533e-01 3.87366652e-01 1.46474195e+00 -5.06976843e-01 -4.37750340e-01 3.54515076e-01 -2.56906480e-01 -3.50986451e-01 -5.75629056e-01 2.00448167e-02 9.94711444e-02 -1.20601254e-02 2.97944397e-01 -3.17639500e-01 -7.47771561e-01 6.89700663e-01 -9.07772362e-01 -5.85808232e-02 5.44577718e-01 9.29012597e-01 7.25927293e-01 -5.29682517e-01 6.30804002e-01 -1.00012124e+00 1.46817416e-01 -2.85446018e-01 -1.03271353e+00 8.55171233e-02 -7.75508702e-01 7.10863322e-02 4.96860951e-01 -7.56720185e-01 -1.39825118e+00 4.95697826e-01 1.98867917e-01 -5.26545644e-01 -3.89236718e-01 -3.57099473e-01 -3.46678883e-01 -2.61505187e-01 1.55207014e+00 1.65067837e-01 -4.24214751e-01 -5.11879683e-01 6.66444898e-01 5.20661473e-01 1.14392364e+00 -7.59090543e-01 1.14440811e+00 6.42110825e-01 -4.10072267e-01 -5.84964156e-01 -8.93892586e-01 -6.46957278e-01 -7.64008880e-01 6.91119656e-02 8.94882321e-01 -1.59595811e+00 1.88693568e-01 7.69458592e-01 -1.11132932e+00 -7.29176164e-01 -3.39850575e-01 7.13925287e-02 -5.06108701e-01 3.06638539e-01 -4.90275025e-01 -2.31944293e-01 -1.46520466e-01 -1.07225323e+00 1.15755892e+00 1.02098905e-01 8.23388845e-02 -6.73463285e-01 1.66987121e-01 4.10959840e-01 4.43014562e-01 3.34451467e-01 4.22498107e-01 -5.65845251e-01 -4.63448733e-01 -5.15505783e-02 -6.93305492e-01 7.15031743e-01 -9.51018035e-02 -1.37520581e-01 -1.14878368e+00 -5.13413668e-01 -5.99520683e-01 -6.50162160e-01 9.56964672e-01 -1.55220807e-01 1.13864577e+00 -3.25300880e-02 -2.96016186e-01 8.66503298e-01 1.42716098e+00 -1.56814501e-01 7.26517260e-01 6.87841892e-01 5.95893741e-01 9.79314670e-02 8.13487411e-01 3.41619194e-01 3.24400991e-01 9.36461985e-01 4.25633848e-01 -3.29276770e-01 -7.84146786e-01 -1.48153052e-01 4.43882436e-01 8.18536729e-02 2.17047423e-01 -3.28491293e-02 -1.09544492e+00 9.45930660e-01 -1.65045249e+00 -7.25767434e-01 -1.16508819e-01 2.17068315e+00 1.11859298e+00 3.38132262e-01 4.47421104e-01 -2.70924807e-01 7.27294207e-01 -1.52703404e-01 -7.91283846e-01 -4.36092578e-02 -4.70177293e-01 1.14186488e-01 8.57944787e-01 2.85661340e-01 -1.54897547e+00 1.38744950e+00 6.55094957e+00 8.07220340e-01 -1.03549874e+00 4.84784454e-01 6.12095654e-01 -8.66716206e-02 2.71185368e-01 -2.60532200e-01 -9.40392017e-01 5.01553237e-01 7.36043870e-01 1.50913030e-01 2.26988167e-01 1.11334026e+00 -5.18053286e-02 -1.77562788e-01 -9.90953684e-01 9.74615097e-01 9.91478488e-02 -1.08082700e+00 -1.02085099e-01 -1.56768158e-01 1.42929614e+00 5.68811357e-01 2.54309058e-01 4.94526893e-01 9.09924686e-01 -7.14945018e-01 6.68779969e-01 -7.82807916e-02 1.19453239e+00 -5.25472820e-01 5.04334867e-01 3.91184092e-02 -8.32776964e-01 -4.91511784e-02 -4.53385860e-01 4.02528077e-01 -1.81452915e-01 3.38696748e-01 -9.73762155e-01 1.90316737e-01 1.01569688e+00 6.14507198e-01 -1.02270734e+00 1.10156393e+00 -4.89851743e-01 8.67035568e-01 -4.02145892e-01 6.36584163e-01 2.49847874e-01 3.97173256e-01 6.65634274e-01 1.86145914e+00 -1.43232808e-01 -6.18102066e-02 1.71219617e-01 7.27604270e-01 -3.21976125e-01 -2.25395709e-01 -3.75758976e-01 3.96035492e-01 5.48447132e-01 1.26904130e+00 -5.24449587e-01 -5.21840632e-01 -4.59248304e-01 1.27874589e+00 2.97056556e-01 6.04134500e-01 -1.17939520e+00 -3.29305112e-01 7.65568256e-01 2.51479268e-01 5.46929657e-01 -8.99445862e-02 -2.98075169e-01 -1.13838160e+00 -9.65495333e-02 -1.28125775e+00 4.46639240e-01 -7.08814621e-01 -1.45079267e+00 5.74088037e-01 1.02684267e-01 -1.52063882e+00 5.78306476e-03 -5.66185892e-01 -2.85713375e-01 7.63402224e-01 -1.71543026e+00 -1.29883718e+00 -6.23612940e-01 7.31126487e-01 6.94614112e-01 -5.00546992e-01 7.90615857e-01 4.63707030e-01 -5.99893093e-01 9.22049880e-01 3.08086455e-01 2.42711514e-01 1.38706017e+00 -1.37618470e+00 8.91468704e-01 1.32881379e+00 2.97703333e-02 1.14194080e-02 6.11670911e-01 -4.43450063e-01 -8.76830161e-01 -1.77717853e+00 1.87011942e-01 -8.70580435e-01 6.13939703e-01 -4.62120056e-01 -1.05707407e+00 6.31668985e-01 3.66199136e-01 7.59420455e-01 3.11865747e-01 -1.62596330e-01 -1.11410749e+00 -3.29605550e-01 -1.41956699e+00 6.06203556e-01 1.07090473e+00 -3.45764309e-01 -5.57208121e-01 4.88535732e-01 8.89884531e-01 -6.17414296e-01 -6.69476449e-01 4.59718347e-01 2.23248437e-01 -7.61502802e-01 1.08543169e+00 -6.46872759e-01 3.52778614e-01 -7.07551241e-01 -3.39308202e-01 -1.38680756e+00 -5.26167095e-01 -6.18086278e-01 9.60267931e-02 1.28106570e+00 4.26424205e-01 -3.54074270e-01 6.91115916e-01 3.10070097e-01 -1.02447145e-01 1.12558804e-01 -8.16977024e-01 -1.11075974e+00 1.70679688e-01 -4.39555466e-01 2.64714152e-01 9.95945036e-01 -6.71775460e-01 2.56318927e-01 -3.18756044e-01 4.60542738e-01 9.68087673e-01 -3.25208217e-01 1.28748655e+00 -8.52424264e-01 -3.32800657e-01 -2.52006114e-01 -3.28278661e-01 -9.97512937e-01 8.09794068e-02 -6.04718506e-01 1.91892847e-01 -1.04815924e+00 2.41834179e-01 -2.14119047e-01 -4.15363908e-01 5.20240307e-01 -4.04154152e-01 6.61933720e-01 3.79062623e-01 1.43319532e-01 -1.13708162e+00 3.82322133e-01 1.02296925e+00 -4.07477111e-01 -2.42714360e-01 -3.24457973e-01 -6.91620290e-01 6.83713734e-01 5.85650980e-01 -8.15108716e-01 -2.17320710e-01 -7.52208471e-01 -2.45636195e-01 -4.87799436e-01 5.15035212e-01 -1.42045367e+00 6.12701923e-02 -1.37779728e-01 5.42538106e-01 -4.72830445e-01 2.56693065e-01 -7.34353781e-01 -2.52999961e-01 3.57398510e-01 -2.57770211e-01 -1.47751987e-01 8.59664619e-01 8.46492290e-01 -4.49980497e-02 2.86457717e-01 1.49008667e+00 3.73316966e-02 -1.24854064e+00 5.75275086e-02 -1.14149325e-01 6.54804051e-01 1.07575500e+00 -8.08146298e-02 -5.83188891e-01 -3.27966437e-02 -5.25817454e-01 4.07935679e-01 6.60630286e-01 6.60758853e-01 4.11353737e-01 -1.30503201e+00 -1.08636951e+00 1.43032148e-01 8.31989050e-01 2.05293611e-01 4.75096926e-02 5.45777798e-01 -5.59841216e-01 4.11730073e-02 -3.74088466e-01 -1.12976885e+00 -1.21701109e+00 4.67003882e-01 4.67360705e-01 -1.89090014e-01 -3.98074567e-01 1.12906265e+00 4.05271739e-01 -5.17369807e-01 2.68224239e-01 -1.99262396e-01 2.61644185e-01 -2.26818100e-01 6.90342724e-01 2.94072032e-01 2.09361658e-01 -5.46309531e-01 -5.75242162e-01 4.91586775e-01 -4.00617689e-01 -2.08618656e-01 1.20658863e+00 -8.34342316e-02 5.08491516e-01 5.54966480e-02 9.52392697e-01 -1.62045673e-01 -2.01154804e+00 -5.98800600e-01 5.01535386e-02 -4.93740529e-01 -6.04132004e-02 -1.54559219e+00 -8.73163879e-01 7.27744579e-01 1.10582435e+00 -5.71092010e-01 1.29812241e+00 -7.79920220e-02 4.02144432e-01 3.23573142e-01 2.57341653e-01 -1.34991860e+00 5.68929076e-01 5.46969414e-01 9.04928505e-01 -1.81267309e+00 2.13743653e-02 -2.13845611e-01 -1.00758708e+00 5.89355469e-01 1.05467141e+00 -2.04795867e-01 2.11584374e-01 3.59550178e-01 3.49512428e-01 3.98032337e-01 -7.11375296e-01 -4.11553800e-01 3.05315554e-01 1.17397583e+00 2.29213592e-02 -8.06323886e-02 2.42399469e-01 3.52837592e-01 3.80886525e-01 -5.37127592e-02 4.59321886e-01 5.64438701e-01 -3.51493478e-01 -1.00892782e+00 -6.18466377e-01 -1.84986696e-01 -4.64333504e-01 -5.30661494e-02 -3.53775948e-01 1.15275180e+00 2.04737514e-01 9.19800460e-01 -4.64052800e-03 -2.41522357e-01 6.31794333e-01 5.25756832e-03 4.20448869e-01 -7.40161955e-01 -5.31431496e-01 1.69740662e-01 4.48742099e-02 -6.73144877e-01 -3.76449883e-01 -8.00861180e-01 -9.92286861e-01 -1.03506893e-02 -1.22287951e-01 -3.86956245e-01 5.86994410e-01 5.54043531e-01 5.72336793e-01 5.15473425e-01 3.92347306e-01 -8.34186375e-01 -5.35540223e-01 -1.09818482e+00 -1.27083436e-01 7.84705400e-01 2.44633228e-01 -6.46592319e-01 -1.84804186e-01 4.61499184e-01]
[9.72154712677002, 1.7019108533859253]
efa0e3cc-36f4-4584-887d-f1ad223fe90e
automatic-cattle-identification-using-yolov5
2210.11939
null
https://arxiv.org/abs/2210.11939v1
https://arxiv.org/pdf/2210.11939v1.pdf
Automatic Cattle Identification using YOLOv5 and Mosaic Augmentation: A Comparative Analysis
You Only Look Once (YOLO) is a single-stage object detection model popular for real-time object detection, accuracy, and speed. This paper investigates the YOLOv5 model to identify cattle in the yards. The current solution to cattle identification includes radio-frequency identification (RFID) tags. The problem occurs when the RFID tag is lost or damaged. A biometric solution identifies the cattle and helps to assign the lost or damaged tag or replace the RFID-based system. Muzzle patterns in cattle are unique biometric solutions like a fingerprint in humans. This paper aims to present our recent research in utilizing five popular object detection models, looking at the architecture of YOLOv5, investigating the performance of eight backbones with the YOLOv5 model, and the influence of mosaic augmentation in YOLOv5 by experimental results on the available cattle muzzle images. Finally, we concluded with the excellent potential of using YOLOv5 in automatic cattle identification. Our experiments show YOLOv5 with transformer performed best with mean Average Precision (mAP) 0.5 (the average of AP when the IoU is greater than 50%) of 0.995, and mAP 0.5:0.95 (the average of AP from 50% to 95% IoU with an interval of 5%) of 0.9366. In addition, our experiments show the increase in accuracy of the model by using mosaic augmentation in all backbones used in our experiments. Moreover, we can also detect cattle with partial muzzle images.
['Will Swain', 'Dave Swain', 'Jonathan Medway', 'Shawn McGrath', 'Muhammad Ashad Kabir', 'Lihong Zheng', 'Rabin Dulal']
2022-10-21
null
null
null
null
['real-time-object-detection']
['computer-vision']
[-1.31036593e-02 -1.67553842e-01 -3.15388739e-02 -2.04203531e-01 2.78675072e-02 -4.95779037e-01 1.94966316e-01 6.37144819e-02 -2.40206122e-01 3.17709148e-01 -5.23225069e-01 -1.04357518e-01 -3.78188640e-01 -9.21546102e-01 -9.19338405e-01 -6.42999411e-01 -4.26375002e-01 5.84193468e-01 4.14171576e-01 -8.71049911e-02 9.29124132e-02 6.97859824e-01 -1.68118310e+00 2.40188137e-01 3.90557945e-01 1.23039937e+00 5.08511364e-01 6.72031701e-01 7.54042491e-02 4.70395476e-01 -6.77083731e-01 -3.96961391e-01 2.97539979e-01 -4.94558476e-02 -5.59438765e-01 -2.55446643e-01 4.30518478e-01 -6.75773501e-01 2.60329008e-01 7.60543942e-01 3.76928598e-01 -6.13734901e-01 5.61872244e-01 -1.46390855e+00 -7.32280791e-01 7.81794369e-01 -7.93458521e-01 -1.15380749e-01 2.55907863e-01 -1.63015276e-01 2.65484929e-01 -4.45482254e-01 5.05062521e-01 1.27790511e+00 1.31527126e+00 2.50628263e-01 -1.54386437e+00 -7.97837794e-01 -7.40600884e-01 2.10206449e-01 -1.59414816e+00 -5.01411855e-01 1.62572965e-01 -4.28117037e-01 5.81925631e-01 5.11124909e-01 4.39905286e-01 2.37718552e-01 5.72070837e-01 6.13670766e-01 9.38431978e-01 -8.86481285e-01 -3.05530816e-01 3.78647447e-01 3.43583196e-01 6.22637928e-01 6.00012779e-01 2.33502910e-01 -1.63004472e-04 -1.48594692e-01 9.24954891e-01 2.00159382e-02 2.85109669e-01 -6.84389174e-01 -1.16798568e+00 5.12967706e-01 1.02014914e-01 3.26849312e-01 -5.87593198e-01 3.34102154e-01 1.16291463e-01 3.77261937e-01 -3.40201668e-02 5.38784862e-01 -4.21410114e-01 3.47925186e-01 -8.30301106e-01 2.28589341e-01 7.31619239e-01 1.35712194e+00 6.27311945e-01 -2.05856040e-01 -1.94772094e-01 9.64402139e-01 5.24737060e-01 1.30874920e+00 -1.46230221e-01 -9.46509600e-01 -2.33808264e-01 4.01527643e-01 6.04720771e-01 -1.02614152e+00 -3.32199305e-01 -1.17809638e-01 -5.22114396e-01 -2.34703980e-02 8.33957493e-01 6.74465522e-02 -8.89599621e-01 1.31512892e+00 2.14449778e-01 -2.97265440e-01 -1.10833310e-01 5.71325660e-01 7.76693285e-01 7.22516477e-01 5.50080510e-03 5.27619384e-02 2.09291768e+00 -4.37331647e-01 -8.50828290e-01 2.29743317e-01 4.92751479e-01 -1.40150273e+00 2.08379582e-01 2.69493401e-01 -6.10179305e-01 -6.80926740e-01 -8.97483945e-01 6.66522980e-01 -2.65903831e-01 7.34912872e-01 4.22671258e-01 1.17004037e+00 -1.05980265e+00 2.35307753e-01 -6.92259729e-01 -1.02604699e+00 -2.21224338e-01 5.99230647e-01 -4.06029522e-01 -2.28498019e-02 -1.05255830e+00 1.19715405e+00 1.97181910e-01 3.65258485e-01 -3.65169644e-01 -6.05758250e-01 -4.67727214e-01 -1.43202692e-01 1.24359131e-01 -1.96208194e-01 1.06397903e+00 -6.25922918e-01 -1.22932172e+00 1.18057191e+00 1.66299224e-01 -7.55799592e-01 5.24979055e-01 -3.84684116e-01 -6.56068206e-01 2.27522209e-01 2.48624787e-01 7.25388944e-01 4.87869710e-01 -1.24145472e+00 -8.30217242e-01 -2.46561155e-01 -3.48992765e-01 -8.34065676e-01 4.66641039e-01 2.69998848e-01 -1.70186564e-01 -4.02728051e-01 4.34349000e-01 -1.02279234e+00 2.95749187e-01 -1.50833115e-01 1.90780655e-01 -4.79211025e-02 7.88726091e-01 -8.27420950e-01 9.57453549e-01 -1.90939665e+00 -8.96753550e-01 4.74492311e-01 -4.22712207e-01 5.11977255e-01 -1.30979434e-01 6.07617557e-01 2.96426952e-01 -4.50740039e-01 2.93651283e-01 6.62438691e-01 1.01628095e-01 -1.46353338e-02 -9.19967424e-03 5.52416682e-01 8.91491398e-02 7.43447602e-01 -6.65299356e-01 -4.65700120e-01 4.55240667e-01 4.77959871e-01 -2.30176374e-01 1.30577669e-01 4.79257792e-01 -3.30561191e-01 -2.14063928e-01 1.15827918e+00 1.30589318e+00 2.41820306e-01 3.03957522e-01 -7.30664790e-01 -4.45291817e-01 -5.31260133e-01 -1.35530555e+00 8.23684573e-01 -2.11560950e-01 4.47573364e-01 2.12723657e-01 -7.83456564e-01 1.29033911e+00 3.39517683e-01 7.42980540e-01 -6.98380232e-01 3.12849507e-02 2.83223391e-01 -1.17238589e-01 -5.55916190e-01 7.09447384e-01 2.87778944e-01 -1.73639968e-01 2.21527204e-01 3.37678909e-01 5.62166989e-01 3.09867471e-01 -2.17773587e-01 8.09693098e-01 3.17837924e-01 1.93101078e-01 -8.37293684e-01 2.83276945e-01 2.02062145e-01 2.88692117e-01 1.17480099e+00 -1.27002120e-01 4.50559855e-01 1.63031146e-01 -3.81626725e-01 -1.03282702e+00 -1.02168679e+00 -5.79088926e-01 1.04806566e+00 3.93398345e-01 1.88886166e-01 -8.51517320e-01 -5.19958884e-02 5.58771908e-01 5.23447275e-01 -4.93096024e-01 -2.70526204e-02 -5.49114883e-01 -9.91977811e-01 8.58903825e-01 4.23382461e-01 7.97403395e-01 -9.24602687e-01 -1.07144701e+00 5.24564743e-01 -3.07597041e-01 -8.85730267e-01 -1.43195800e-02 2.13500589e-01 -8.67851973e-01 -9.49967623e-01 -1.00723958e+00 -8.94827187e-01 6.15939319e-01 4.50806111e-01 9.38217759e-01 -3.94216217e-02 -4.38534349e-01 3.75305682e-01 -5.76704085e-01 -3.70883048e-01 -5.72620988e-01 -1.75734222e-01 9.52565484e-03 -1.74968749e-01 9.55261409e-01 2.46407151e-01 -4.70607132e-01 9.59808528e-01 -5.70645750e-01 -3.31103921e-01 7.55020261e-01 9.32434082e-01 1.01592734e-01 -2.65699536e-01 5.41355848e-01 -7.25502491e-01 -9.41258669e-02 -3.73729199e-01 -1.00788343e+00 6.58839047e-01 -6.71397448e-01 9.44435298e-02 -1.62877724e-01 -2.76086181e-01 -7.01831639e-01 1.61934212e-01 -5.67645393e-02 8.15250427e-02 -4.30751741e-01 -9.47219133e-02 -2.44940985e-02 -3.50038320e-01 2.81841993e-01 -9.36764479e-02 3.46618205e-01 -8.94725859e-01 2.72267740e-02 7.80889153e-01 8.73834312e-01 -4.68364298e-01 2.49290124e-01 1.29634783e-01 -1.15039356e-01 -9.78679538e-01 2.59554297e-01 -5.96621335e-01 -5.27973175e-01 -5.04222870e-01 5.73104978e-01 -6.78856552e-01 -1.29599202e+00 1.04118514e+00 -1.02507591e+00 1.23925857e-01 2.19267368e-01 7.88088322e-01 -3.72920275e-01 2.12874696e-01 -6.91955149e-01 -1.06087410e+00 -3.50627244e-01 -7.86307037e-01 9.40571427e-01 7.65326098e-02 -1.98907658e-01 -1.43517077e-01 -2.66954243e-01 3.29653352e-01 3.77457201e-01 1.39523447e-01 6.58690393e-01 -3.20827305e-01 -3.60718369e-01 -5.27744532e-01 -5.03475070e-01 -6.30060062e-02 1.78872466e-01 2.03104526e-01 -8.36401343e-01 -2.42369890e-01 -4.11967874e-01 3.30961287e-01 7.99149573e-01 8.00723910e-01 3.64472777e-01 -3.04168817e-02 -6.43296540e-01 3.71450961e-01 1.56742966e+00 7.70211399e-01 8.82234216e-01 6.60284221e-01 1.44893304e-01 6.65549040e-01 1.26940012e+00 2.50168145e-01 2.75051504e-01 1.00800896e+00 2.67617613e-01 5.85556999e-02 1.13393888e-02 -1.00762188e-01 2.94694066e-01 2.64787018e-01 -4.45508212e-01 -1.78808197e-01 -9.76792157e-01 5.85783064e-01 -1.59047306e+00 -1.17291343e+00 -3.31380188e-01 2.25817823e+00 3.88704658e-01 -2.21390218e-01 2.15610728e-01 2.00529560e-01 1.24995732e+00 -5.47534943e-01 2.26511657e-01 -6.57523453e-01 1.36153147e-01 1.44558260e-02 1.35904336e+00 7.73907229e-02 -1.25830984e+00 7.17771888e-01 6.55574799e+00 5.06825626e-01 -9.01234686e-01 -1.92495480e-01 1.88012451e-01 4.14051473e-01 4.52155530e-01 -3.17512989e-01 -9.87534165e-01 6.43194199e-01 6.40877128e-01 4.60809916e-01 6.11291192e-02 6.88200653e-01 -1.39431953e-01 -7.72331655e-01 -9.63893354e-01 7.10278332e-01 -1.67561367e-01 -8.52651596e-01 -3.39678913e-01 -3.32111828e-02 3.36454183e-01 -5.13427317e-01 -2.01043010e-01 4.54112627e-02 2.85571873e-01 -5.22787333e-01 9.74376857e-01 6.57231212e-01 8.52582693e-01 -6.64910614e-01 1.45652151e+00 -4.90779877e-02 -1.29228127e+00 2.02826858e-01 -5.84140897e-01 3.12543899e-01 -7.78956339e-02 1.94022551e-01 -1.12899363e+00 4.76423055e-01 1.03845263e+00 -1.77757934e-01 -4.92542595e-01 1.31880069e+00 5.06380916e-01 3.80589664e-01 -7.55338788e-01 -2.17479676e-01 6.93553537e-02 -1.45026281e-01 1.44855425e-01 1.48672998e+00 8.65593314e-01 6.29026666e-02 -5.76835990e-01 6.72545969e-01 4.77449328e-01 -1.29560724e-01 -3.96981359e-01 2.57847816e-01 8.64052653e-01 8.01787555e-01 -9.42448378e-01 1.32119969e-01 -1.32210925e-01 5.42149782e-01 -8.56156647e-01 1.38203502e-02 -8.15095425e-01 -6.48258686e-01 2.75305271e-01 3.41724426e-01 6.76179171e-01 2.57070929e-01 4.37630005e-02 -3.84350240e-01 -1.92121223e-01 -5.17183244e-01 1.72730431e-01 -7.29146004e-01 -8.80041718e-01 1.89460859e-01 3.33321393e-01 -1.24149835e+00 -2.58457541e-01 -5.12193322e-01 9.57162157e-02 7.70437598e-01 -7.07386851e-01 -1.54418695e+00 -1.63608089e-01 -1.15535229e-01 -7.88786486e-02 -2.48713955e-01 8.97431314e-01 6.24822021e-01 8.96586478e-02 9.43487525e-01 6.85411692e-01 1.55122012e-01 7.83985019e-01 -7.85823882e-01 2.44381890e-01 5.84374905e-01 -3.64056140e-01 7.49173939e-01 8.94228816e-01 -6.62496328e-01 -1.70577073e+00 -8.41585994e-01 1.08055627e+00 -1.68140024e-01 1.41901180e-01 -1.35321721e-01 -5.88496864e-01 5.48401415e-01 -1.42459661e-01 -3.22194993e-01 2.76169866e-01 -9.93227288e-02 9.34103653e-02 -7.32920587e-01 -1.77599621e+00 -1.37876257e-01 5.36196828e-01 -7.25679919e-02 -3.89372587e-01 -4.71932888e-01 -8.99224952e-02 -1.09280713e-01 -1.09990752e+00 6.62005246e-01 1.79875684e+00 -8.45668674e-01 1.15958655e+00 1.50608584e-01 -1.03403814e-01 -4.40835267e-01 -1.96157262e-01 -8.31858993e-01 -5.57430506e-01 -5.95908090e-02 3.91681463e-01 1.74530268e+00 6.15938120e-02 -6.67590916e-01 4.95928466e-01 1.19852647e-01 4.07798707e-01 1.81016196e-02 -5.68979740e-01 -9.52494264e-01 -6.69009626e-01 7.52679035e-02 9.96087849e-01 7.59757578e-01 -3.52990001e-01 -4.44092870e-01 -7.70606756e-01 3.16381723e-01 9.98395562e-01 2.95539856e-01 7.76539981e-01 -1.24947524e+00 1.28111303e-01 1.57677680e-02 -9.09145236e-01 -6.78825438e-01 -8.26682687e-01 -4.18972343e-01 2.48732239e-01 -1.18447781e+00 1.61086544e-01 -7.72196531e-01 -4.02406365e-01 4.26624805e-01 2.72643894e-01 4.63106811e-01 2.83694416e-01 1.43415481e-01 -1.38694584e-01 -5.11223197e-01 5.95234334e-01 -1.78417653e-01 -2.62295995e-02 3.71221036e-01 -2.76065636e-02 4.05970037e-01 6.08962119e-01 -5.79241574e-01 3.65423352e-01 -3.94412339e-01 -1.82690218e-01 2.64396936e-01 6.34183884e-01 -1.02688694e+00 -8.27527195e-02 7.64899552e-02 6.44178510e-01 -1.14772487e+00 1.11159952e-02 -1.30451500e+00 1.06219530e+00 1.00712180e+00 1.22992754e-01 -9.77379829e-02 1.53540537e-01 6.61697909e-02 7.02417940e-02 -3.85446310e-01 7.73043275e-01 4.51747961e-02 -1.09866250e+00 -5.10789990e-01 -6.39119267e-01 -9.89348948e-01 9.28489804e-01 -6.77061021e-01 -3.32956582e-01 2.13818848e-01 -6.11918807e-01 2.86660135e-01 6.61054909e-01 4.66933370e-01 1.19531460e-01 -1.44123352e+00 -6.25524819e-01 6.06582165e-01 3.22571307e-01 -8.22545230e-01 7.67435357e-02 8.26568484e-01 -1.27371907e+00 8.03808510e-01 -1.06521177e+00 -1.08320534e+00 -1.68894947e+00 3.94393712e-01 2.07810774e-01 2.34963402e-01 -1.06615745e-01 5.06239951e-01 3.33131365e-02 -3.27606648e-01 2.36572608e-01 -4.43977982e-01 -2.84353435e-01 2.93174416e-01 2.86085308e-01 9.86551404e-01 1.28160551e-01 -7.13764608e-01 -7.83381581e-01 8.03623915e-01 1.78378969e-01 2.46097952e-01 1.21343851e+00 -2.22379103e-01 -2.40654662e-01 1.53677091e-01 7.44559526e-01 -3.57386544e-02 -6.17869854e-01 -6.34764656e-02 3.88160944e-01 -3.61312509e-01 -4.55433846e-01 -9.23710465e-01 -8.95299315e-01 4.05694962e-01 1.51880360e+00 4.91912782e-01 9.89090085e-01 -8.22011530e-02 6.32329226e-01 4.26396519e-01 9.38207924e-01 -8.68771434e-01 -8.89257789e-01 1.56299129e-01 5.78795612e-01 -9.77704227e-01 1.33011475e-01 -3.37335587e-01 -3.28665107e-01 1.00632811e+00 2.41095141e-01 -7.13468120e-02 4.38424915e-01 4.21163112e-01 3.08964401e-01 -1.07435055e-01 -2.13279083e-01 -1.33670613e-01 -3.20450738e-02 8.76969934e-01 5.10863841e-01 5.36753476e-01 -7.04289496e-01 4.83845085e-01 5.86130954e-02 4.44201708e-01 7.23319426e-02 1.10016894e+00 -8.05245817e-01 -1.03752673e+00 -1.16540670e+00 5.24346411e-01 -6.89936161e-01 4.07027096e-01 -6.32227659e-02 8.83507133e-01 5.04567921e-01 1.00646603e+00 2.00300306e-01 -4.84882474e-01 4.82481480e-01 4.24672849e-03 7.91525066e-01 1.01379097e-01 -7.08515942e-01 3.39796841e-01 2.02603593e-01 -3.34883809e-01 -7.04673052e-01 -6.91873193e-01 -7.20819414e-01 -8.60118866e-01 -9.59297836e-01 3.68345320e-01 7.79878676e-01 3.48526418e-01 -3.26594785e-02 5.43625504e-02 4.88565266e-01 -3.72520238e-01 -3.43339741e-01 -1.08825314e+00 -1.15754712e+00 -1.35371864e-01 1.65247068e-01 -9.13035750e-01 2.74756968e-01 3.73318523e-01]
[12.911779403686523, 0.938870370388031]
d9a01af5-752e-4a90-a815-bd1e32bdff10
nuclear-instance-segmentation-using-a
1908.10356
null
https://arxiv.org/abs/1908.10356v1
https://arxiv.org/pdf/1908.10356v1.pdf
Nuclear Instance Segmentation using a Proposal-Free Spatially Aware Deep Learning Framework
Nuclear segmentation in histology images is a challenging task due to significant variations in the shape and appearance of nuclei. One of the main hurdles in nuclear instance segmentation is overlapping nuclei where a smart algorithm is needed to separate each nucleus. In this paper, we introduce a proposal-free deep learning based framework to address these challenges. To this end, we propose a spatially-aware network (SpaNet) to capture spatial information in a multi-scale manner. A dual-head variation of the SpaNet is first utilized to predict the pixel-wise segmentation and centroid detection maps of nuclei. Based on these outputs, a single-head SpaNet predicts the positional information related to each nucleus instance. Spectral clustering method is applied on the output of the last SpaNet, which utilizes the nuclear mask and the Gaussian-like detection map for determining the connected components and associated cluster identifiers, respectively. The output of the clustering method is the final nuclear instance segmentation mask. We applied our method on a publicly available multi-organ data set and achieved state-of-the-art performance for nuclear segmentation.
['Ali Gooya', 'Navid Alemi Koohbanani', 'Mostafa Jahanifar', 'Nasir Rajpoot']
2019-08-27
null
null
null
null
['nuclear-segmentation']
['medical']
[ 1.06912851e-01 -4.32534851e-02 2.32371435e-01 -5.09780407e-01 -9.27640736e-01 -7.18079805e-01 3.20118725e-01 4.16617721e-01 -6.77202344e-01 4.16685313e-01 -1.98812410e-01 1.28649399e-01 -5.55911809e-02 -7.10458577e-01 -5.71592331e-01 -1.13721287e+00 1.56740054e-01 6.57682240e-01 5.88247836e-01 2.67764300e-01 2.31632844e-01 8.27074647e-01 -9.24267471e-01 3.06283832e-01 7.57025659e-01 8.89559209e-01 5.08214891e-01 6.53291881e-01 -1.51465431e-01 1.14725694e-01 -1.41538426e-01 8.39022249e-02 1.19966082e-01 -3.52122694e-01 -7.13313580e-01 1.30398482e-01 2.73582995e-01 -4.90756258e-02 -1.93024844e-01 1.08184087e+00 6.26453876e-01 4.76712063e-02 9.60021079e-01 -7.66056836e-01 5.45648560e-02 8.41678798e-01 -8.32886577e-01 5.19453704e-01 -6.10007226e-01 1.91934124e-01 7.15066791e-01 -7.07481742e-01 6.30695164e-01 7.94255972e-01 6.30051732e-01 2.21480280e-01 -1.35010159e+00 -5.24856448e-01 -2.12189972e-01 -8.77272263e-02 -1.71819973e+00 -2.85323918e-01 5.53996205e-01 -6.49887443e-01 5.28569520e-01 5.03094457e-02 6.81766450e-01 2.83347458e-01 3.14851940e-01 6.39941990e-01 1.01876271e+00 -1.44132137e-01 4.08977598e-01 -1.49898142e-01 7.38272965e-02 6.91700280e-01 -7.68758878e-02 -6.42418027e-01 9.75561291e-02 7.62792602e-02 9.75753427e-01 1.91750869e-01 -1.09296553e-01 -4.43530291e-01 -1.29423261e+00 6.28529429e-01 8.52315426e-01 5.92772603e-01 -4.24143821e-01 1.12691261e-01 3.92521560e-01 -4.80206549e-01 3.06977153e-01 1.38148576e-01 -3.14271450e-01 2.11177915e-01 -1.46054053e+00 6.95002005e-02 4.87585813e-01 2.78915554e-01 8.73448789e-01 -4.06154305e-01 -5.61669052e-01 7.98907280e-01 3.87207776e-01 -5.36804348e-02 4.62039500e-01 -7.91174650e-01 -2.83966330e-03 1.00184321e+00 -1.88651487e-01 -8.89838815e-01 -1.07756722e+00 -8.74991417e-01 -1.05882740e+00 1.26664966e-01 7.34348953e-01 -2.23022491e-01 -1.28928053e+00 1.34066176e+00 7.48421192e-01 4.83691216e-01 -3.42186332e-01 1.12259424e+00 7.78324187e-01 4.63044882e-01 1.09667078e-01 9.31858458e-03 1.45948589e+00 -9.15067255e-01 -2.98485607e-01 7.79906660e-02 8.56231213e-01 -5.21608829e-01 4.65325683e-01 -2.49881342e-01 -8.33354354e-01 -1.86691001e-01 -8.76377463e-01 -4.46449183e-02 -3.32940608e-01 5.50675511e-01 2.23532408e-01 2.35441491e-01 -1.18061125e+00 6.27922595e-01 -1.20020497e+00 -4.55701381e-01 7.33387828e-01 6.83035791e-01 -4.34257179e-01 6.55225739e-02 -5.89286506e-01 6.57697141e-01 6.47227645e-01 3.81188452e-01 -7.45596349e-01 -7.57999241e-01 -7.09375501e-01 1.96488082e-01 7.44284242e-02 -6.23547554e-01 8.66156042e-01 -5.38855970e-01 -1.05041635e+00 9.98187482e-01 -1.08873025e-01 -4.00065660e-01 5.49072802e-01 6.01802468e-01 1.14427857e-01 3.30665350e-01 2.27111608e-01 1.10313129e+00 3.53461653e-01 -1.10557830e+00 -6.42243505e-01 -7.02553868e-01 -5.25232553e-01 3.45463037e-01 5.94076775e-02 -1.90984577e-01 -8.09316576e-01 -3.89635891e-01 3.31876844e-01 -9.60358620e-01 -5.02370894e-01 -2.36209445e-02 -7.67859340e-01 -2.14661807e-01 8.86212647e-01 -7.06322730e-01 1.01070106e+00 -2.31365204e+00 3.60494256e-01 5.63636661e-01 4.25518095e-01 9.11526754e-02 5.90878874e-02 -4.58715111e-02 5.62471710e-02 5.33356890e-02 -4.13379222e-01 -4.81736541e-01 -2.38501742e-01 -4.47749011e-02 4.38304454e-01 9.31117773e-01 -2.59787105e-02 8.31858337e-01 -7.15389192e-01 -8.47807467e-01 2.71864593e-01 5.40220082e-01 -3.95406991e-01 -1.41405817e-02 -2.28660196e-01 7.39016175e-01 -3.08029801e-01 6.23452246e-01 7.34516799e-01 -4.21760082e-01 3.17275643e-01 -5.46313882e-01 -2.88299054e-01 -3.40908110e-01 -1.06179094e+00 1.85415792e+00 6.54320121e-02 3.65316093e-01 3.59556019e-01 -1.00898159e+00 6.67399585e-01 2.36558422e-01 7.33059824e-01 -1.69260904e-01 4.71084028e-01 2.57350802e-01 2.33704045e-01 -2.59675920e-01 2.05958635e-02 -1.25394955e-01 1.06054209e-01 3.99245799e-01 1.63836002e-01 2.10481174e-02 3.81294817e-01 1.54875308e-01 1.15628290e+00 7.08012842e-04 1.40333042e-01 -4.34121549e-01 4.74892467e-01 -4.92907763e-02 7.84748077e-01 4.88493800e-01 -3.97208571e-01 1.05302787e+00 6.51258349e-01 -4.49709743e-01 -1.02784705e+00 -8.07978690e-01 -2.23795772e-01 8.11776221e-01 -5.65468483e-02 -1.61928087e-02 -1.12100041e+00 -8.48540366e-01 -8.20052251e-02 1.75686508e-01 -9.21773136e-01 3.14223170e-01 -4.75880861e-01 -9.87130225e-01 4.98434782e-01 4.15398896e-01 3.63895923e-01 -9.60770428e-01 -5.98663211e-01 2.41157249e-01 -7.18104765e-02 -1.01804936e+00 -5.60703576e-01 5.08147001e-01 -6.10375226e-01 -8.85133326e-01 -8.93380940e-01 -9.10113752e-01 9.70350802e-01 -1.14826187e-01 5.11740863e-01 1.38848305e-01 -5.24815381e-01 -3.05041701e-01 1.48159996e-01 6.18644711e-03 -6.42850175e-02 3.45860630e-01 -2.78327346e-01 2.39970222e-01 3.79145183e-02 -6.02799535e-01 -1.04057455e+00 1.58014029e-01 -9.82748568e-01 7.05022067e-02 7.49322772e-01 5.52852154e-01 1.19108772e+00 1.91425502e-01 3.04323941e-01 -9.54181612e-01 3.81746851e-02 -6.20638371e-01 -7.44658649e-01 2.34553710e-01 7.99427330e-02 -5.23162214e-03 6.01219416e-01 -2.04507604e-01 -5.47449350e-01 8.02625656e-01 -7.73065835e-02 -2.61040956e-01 -2.63278216e-01 6.55628622e-01 -8.39896873e-02 -1.85306147e-01 3.10226440e-01 2.69399911e-01 3.74754034e-02 -3.04084808e-01 1.44104794e-01 5.57200372e-01 6.77452028e-01 -1.25529960e-01 5.46895802e-01 8.22047889e-01 2.50590593e-01 -6.80586934e-01 -5.72898746e-01 -6.99635386e-01 -1.14340568e+00 -2.75726736e-01 1.27876484e+00 -6.84883773e-01 -5.95997751e-01 5.81498504e-01 -9.33796108e-01 -4.67142224e-01 9.82256010e-02 4.00191188e-01 -3.01545501e-01 1.73092842e-01 -7.57505357e-01 -3.52205813e-01 -5.13172448e-01 -1.47515011e+00 1.11737943e+00 6.48417592e-01 -5.97552471e-02 -9.42414403e-01 2.41873726e-01 3.73024166e-01 1.25756368e-01 4.00042117e-01 1.05232370e+00 -8.93380165e-01 -5.58293939e-01 -1.73027784e-01 -4.76670414e-01 -2.57395431e-02 7.30467439e-02 1.38208061e-01 -7.20100284e-01 -2.17700869e-01 -4.89046514e-01 1.16495416e-02 1.04771793e+00 9.71050203e-01 1.27173030e+00 1.55830935e-01 -7.41059899e-01 8.98332357e-01 1.59521806e+00 6.50363490e-02 3.41183484e-01 2.24505574e-01 8.73713315e-01 4.39040452e-01 2.48932987e-01 3.46609563e-01 3.93036962e-01 6.74239278e-01 3.20039868e-01 -5.60900569e-01 -2.38678694e-01 1.87211543e-01 -2.85653383e-01 3.97157341e-01 2.42143750e-01 -3.46399844e-02 -1.21289670e+00 8.67399037e-01 -1.82091784e+00 -6.44354463e-01 -2.08456904e-01 1.80229735e+00 8.56423855e-01 -1.32895097e-01 1.83891192e-01 -2.81162083e-01 9.97561395e-01 -1.37472942e-01 -7.84151971e-01 1.35589570e-01 9.46748927e-02 2.58441386e-03 4.77953285e-01 2.57854789e-01 -1.38061583e+00 9.34835911e-01 5.32197475e+00 9.77260947e-01 -1.14558923e+00 2.78626923e-02 1.04668784e+00 -1.20488748e-01 1.08238034e-01 -1.24405168e-01 -8.69491816e-01 6.19996727e-01 4.82705414e-01 2.67954946e-01 2.44315833e-01 3.97249252e-01 4.61521596e-01 -3.24422419e-01 -9.34169114e-01 6.85453773e-01 -1.61099404e-01 -1.44861758e+00 -1.96520388e-01 2.17670456e-01 7.49846458e-01 3.39733183e-01 -4.47098203e-02 -1.14758149e-01 1.87397927e-01 -1.09804046e+00 4.33479190e-01 5.92439294e-01 8.16584170e-01 -9.37247694e-01 9.93376017e-01 3.78023893e-01 -1.22532666e+00 3.92195471e-02 -3.44051808e-01 4.99957860e-01 -2.37768553e-02 7.68322229e-01 -1.13227880e+00 2.76214480e-01 4.48434204e-01 4.35313791e-01 -6.33095920e-01 1.52764475e+00 1.74728394e-01 4.58679199e-01 -6.30156159e-01 2.23242208e-01 3.74931961e-01 -2.94253141e-01 3.05222601e-01 1.49213541e+00 2.92661995e-01 1.69544339e-01 3.25065076e-01 1.01192009e+00 -3.01339895e-01 5.82503155e-02 -1.91058107e-02 7.90284500e-02 5.01016676e-01 1.99665225e+00 -1.73631668e+00 -1.63142860e-01 -1.42677382e-01 7.91974366e-01 6.54168069e-01 2.47495309e-01 -7.17028797e-01 -2.62368530e-01 3.34158331e-01 2.07507148e-01 5.13577759e-01 -9.32248607e-02 -4.70860392e-01 -6.95182264e-01 -4.18099731e-01 -3.38932276e-01 5.30328095e-01 -4.02688861e-01 -1.08411229e+00 3.65265638e-01 -4.05715883e-01 -7.59408355e-01 2.74113238e-01 -1.66973501e-01 -6.97719336e-01 9.85271990e-01 -1.20826864e+00 -1.17822790e+00 -2.88073689e-01 3.37660491e-01 3.77499849e-01 1.01891540e-01 6.66930795e-01 2.30638579e-01 -9.03277874e-01 2.89241493e-01 2.49185428e-01 5.77342868e-01 4.09929663e-01 -1.39692616e+00 -5.19868918e-02 7.13540971e-01 -3.44086140e-02 6.23984337e-01 3.73226643e-01 -7.59167731e-01 -8.95564795e-01 -1.35281193e+00 5.92301369e-01 -3.30298431e-02 5.58356285e-01 -3.44729185e-01 -7.39760160e-01 5.62016010e-01 -3.45866010e-02 5.37427783e-01 7.56378710e-01 -3.86020064e-01 2.98087239e-01 -1.59567609e-01 -1.34756625e+00 5.23665786e-01 2.78515458e-01 -1.77313849e-01 -1.10209942e-01 2.86578625e-01 3.69091570e-01 -6.72751307e-01 -8.64903867e-01 2.53912985e-01 3.72623771e-01 -7.61731863e-01 7.65322626e-01 5.71792983e-02 3.00176680e-01 -7.90280640e-01 2.22890675e-01 -1.23779857e+00 -6.17796719e-01 3.26202326e-02 3.46142292e-01 1.03611827e+00 4.19883281e-01 -1.22339867e-01 1.01058221e+00 4.92624700e-01 -3.93349260e-01 -1.03509915e+00 -1.09259057e+00 -9.99416783e-02 5.71019091e-02 7.74616748e-02 2.74853408e-01 7.21043408e-01 -1.97436988e-01 2.91907102e-01 2.09441140e-01 3.05241108e-01 7.78915048e-01 1.06264576e-01 3.52901995e-01 -1.07427442e+00 -1.76807623e-02 -4.69610691e-01 -6.02175772e-01 -6.69259965e-01 1.02281675e-01 -1.29623330e+00 1.60803065e-01 -1.82585788e+00 6.11910760e-01 -3.81761879e-01 -5.84158957e-01 6.32408023e-01 -1.48807839e-01 6.50680482e-01 2.29252018e-02 2.39339784e-01 -7.68337309e-01 2.26531431e-01 1.14098084e+00 -1.89243302e-01 -2.20673919e-01 3.46113592e-02 -3.69821429e-01 7.49874651e-01 7.62093365e-01 -5.08529723e-01 -2.43414957e-02 -2.32378229e-01 -2.47418955e-01 1.13632664e-01 4.97643709e-01 -1.25064218e+00 6.65144384e-01 2.13527769e-01 9.49023545e-01 -1.05591571e+00 1.60602868e-01 -7.67063618e-01 1.31744921e-01 3.86477113e-01 -2.63930589e-01 -3.15573484e-01 1.00875787e-01 5.05018234e-01 3.07940952e-02 -6.37975186e-02 1.12322736e+00 -3.76401097e-01 -1.97746515e-01 5.33236563e-01 -3.78568083e-01 -3.00233603e-01 1.23407900e+00 -2.67980784e-01 -9.39597711e-02 1.55696467e-01 -9.10435200e-01 4.46337283e-01 3.98217201e-01 -3.19517940e-01 4.14267331e-01 -1.09186375e+00 -6.55750692e-01 -3.62493889e-03 -1.89143613e-01 4.85637277e-01 5.89037955e-01 1.25182247e+00 -7.09694207e-01 4.95226502e-01 -1.56677902e-01 -9.65979755e-01 -1.27563977e+00 2.11388484e-01 7.75839329e-01 -4.31880653e-01 -5.82591295e-01 1.07367051e+00 3.81463140e-01 -4.34461355e-01 1.10249557e-01 -5.03807545e-01 -5.41926861e-01 3.98027785e-02 3.29936653e-01 1.09000161e-01 1.27531990e-01 -8.40347767e-01 -5.77502251e-01 6.39198422e-01 -3.07798266e-01 1.14813475e-02 1.51224422e+00 -9.86823365e-02 -4.91832137e-01 3.51565599e-01 1.24088013e+00 -2.80748338e-01 -1.44468176e+00 -2.56424695e-01 -5.96407615e-02 -4.54282239e-02 3.94333154e-01 -1.00390279e+00 -1.43610764e+00 6.64109051e-01 7.05951452e-01 -1.52430385e-01 9.41488743e-01 1.17238343e-01 8.01791370e-01 -1.70797333e-01 -5.92223294e-02 -9.73140895e-01 -4.59678978e-01 3.46771836e-01 2.48370007e-01 -1.04477167e+00 -6.45573884e-02 -3.20540726e-01 -3.99051636e-01 1.09171700e+00 6.47626817e-01 5.69901392e-02 5.39385080e-01 3.83213967e-01 1.20279469e-01 -4.14897680e-01 -3.14740032e-01 -1.01313956e-01 2.85089046e-01 3.34145576e-01 4.36390698e-01 1.28609926e-01 -1.56702533e-01 6.06861472e-01 1.41990066e-01 -1.32275373e-01 2.52535403e-01 5.71685910e-01 -5.65009892e-01 -9.11161602e-01 -4.17199194e-01 5.27161658e-01 -6.87089741e-01 4.74310890e-02 -3.77689302e-01 4.47906017e-01 3.69236976e-01 4.87589866e-01 2.20469266e-01 -1.60049334e-01 -1.06276296e-01 -9.13866237e-02 2.58641064e-01 -6.93620086e-01 -7.45174646e-01 3.82388949e-01 -5.05632997e-01 -2.57745117e-01 -1.43241987e-01 -6.80578232e-01 -1.95590270e+00 9.29239169e-02 -4.61303852e-02 -1.76840881e-03 7.01963127e-01 1.05947375e+00 2.93366790e-01 5.53240955e-01 3.64997625e-01 -9.40512300e-01 -1.28759637e-01 -9.23321187e-01 -8.86180997e-01 2.24113777e-01 2.34735191e-01 -3.13628376e-01 -3.11873760e-02 -3.36260200e-02]
[14.90174674987793, -3.0397047996520996]
1573ab7d-1bb6-4fa0-bf3d-bd824eb2bd6e
c-textgen-conditional-text-generation-for
1909.03409
null
https://arxiv.org/abs/1909.03409v2
https://arxiv.org/pdf/1909.03409v2.pdf
Conditional Text Generation for Harmonious Human-Machine Interaction
In recent years, with the development of deep learning, text generation technology has undergone great changes and provided many kinds of services for human beings, such as restaurant reservation and daily communication. The automatically generated text is becoming more and more fluent so researchers begin to consider more anthropomorphic text generation technology, that is the conditional text generation, including emotional text generation, personalized text generation, and so on. Conditional Text Generation (CTG) has thus become a research hotspot. As a promising research field, we find that many efforts have been paid to exploring it. Therefore, we aim to give a comprehensive review of the new research trends of CTG. We first summary several key techniques and illustrate the technical evolution route in the field of neural text generation, based on the concept model of CTG. We further make an investigation of existing CTG fields and propose several general learning models for CTG. Finally, we discuss the open issues and promising research directions of CTG.
['Wei Wu', 'Zhiwen Yu', 'Shaoyang Hao', 'Yueqi Sun', 'Bin Guo', 'Hao Wang', 'Yasan Ding']
2019-09-08
null
null
null
null
['conditional-text-generation']
['natural-language-processing']
[ 6.00161143e-02 3.84864390e-01 -1.16670832e-01 -2.59091765e-01 -3.18618178e-01 7.68687055e-02 7.15284526e-01 -1.10021114e-01 6.13541491e-02 1.06137550e+00 5.06178737e-01 -3.85317281e-02 3.57760310e-01 -1.11508930e+00 -2.11884633e-01 -6.82393551e-01 1.44434988e-01 3.28271687e-01 -3.63744050e-01 -6.02572680e-01 2.92499900e-01 3.42004076e-02 -1.50664198e+00 4.22885381e-02 1.21021211e+00 8.80667508e-01 2.48671085e-01 5.00841141e-01 -4.99746948e-01 4.51893449e-01 -9.28733945e-01 -6.12816691e-01 -3.64163846e-01 -8.12732995e-01 -9.42667365e-01 -8.45958143e-02 -2.26590291e-01 -2.84007460e-01 -2.53931940e-01 8.57122302e-01 1.12700009e+00 5.05703032e-01 6.91689491e-01 -1.50555229e+00 -1.01629055e+00 1.11143422e+00 -4.03145880e-01 -2.69618839e-01 4.40828204e-01 -1.60332009e-01 9.47807789e-01 -7.02477992e-01 6.16794229e-01 1.16720426e+00 4.19144571e-01 1.07959747e+00 -5.06148219e-01 -6.29299700e-01 3.47595304e-01 5.79793528e-02 -9.40732479e-01 -1.59634635e-01 1.00872254e+00 -5.74773960e-02 1.08869255e+00 2.83428431e-01 9.48692381e-01 1.47004378e+00 5.30014634e-01 1.29493034e+00 6.97126985e-01 -5.63312650e-01 1.80928260e-01 1.70458540e-01 -2.24084258e-01 5.06883621e-01 2.67646819e-01 -3.12581696e-02 -6.02074742e-01 2.75439192e-02 5.98100066e-01 -2.99867243e-01 -1.70269415e-01 3.25317413e-01 -1.22081637e+00 1.13112998e+00 1.93959460e-01 6.31596804e-01 -2.07665116e-01 2.94006586e-01 4.74091977e-01 -4.35470007e-02 9.01732087e-01 5.78393579e-01 -1.55264601e-01 -4.36756819e-01 -8.34427476e-01 6.00089788e-01 7.88294554e-01 1.17162919e+00 4.39903915e-01 5.59939146e-01 -3.52135599e-01 9.82449710e-01 1.24328300e-01 3.97437871e-01 1.16599631e+00 -3.72767866e-01 3.79211038e-01 4.15043056e-01 -1.47904530e-01 -9.49690282e-01 -3.88258606e-01 -1.87759832e-01 -1.27099752e+00 -3.09198290e-01 -3.42860729e-01 -9.57063794e-01 -6.82766199e-01 1.41936886e+00 1.55197173e-01 -1.73114315e-01 6.91492558e-02 4.39633429e-01 1.24603784e+00 1.08062220e+00 1.08735122e-01 -2.39882991e-01 9.46760058e-01 -9.11970735e-01 -1.03414345e+00 -1.45413846e-01 5.12841225e-01 -8.23107898e-01 9.73977983e-01 2.29977295e-01 -8.98178041e-01 -4.88365144e-01 -8.94403934e-01 -1.33872658e-01 -7.09825873e-01 5.72033077e-02 1.17223907e+00 7.16255605e-01 -1.02209318e+00 6.87817574e-01 -6.52721763e-01 -6.90813005e-01 2.10640147e-01 1.39329314e-01 1.44167319e-01 3.58317077e-01 -1.71528709e+00 6.95310771e-01 6.25122309e-01 2.10571021e-01 -3.49601835e-01 -1.45849213e-01 -8.55447292e-01 -3.71745862e-02 2.14232281e-02 -8.93021703e-01 1.37267947e+00 -7.23722517e-01 -1.96048856e+00 5.57808340e-01 -5.85959591e-02 -2.15472013e-01 3.62365752e-01 -4.67262775e-01 -7.71850526e-01 -2.16428876e-01 -4.15788107e-02 8.72062087e-01 7.52487898e-01 -1.07638383e+00 -4.94498074e-01 8.43034908e-02 -1.92814246e-01 4.31771070e-01 -5.77734411e-01 -1.11622274e-01 3.80543247e-02 -1.06253362e+00 -3.23704392e-01 -7.35266149e-01 -4.28593785e-01 -4.29101110e-01 -7.05076635e-01 -7.20272005e-01 9.69106734e-01 -4.21645075e-01 1.43770361e+00 -1.84940386e+00 -6.06826833e-03 -2.47472972e-01 1.34380296e-01 1.56908065e-01 2.00813949e-01 8.45063627e-01 -1.39039099e-01 4.90179926e-01 -9.02568027e-02 -3.61404955e-01 1.82237610e-01 -6.16606586e-02 -6.12024605e-01 -3.35604757e-01 -1.05680563e-02 1.28782868e+00 -1.04869616e+00 -5.94603419e-01 2.05292761e-01 4.02879238e-01 -3.58365685e-01 2.06688434e-01 -6.24145567e-01 1.90334454e-01 -9.55548823e-01 4.82842922e-01 3.02531034e-01 -2.22318962e-01 -1.56282544e-01 2.16429383e-01 -2.79648632e-01 1.56665877e-01 -5.84597170e-01 1.72006536e+00 -4.45171326e-01 6.91725969e-01 -4.57458884e-01 -7.70469248e-01 1.32224047e+00 5.37067771e-01 4.94997829e-01 -4.62732762e-01 5.01337707e-01 2.71036834e-01 -2.82040656e-01 -6.72617018e-01 1.22697067e+00 -4.40233916e-01 -3.20465297e-01 8.17126572e-01 -1.14951603e-01 -4.44534063e-01 9.42329541e-02 3.72331351e-01 3.41265917e-01 3.05916429e-01 3.65789205e-01 9.30291116e-02 1.62703022e-01 -2.08633348e-01 5.06470911e-02 2.86236525e-01 -3.17899138e-02 5.05072474e-01 2.40459070e-01 -2.22619727e-01 -6.51289105e-01 -6.70217276e-01 2.36902609e-01 9.07621443e-01 1.72327965e-01 -5.60218215e-01 -8.79701436e-01 -5.56348324e-01 -4.41181272e-01 8.83354843e-01 -5.97421408e-01 -4.74015832e-01 -4.89011973e-01 -1.09209502e+00 4.85190570e-01 5.60602188e-01 9.29195523e-01 -1.72080076e+00 -3.13450664e-01 3.81843388e-01 -7.24851251e-01 -7.12662101e-01 -3.93619835e-01 -2.50126064e-01 -1.10508072e+00 -2.54552513e-01 -1.15024626e+00 -1.06130087e+00 4.41760302e-01 1.18317015e-01 1.08342683e+00 7.53599778e-02 -1.28707409e-01 1.71757922e-01 -8.96146059e-01 -7.40609169e-01 -3.73104334e-01 5.45265615e-01 -4.90041040e-02 -2.04551876e-01 1.40381649e-01 -6.11372650e-01 -5.42226434e-01 -2.10987285e-01 -8.78097296e-01 4.41015691e-01 6.83428943e-01 7.48463690e-01 2.80116022e-01 1.90276116e-01 1.18233538e+00 -8.47776592e-01 1.34608233e+00 -4.87980902e-01 -7.36420881e-03 1.39724478e-01 -7.18629658e-01 2.40808260e-02 6.90091670e-01 -3.50335747e-01 -1.64617360e+00 -5.20206630e-01 -5.88868260e-01 2.70438373e-01 -5.19212633e-02 7.18068719e-01 -2.07774252e-01 2.94907242e-01 5.98636746e-01 3.70222986e-01 -4.36042279e-01 -3.56154621e-01 6.19888008e-01 1.07489181e+00 1.37304336e-01 -5.88333964e-01 5.15545547e-01 1.13624811e-01 -3.83231848e-01 -9.19431269e-01 -5.73229611e-01 1.77655309e-01 -1.39106005e-01 -5.54667890e-01 8.98196459e-01 -3.51256460e-01 -5.13818324e-01 7.81951070e-01 -1.24344528e+00 -4.63837057e-01 -2.84543246e-01 2.91916192e-01 -5.20612299e-01 4.92618531e-01 -6.43754363e-01 -7.96927333e-01 -9.86753285e-01 -6.92082107e-01 1.18987775e+00 7.39581168e-01 -3.15020949e-01 -1.26167285e+00 1.02686770e-01 1.42870009e-01 6.23865366e-01 3.80789191e-01 1.02736759e+00 -3.89770895e-01 -3.63841087e-01 -2.37671018e-01 7.29292855e-02 1.67626470e-01 2.18808129e-01 -9.27610416e-03 -9.62004900e-01 6.17753714e-02 -3.48304659e-02 -2.89145470e-01 6.45392239e-01 3.71638656e-01 1.39589167e+00 -3.32504153e-01 -6.48587286e-01 4.14084584e-01 1.00803173e+00 5.90478063e-01 7.88095057e-01 4.61231321e-02 5.84129453e-01 4.90619361e-01 5.97433507e-01 5.96901357e-01 4.26823884e-01 2.36617804e-01 1.94470406e-01 -1.50350884e-01 -1.21732650e-03 -7.06031084e-01 3.66273940e-01 1.18625605e+00 -3.50890219e-01 -7.91316450e-01 -4.53565031e-01 3.60973775e-01 -1.74422371e+00 -1.11907017e+00 -9.08718556e-02 1.78004515e+00 9.56792355e-01 -1.26489684e-01 -1.72320023e-01 3.15188356e-02 9.51621711e-01 4.04378116e-01 -5.82829356e-01 -4.80028838e-01 -6.43774942e-02 3.34083825e-01 -2.25717530e-01 1.26876816e-01 -7.90055156e-01 1.16086090e+00 7.18282032e+00 9.42080259e-01 -1.34233022e+00 -2.46866688e-01 9.49665844e-01 1.65526137e-01 -4.67885673e-01 -4.11338985e-01 -7.40842402e-01 6.82995737e-01 6.73411787e-01 -8.16701651e-01 2.35381365e-01 1.02676916e+00 3.32014084e-01 2.39586085e-02 -7.70904601e-01 1.08667314e+00 3.08313787e-01 -1.28941822e+00 4.40093130e-01 -1.44154951e-01 1.13139224e+00 -3.29395890e-01 1.02399848e-01 5.18482566e-01 2.72856086e-01 -1.03507626e+00 4.00040716e-01 3.69136959e-01 9.52925801e-01 -8.55457664e-01 6.52289450e-01 2.69109875e-01 -1.15202618e+00 3.57752413e-01 -3.67654622e-01 -1.66897386e-01 4.87344414e-01 8.41575444e-01 -6.92443550e-01 8.51970255e-01 4.66002107e-01 8.31432819e-01 -3.14543694e-01 9.98026729e-01 -3.68679732e-01 5.51522493e-01 1.69168934e-01 -7.98689663e-01 -2.66918745e-02 -2.00377762e-01 3.87560397e-01 1.05617762e+00 7.47421503e-01 1.30300865e-01 -3.45473103e-02 8.49863052e-01 -2.31024414e-01 3.58903050e-01 -4.53074306e-01 -5.06065190e-01 1.17061980e-01 1.34484148e+00 -8.89699101e-01 -4.40217763e-01 -1.28302453e-02 1.27946568e+00 -4.13226970e-02 4.75488603e-01 -8.58135104e-01 -1.03140473e+00 2.82133430e-01 -2.71367043e-01 -2.93740839e-01 -2.24849746e-01 -3.19288373e-01 -1.32839155e+00 -1.66223511e-01 -6.33585334e-01 1.46604152e-02 -1.15080047e+00 -1.25613058e+00 8.00228775e-01 2.73536108e-02 -1.09905326e+00 -5.58431268e-01 -3.72525245e-01 -9.57953811e-01 8.60621393e-01 -1.15604377e+00 -9.24474239e-01 -3.85208011e-01 2.81454653e-01 8.16341877e-01 -8.93824995e-02 8.29561949e-01 2.24776492e-02 -4.39974040e-01 5.13899446e-01 7.03381673e-02 8.86947140e-02 6.59985840e-01 -1.02115238e+00 9.53972042e-01 6.31338418e-01 -5.97675443e-02 6.43736720e-01 5.76323926e-01 -8.47846627e-01 -9.81783628e-01 -1.12874031e+00 1.28682828e+00 1.27249202e-02 3.28200758e-01 -5.41119635e-01 -4.94784892e-01 6.94979429e-01 6.37683511e-01 -9.34415579e-01 6.30580544e-01 -3.89319621e-02 3.97285759e-01 9.48796347e-02 -9.95839357e-01 1.01083696e+00 1.11715484e+00 -2.58297357e-03 -4.76471752e-01 4.97481138e-01 1.07290494e+00 -3.86199176e-01 -4.03602958e-01 6.93619996e-02 3.87550920e-01 -8.08041632e-01 3.07807654e-01 -2.04274788e-01 5.89946747e-01 3.36688384e-02 4.77608204e-01 -1.78085387e+00 -7.54526407e-02 -1.10700452e+00 -6.38369769e-02 1.47869408e+00 3.30909461e-01 -7.88348973e-01 8.53311598e-01 4.84592259e-01 -4.62998360e-01 -1.04128683e+00 -4.53171819e-01 -4.32680130e-01 2.73778528e-01 -4.65902805e-01 8.24369013e-01 8.79018307e-01 6.00089610e-01 7.30831563e-01 -7.35326648e-01 -8.49149346e-01 -2.66285911e-02 2.34417081e-01 5.82507432e-01 -1.22449911e+00 -1.63280398e-01 -6.58823252e-01 2.69117113e-02 -1.52515793e+00 1.00513995e-02 -1.00165045e+00 2.08982617e-01 -2.01541877e+00 7.97866806e-02 -2.29459301e-01 3.64959598e-01 3.68418664e-01 -3.59847933e-01 3.07317860e-02 5.06227510e-03 -2.25102782e-01 -2.90619135e-01 1.15645468e+00 1.72480917e+00 4.08106344e-03 -3.46013337e-01 1.14220381e-01 -1.15836656e+00 4.56820309e-01 1.16430032e+00 6.28341362e-02 -6.21357977e-01 -2.34137595e-01 7.07261264e-01 -1.93962194e-02 -2.23846778e-01 -8.08085084e-01 -8.75873957e-03 -2.12883651e-01 4.53510106e-01 -7.24854052e-01 1.59165934e-01 -2.83912629e-01 -1.11859508e-01 1.27652138e-01 -4.60297644e-01 1.93117093e-02 -3.71534191e-02 3.72690648e-01 -1.71584383e-01 -2.67929196e-01 5.14931023e-01 -2.04978362e-01 -5.91429412e-01 5.37084222e-01 -7.54421353e-01 1.99275166e-02 9.80484247e-01 -1.10015906e-01 -2.75611132e-01 -8.69731605e-01 -5.43421865e-01 3.32829714e-01 -4.18585166e-02 7.13862300e-01 8.45905066e-01 -1.37882137e+00 -5.48177004e-01 -7.10083544e-02 -1.52606457e-01 1.37697412e-02 3.23584706e-01 2.22723678e-01 -2.51861751e-01 4.13774043e-01 -2.82056239e-02 9.71237794e-02 -9.01061535e-01 3.50247741e-01 2.96815634e-01 -3.92725527e-01 -6.37885094e-01 7.39072442e-01 6.93327487e-02 -2.22358584e-01 -1.33989574e-02 -3.72371614e-01 -4.38519359e-01 -5.37838563e-02 4.79999125e-01 5.25276423e-01 -1.34313434e-01 -3.12968194e-01 2.34779805e-01 2.66065568e-01 -7.26826042e-02 -2.04056993e-01 1.09322178e+00 -6.83394298e-02 -1.23229556e-01 5.06292343e-01 8.16185653e-01 -2.41135165e-01 -6.04721129e-01 3.40273857e-01 -2.65143037e-01 8.33936706e-02 -1.55616254e-01 -1.07300377e+00 -1.21602476e+00 1.05345428e+00 1.27774134e-01 5.63256145e-01 1.11178231e+00 -3.86314876e-02 1.43909776e+00 3.36803734e-01 5.16682208e-01 -1.43799233e+00 3.96812141e-01 7.21570730e-01 1.06870413e+00 -8.30965638e-01 -7.01926947e-02 -2.56108820e-01 -7.35062122e-01 1.10798359e+00 7.45582879e-01 1.20270491e-01 5.98596394e-01 4.61151404e-03 1.15454691e-02 -7.79961571e-02 -6.42362475e-01 8.72297771e-03 8.30191895e-02 8.65192533e-01 1.02485633e+00 1.85934365e-01 -7.31156409e-01 7.99262404e-01 -9.66811299e-01 1.66040987e-01 5.07707000e-01 7.12033510e-01 -5.49708247e-01 -1.40388787e+00 -1.35528654e-01 7.52338052e-01 -2.88408667e-01 -2.84490258e-01 -6.52714372e-01 6.65497124e-01 4.38990816e-02 1.14736962e+00 -1.81215361e-01 -5.83656669e-01 1.91979315e-02 3.16760391e-01 2.74023056e-01 -6.86224699e-01 -4.52693194e-01 -1.22337058e-01 1.32040083e-01 -7.12197199e-02 -4.14403856e-01 -3.98033947e-01 -1.56856883e+00 -3.96331310e-01 -5.00555336e-01 4.26994443e-01 7.38965988e-01 9.84043479e-01 3.79570544e-01 5.32687426e-01 5.35499215e-01 -1.07995141e+00 6.91042393e-02 -1.21201503e+00 -7.01849282e-01 -2.96883415e-02 -2.19251290e-01 -4.72799748e-01 -1.77132577e-01 2.27278367e-01]
[11.992610931396484, 9.118602752685547]
8c490cba-3f64-4c1b-b9e3-7f61973974ca
blind-image-deblurring-using-dark-channel
null
null
http://openaccess.thecvf.com/content_cvpr_2016/html/Pan_Blind_Image_Deblurring_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Pan_Blind_Image_Deblurring_CVPR_2016_paper.pdf
Blind Image Deblurring Using Dark Channel Prior
We present a simple and effective blind image deblurring method based on the dark channel prior. Our work is inspired by the interesting observation that the dark channel of blurred images is less sparse. While most image patches in the clean image contain some dark pixels, these pixels are not dark when averaged with neighboring high-intensity pixels during the blur process.Our analysis shows that this change in the sparsity of the dark channel is an inherent property of the blur process, both theoretically and empirically. This change in the sparsity of the dark channel is an inherent property of the blur process, which we both prove mathematically and validate using training data. Therefore, enforcing the sparsity of the dark channel helps blind deblurring on various scenarios, including natural, face, text, and low-illumination images. However, sparsity of the dark channel introduces a non-convex non-linear optimization problem. We introduce a linear approximation of the min operator to compute the dark channel. Our look-up-table-based method converges fast in practice and can be directly extended to non-uniform deblurring. Extensive experiments show that our method achieves state-of-the-art results on deblurring natural images and compares favorably methods that are well-engineered for specific scenarios.
['Ming-Hsuan Yang', 'Deqing Sun', 'Jinshan Pan', 'Hanspeter Pfister']
2016-06-01
null
null
null
cvpr-2016-6
['blind-image-deblurring']
['computer-vision']
[ 3.31706434e-01 -5.66740930e-01 8.33974257e-02 -7.75236934e-02 -2.41729572e-01 -5.73732257e-01 4.42016542e-01 -8.45940351e-01 9.12799849e-04 7.60928750e-01 6.27025366e-01 -9.47117358e-02 -3.81690543e-03 -1.88466892e-01 -6.76442444e-01 -1.04622495e+00 -1.99971534e-02 -3.61886680e-01 -1.16704263e-01 2.07157061e-02 3.19270879e-01 3.36800039e-01 -1.12518346e+00 1.23039275e-01 9.53166544e-01 7.12998629e-01 3.38449836e-01 8.58752847e-01 3.02915752e-01 8.90438199e-01 -4.17275667e-01 -1.26791656e-01 6.63085699e-01 -6.82976723e-01 -6.73818409e-01 4.37231064e-01 7.65909731e-01 -7.45750010e-01 -8.31358135e-01 1.61951900e+00 4.88667756e-01 8.49629492e-02 4.67444658e-01 -7.38709927e-01 -1.27489901e+00 2.02393234e-01 -9.69521224e-01 4.84055340e-01 3.78391594e-01 3.20282638e-01 5.77124774e-01 -1.12216222e+00 5.12833178e-01 1.09493911e+00 8.08462024e-01 4.12956744e-01 -1.17729795e+00 -4.40871179e-01 -3.52842629e-01 4.16857034e-01 -1.58051038e+00 -7.36078501e-01 5.63965619e-01 -3.78897250e-01 2.94899583e-01 4.82562035e-01 3.47719401e-01 8.40696812e-01 2.33012021e-01 5.30646503e-01 1.64825523e+00 -5.45191348e-01 2.03193933e-01 -1.43868551e-01 1.47322223e-01 4.72197473e-01 5.97735405e-01 4.94919091e-01 -6.91644132e-01 -1.53181210e-01 9.66425300e-01 1.23407997e-01 -1.11363769e+00 -1.78196907e-01 -1.27567649e+00 4.25735593e-01 3.89732808e-01 3.21845531e-01 -2.66181111e-01 3.39646399e-01 -1.90815106e-01 3.36504310e-01 5.40222764e-01 3.12194049e-01 -1.25188932e-01 -1.63089961e-01 -1.33127844e+00 -1.03518926e-01 9.51482594e-01 6.53154075e-01 9.48873818e-01 4.29694280e-02 -3.64875317e-01 8.48135114e-01 1.41370460e-01 8.98686588e-01 4.26923394e-01 -1.27986217e+00 2.10836809e-02 -3.03079069e-01 5.12798488e-01 -1.01897693e+00 2.95969993e-01 -4.63474482e-01 -1.27601016e+00 2.27218628e-01 6.12663984e-01 -3.35710973e-01 -9.23051596e-01 1.43142056e+00 -5.94895817e-02 6.56154633e-01 -5.53248683e-03 1.38309109e+00 3.42913419e-01 6.03895545e-01 -6.87775016e-01 -5.68279922e-01 1.26321900e+00 -9.22078311e-01 -1.23641551e+00 -3.78511637e-01 -2.06199914e-01 -1.05120945e+00 6.80015504e-01 3.18921357e-01 -1.24658871e+00 -4.11049247e-01 -9.81424987e-01 -4.71089967e-02 1.71688542e-01 6.88734129e-02 5.06912589e-01 9.22780335e-01 -1.26949155e+00 4.29163218e-01 -4.70616281e-01 -2.99043387e-01 3.33878756e-01 -2.87057180e-03 -2.08038807e-01 -6.50358975e-01 -1.10822499e+00 9.93442714e-01 -1.83777273e-01 3.44400346e-01 -8.28197300e-01 -6.07452452e-01 -7.30377614e-01 1.39412731e-01 5.92932068e-02 -7.23205328e-01 9.99115705e-01 -1.06348825e+00 -1.49033844e+00 6.85851514e-01 -7.56069541e-01 -4.07523215e-01 5.29639184e-01 -4.52028036e-01 -2.88210779e-01 3.62637907e-01 -3.59450355e-02 1.96897134e-01 1.59991193e+00 -1.54580569e+00 -1.31364122e-01 -1.54042736e-01 -2.62223661e-01 1.62493125e-01 -2.05624327e-01 -8.38965923e-03 -5.19395351e-01 -1.03663230e+00 1.25907063e-01 -8.70076120e-01 -6.43013790e-02 2.01675836e-02 -2.74869591e-01 6.94696128e-01 9.70247149e-01 -9.39290822e-01 1.38996637e+00 -2.45865512e+00 -6.72828928e-02 3.87520110e-03 3.17281276e-01 1.41759083e-01 -2.77883802e-02 2.97459722e-01 -2.29167044e-01 -2.72593528e-01 -5.53680718e-01 -8.83481950e-02 -3.62054408e-01 4.68467586e-02 -5.83502948e-01 1.14405119e+00 -1.70846522e-01 8.17102373e-01 -1.01798630e+00 -1.19344272e-01 2.60412633e-01 5.29349625e-01 -4.57005292e-01 2.86379039e-01 3.71088654e-01 5.57133257e-01 1.53223053e-01 5.27524590e-01 1.43912101e+00 -5.08814275e-01 -8.67591202e-02 -4.15204376e-01 -2.54504621e-01 -3.29724818e-01 -1.12803972e+00 1.57720971e+00 -3.22161883e-01 1.16672719e+00 4.98616308e-01 -5.28194368e-01 3.51449549e-01 2.50595987e-01 1.84668988e-01 -3.35261524e-01 -5.40780183e-03 3.09435517e-01 -1.61686391e-01 -6.62442088e-01 6.71445906e-01 -1.32208541e-01 5.49543142e-01 5.20630240e-01 -2.56829917e-01 -4.70408827e-01 4.19726968e-03 1.70624614e-01 1.07484448e+00 -3.55841279e-01 2.24209949e-01 -4.79642004e-01 4.53738809e-01 -4.06370163e-01 2.05425188e-01 1.00953364e+00 -2.66615421e-01 1.08625770e+00 6.62165657e-02 -1.07586823e-01 -9.52147722e-01 -8.25165570e-01 -3.69713426e-01 5.25575817e-01 7.54659593e-01 -2.60442961e-02 -1.13983822e+00 -1.88854694e-01 -1.94156498e-01 3.86320025e-01 -6.47248447e-01 7.86697343e-02 -4.45350051e-01 -1.07870805e+00 2.91242599e-01 2.00333241e-02 1.01240063e+00 -3.14455420e-01 -3.63672882e-01 -2.05301568e-01 -3.95110488e-01 -1.39307225e+00 -1.07876801e+00 -3.25177126e-02 -6.67075455e-01 -1.02975106e+00 -1.38579428e+00 -7.27343082e-01 1.03272128e+00 1.01322806e+00 7.63066709e-01 1.09995432e-01 -1.51271880e-01 5.06274700e-01 -2.00948998e-01 1.97430223e-01 -3.94732565e-01 -6.57220244e-01 -8.32544565e-02 4.48933929e-01 1.15858331e-01 -4.55160111e-01 -9.70316947e-01 4.01764631e-01 -1.01932168e+00 2.03391146e-02 5.87833822e-01 1.19044268e+00 1.12536319e-01 3.08594346e-01 -1.30350724e-01 -4.46805865e-01 7.05885530e-01 -3.31076413e-01 -5.11967599e-01 1.08718097e-01 -6.07587159e-01 1.97884351e-01 2.94191569e-01 -5.60255527e-01 -1.32608795e+00 5.94214126e-02 4.26983327e-01 -5.47389209e-01 5.20497188e-02 1.38170466e-01 8.28905031e-02 -4.98750329e-01 7.15813339e-01 3.54245335e-01 8.79070386e-02 -5.84292650e-01 5.46742499e-01 9.29807425e-01 9.21309471e-01 -1.48801282e-01 1.26815963e+00 8.66953075e-01 3.79169174e-02 -1.17388320e+00 -6.74980223e-01 -6.35901988e-01 -2.61016309e-01 -6.90612346e-02 5.97700715e-01 -1.07798147e+00 -5.33959806e-01 1.06617081e+00 -1.25447094e+00 -3.64546150e-01 -2.15200886e-01 4.40592736e-01 -2.61278778e-01 1.04508269e+00 -9.00712371e-01 -6.43945634e-01 -2.19951302e-01 -1.11934376e+00 8.80194187e-01 2.57979035e-01 5.47490157e-02 -9.95785713e-01 8.01417455e-02 2.50267982e-01 8.70417774e-01 -3.30040097e-01 3.65022928e-01 2.95454025e-01 -7.55178869e-01 -1.51401488e-02 -7.52278030e-01 6.90732241e-01 5.08416414e-01 -4.62352693e-01 -1.25062001e+00 -5.18796146e-01 5.44021964e-01 2.47588784e-01 1.04301667e+00 7.61100531e-01 1.06955802e+00 -6.12807572e-01 -1.02566756e-01 8.95503700e-01 1.50823295e+00 -4.09086615e-01 9.94376838e-01 6.22288249e-02 6.69227421e-01 2.07692370e-01 1.11553334e-01 4.27919805e-01 -9.26729590e-02 7.42887259e-01 3.47597361e-01 -3.97322237e-01 -7.58832216e-01 2.66349077e-01 5.68804979e-01 5.43915987e-01 -1.17307557e-02 -2.34823987e-01 -5.17232895e-01 6.23962045e-01 -1.51819468e+00 -1.17365980e+00 -4.11101669e-01 2.26224327e+00 1.31932604e+00 -5.63621998e-01 -5.59170246e-01 -8.69788378e-02 9.85044241e-01 4.27859306e-01 -3.72943789e-01 -4.33329167e-03 -6.24218702e-01 1.65765554e-01 1.00035048e+00 8.67803633e-01 -9.90868509e-01 7.38713443e-01 7.37053776e+00 7.66378701e-01 -1.25926960e+00 2.53714561e-01 5.64310312e-01 1.11764319e-01 -1.07280254e-01 1.29373357e-01 -3.02225709e-01 8.99893820e-01 5.29213667e-01 -2.18814030e-01 1.14120579e+00 3.63386720e-01 5.66547036e-01 -4.79322046e-01 -8.47620606e-01 1.49875700e+00 2.95805216e-01 -1.22236407e+00 -2.92517781e-01 1.37413114e-01 1.27423191e+00 8.81060958e-03 3.91868532e-01 -7.04239607e-01 3.12541932e-01 -1.00825369e+00 6.80498004e-01 5.78507066e-01 1.00700116e+00 6.78187236e-02 6.74645722e-01 -3.96270230e-02 -5.56903780e-01 -4.06759456e-02 -1.92931756e-01 -7.27087334e-02 2.10432708e-01 1.27288020e+00 -2.94045806e-01 3.02363545e-01 9.09119964e-01 9.92245197e-01 -4.62187767e-01 1.26940978e+00 -3.29084575e-01 6.72784865e-01 7.04321712e-02 6.57233417e-01 -1.01279326e-01 -2.73862720e-01 8.36277306e-01 1.41526687e+00 5.17621219e-01 2.22347096e-01 -4.70883250e-01 9.45410550e-01 -1.67152375e-01 -5.45098007e-01 -3.22300464e-01 2.73989350e-01 2.69673854e-01 8.95086586e-01 -2.43230209e-01 -3.51432472e-01 -4.51856047e-01 1.60235846e+00 -5.90696812e-01 9.95158494e-01 -5.96762598e-01 -2.30838761e-01 9.43465233e-01 -1.16533618e-02 5.56438446e-01 -3.30054820e-01 -5.54820597e-01 -1.60748100e+00 -8.81743729e-02 -1.12680352e+00 -2.23234639e-01 -1.09647012e+00 -1.47996128e+00 2.58991569e-01 -2.51418799e-01 -1.40309095e+00 3.23771685e-03 -3.90720099e-01 -5.05694032e-01 1.23539352e+00 -2.01359463e+00 -6.29566848e-01 -7.99198806e-01 7.00576067e-01 5.06344080e-01 1.74640656e-01 3.58182043e-01 2.01500848e-01 -4.45402503e-01 2.32802644e-01 6.64875805e-01 8.90279710e-02 1.09307218e+00 -1.07097733e+00 3.14045221e-01 1.53358793e+00 -2.00410143e-01 8.90201092e-01 1.06459737e+00 -5.64851403e-01 -1.69282162e+00 -7.59770036e-01 4.13047403e-01 -1.27956361e-01 6.47347152e-01 7.71025419e-02 -9.64432478e-01 3.30020875e-01 7.88634896e-01 3.99789691e-01 2.88777500e-01 -5.79751790e-01 -4.06432629e-01 2.12613363e-02 -1.10066926e+00 3.11674953e-01 7.29156494e-01 -6.70314491e-01 -4.89083767e-01 2.32642516e-01 3.06230754e-01 -5.34130037e-01 -2.79119611e-01 1.11342572e-01 5.24051309e-01 -1.16127431e+00 1.08536565e+00 1.59813762e-01 4.67939377e-01 -6.29027605e-01 -1.70641437e-01 -1.56519592e+00 -4.87465203e-01 -1.26199889e+00 -5.46013772e-01 7.21804857e-01 -1.24917716e-01 -6.96112931e-01 4.64948833e-01 4.45520192e-01 1.35654017e-01 -3.46131064e-02 -6.35872483e-01 -8.50591362e-01 -2.30145678e-01 5.13686500e-02 2.56289542e-01 1.08055198e+00 -1.57126814e-01 -2.93948837e-02 -9.14745331e-01 5.20453990e-01 1.05247450e+00 9.93508473e-02 4.72685844e-01 -7.69386530e-01 -3.51522654e-01 -1.67948052e-01 -1.53531834e-01 -1.87215984e+00 -6.33913651e-02 -3.65255386e-01 3.40302080e-01 -1.26960456e+00 4.59714890e-01 -1.75850138e-01 8.78915191e-02 9.58894938e-02 -4.52017605e-01 5.07875502e-01 -7.08312392e-02 7.13869631e-01 -1.06605493e-01 2.94597000e-01 1.47900367e+00 -3.52173179e-01 -3.96752283e-02 -2.35045493e-01 -6.77511752e-01 5.56541979e-01 3.95655513e-01 -2.60399491e-01 -1.33890966e-02 -8.02627742e-01 1.04566716e-01 -2.01037660e-01 5.50781548e-01 -7.86388278e-01 3.72012019e-01 -1.54785097e-01 3.80176783e-01 -1.27441436e-01 3.43484521e-01 -9.04200375e-01 1.32430851e-01 4.32525039e-01 -4.49391939e-02 -3.86544049e-01 -1.65697690e-02 6.98251903e-01 -3.47098261e-01 -2.68638670e-01 1.09169793e+00 -1.45960048e-01 -6.24855459e-01 2.49132551e-02 -3.85601521e-01 5.92116127e-03 6.18013859e-01 -4.69481021e-01 -7.92071462e-01 -8.35075796e-01 -2.62130529e-01 -3.89373988e-01 8.75802100e-01 3.22860591e-02 5.24719834e-01 -9.22919691e-01 -7.91418493e-01 4.02904123e-01 -3.91476244e-01 -4.51036841e-01 3.61661434e-01 1.17552269e+00 -7.07685471e-01 1.16357155e-01 -7.24285096e-02 -4.63606954e-01 -1.12118471e+00 5.59382975e-01 4.27789897e-01 1.83013573e-01 -4.56166208e-01 7.87594676e-01 3.94518584e-01 6.50556326e-01 1.21586025e-01 -2.39617184e-01 2.28068247e-01 -3.00715357e-01 9.87927079e-01 6.92065358e-01 -3.59657407e-02 -7.36446202e-01 -1.71469018e-01 7.53387272e-01 2.48858899e-01 -1.47094727e-01 1.10163021e+00 -7.70451128e-01 -7.16206908e-01 -1.97467104e-01 1.24875379e+00 5.95868826e-01 -1.47118449e+00 -2.62929708e-01 -4.17142659e-01 -1.20399487e+00 5.01444697e-01 -6.71955705e-01 -1.32076514e+00 7.00839043e-01 8.51272225e-01 3.09339255e-01 1.46517253e+00 -1.72651395e-01 8.62019360e-01 7.42984116e-02 1.17651209e-01 -6.41065240e-01 1.90128610e-01 4.49259967e-01 9.64048624e-01 -1.10246456e+00 2.36781210e-01 -5.83319008e-01 -3.49528432e-01 9.91184533e-01 -5.22445701e-02 5.54024875e-02 7.59253323e-01 4.13770795e-01 1.12728551e-01 7.28317723e-02 -1.27579555e-01 -1.15900233e-01 1.65968224e-01 5.30993223e-01 2.15779826e-01 -2.42931411e-01 -1.39966920e-01 -1.48321331e-01 1.57350615e-01 1.99500859e-01 9.67553020e-01 6.86373413e-01 -4.26342458e-01 -7.61602640e-01 -1.05991685e+00 -2.11663861e-02 -6.11194730e-01 -6.68840766e-01 -1.34884343e-01 4.70942222e-02 -1.20007098e-02 1.29748845e+00 -1.81688622e-01 1.86957035e-03 -2.43385106e-01 -3.39442879e-01 5.94981611e-01 -2.86159247e-01 -7.46757165e-02 4.18297462e-02 -3.62398237e-01 -4.94636655e-01 -6.60173893e-01 -5.11664093e-01 -6.47630751e-01 -5.97390354e-01 -6.13475263e-01 6.03278764e-02 5.39491594e-01 6.82238281e-01 4.62085247e-01 2.86097312e-03 8.53514850e-01 -9.61169004e-01 -4.33604956e-01 -9.87522900e-01 -8.27297449e-01 4.71711636e-01 1.07941663e+00 -3.25546890e-01 -1.01742601e+00 7.57912219e-01]
[11.635109901428223, -2.7722268104553223]
84772c99-31d1-47e7-8db4-7497abdb05c7
building-and-modelling-multilingual
null
null
https://aclanthology.org/L14-1340
https://aclanthology.org/L14-1340.pdf
Building and Modelling Multilingual Subjective Corpora
Building multilingual opinionated models requires multilingual corpora annotated with opinion labels. Unfortunately, such kind of corpora are rare. We consider opinions in this work as subjective or objective. In this paper, we introduce an annotation method that can be reliably transferred across topic domains and across languages. The method starts by building a classifier that annotates sentences into subjective/objective label using a training data from {``}movie reviews{''} domain which is in English language. The annotation can be transferred to another language by classifying English sentences in parallel corpora and transferring the same annotation to the same sentences of the other language. We also shed the light on the link between opinion mining and statistical language modelling, and how such corpora are useful for domain specific language modelling. We show the distinction between subjective and objective sentences which tends to be stable across domains and languages. Our experiments show that language models trained on objective (respectively subjective) corpus lead to better perplexities on objective (respectively subjective) test.
['Kamel Sma{\\"\\i}li', 'David Langlois', 'Motaz Saad']
2014-05-01
null
null
null
lrec-2014-5
['subjectivity-analysis']
['natural-language-processing']
[-1.34852052e-01 2.93597102e-01 -3.08912843e-01 -9.27465200e-01 -9.51039433e-01 -9.47268546e-01 6.20726645e-01 5.21192193e-01 -6.86963141e-01 1.22211564e+00 2.19678938e-01 -4.68647242e-01 3.70473176e-01 -5.64670324e-01 -3.86860669e-01 -5.28254688e-01 2.44609237e-01 6.10761821e-01 2.54311621e-01 -4.76152807e-01 1.65717512e-01 -6.21835105e-02 -1.05940151e+00 5.48605740e-01 9.37680721e-01 8.71895552e-01 9.95372683e-02 5.13838649e-01 -5.45307696e-01 9.40812230e-01 -9.25133765e-01 -1.17352390e+00 1.52125358e-04 -3.38498473e-01 -1.03454518e+00 2.78445721e-01 1.17456108e-01 4.48760420e-01 8.48594129e-01 1.20620954e+00 3.62327546e-01 -1.70656681e-01 9.99500453e-01 -1.06025219e+00 -9.28843021e-01 7.73796976e-01 -4.42566335e-01 -1.31313339e-01 2.84070671e-01 -5.86713552e-01 1.07018805e+00 -9.67746139e-01 8.76344919e-01 1.35909379e+00 5.26801288e-01 4.37033206e-01 -9.29813683e-01 -3.05829912e-01 6.02424681e-01 -9.69970301e-02 -9.53326106e-01 -1.53862864e-01 6.25757694e-01 -7.31292129e-01 8.28666747e-01 2.44816989e-02 3.47928166e-01 1.00564003e+00 4.15375471e-01 8.46903861e-01 1.37524688e+00 -7.38259137e-01 4.02491391e-01 1.18246448e+00 5.31362712e-01 3.78543258e-01 2.49430001e-01 -6.67828977e-01 -5.97082615e-01 1.47959784e-01 -9.12142769e-02 -5.86490571e-01 -9.58785042e-03 -2.94308662e-01 -9.44641590e-01 1.09918797e+00 -3.01161110e-01 5.65927982e-01 -2.43413240e-01 -5.57612002e-01 8.04243982e-01 7.81014323e-01 1.30418086e+00 4.48955983e-01 -1.35172582e+00 8.15565884e-02 -6.45597279e-01 -2.34142378e-01 1.20259166e+00 8.53142977e-01 9.01141822e-01 -2.12573200e-01 1.76475406e-01 1.28912175e+00 5.46360552e-01 8.61479402e-01 6.54981136e-01 -7.66237199e-01 4.67440456e-01 5.08663416e-01 2.12041363e-01 -8.68552923e-01 -1.58580810e-01 -2.41365433e-01 -4.20455933e-01 2.17799377e-02 1.29596099e-01 -6.97706699e-01 -5.37342846e-01 1.66818655e+00 2.24001825e-01 -8.20031881e-01 7.22858131e-01 4.19019580e-01 7.49463022e-01 7.89959848e-01 3.83036703e-01 -7.39251912e-01 1.58595979e+00 -1.16775262e+00 -8.97549748e-01 -4.59261924e-01 1.10160458e+00 -1.16981184e+00 9.06532049e-01 6.00569189e-01 -9.69482243e-01 -6.64018869e-01 -7.14589477e-01 -1.60715077e-02 -9.32375968e-01 5.30137420e-01 5.51389039e-01 8.50798488e-01 -1.26420689e+00 -1.98480710e-02 -3.65319252e-01 -6.15365922e-01 -1.94985852e-01 2.81588793e-01 -4.65468526e-01 2.83415109e-01 -1.52945280e+00 1.20177877e+00 3.07840049e-01 -1.24247290e-01 -2.49135226e-01 -4.01729010e-02 -1.00986195e+00 -5.64193964e-01 -2.99354270e-02 -3.19424093e-01 1.38398600e+00 -1.83067441e+00 -1.46269226e+00 1.30954564e+00 -5.39084792e-01 -2.20559850e-01 2.01148331e-01 -2.34166831e-01 -8.27157140e-01 -1.71309739e-01 5.66976070e-01 6.14608824e-01 4.91041362e-01 -1.42510843e+00 -9.67428744e-01 -5.49755633e-01 1.50319979e-01 2.48979792e-01 -6.88963294e-01 4.50094104e-01 -4.28478688e-01 -5.80863535e-01 -1.02940634e-01 -8.22221756e-01 -1.16328888e-01 -4.29911166e-01 -1.12254813e-01 -7.61067629e-01 4.52802271e-01 -7.03609943e-01 1.09219599e+00 -1.79522526e+00 2.48016343e-01 5.02036773e-02 -3.69356722e-01 -3.86672863e-03 -2.12619826e-02 3.71465385e-01 -1.14406720e-01 2.55312264e-01 -1.62527993e-01 -6.61806345e-01 2.98108310e-01 4.00159836e-01 -3.31612617e-01 1.85196579e-01 2.50994593e-01 5.07114410e-01 -9.43054914e-01 -7.66767204e-01 -2.33563498e-01 2.45180652e-01 -3.53718668e-01 1.47597492e-01 -1.05694197e-01 3.56424868e-01 -5.34421742e-01 4.98046756e-01 4.66563404e-01 1.18751965e-01 4.35199916e-01 -1.94642812e-01 -2.32149169e-01 3.50819081e-01 -8.87610376e-01 1.64743567e+00 -9.30092335e-01 7.45134950e-01 1.29831597e-01 -1.19291770e+00 1.26298845e+00 6.33530021e-01 1.83184713e-01 -4.29457933e-01 1.54994264e-01 4.77087617e-01 -2.24580079e-01 -7.83322573e-01 6.29430711e-01 -5.67400992e-01 -3.91413718e-01 4.87228304e-01 3.67672235e-01 -6.02991343e-01 8.34889650e-01 -3.03296037e-02 2.95765877e-01 7.79967234e-02 3.70396495e-01 -6.77193761e-01 1.06775105e+00 8.03145394e-02 3.66177440e-01 2.79896498e-01 -8.80984664e-02 1.65025085e-01 6.41438961e-01 -3.36832643e-01 -8.69397938e-01 -7.15280354e-01 -4.52800184e-01 1.51162446e+00 -1.76682130e-01 -4.44724888e-01 -6.87184572e-01 -1.08114171e+00 -4.89939570e-01 7.47977436e-01 -4.44139361e-01 3.56787413e-01 -9.94901359e-02 -9.06833172e-01 1.05649874e-01 3.68124545e-01 1.98876798e-01 -1.08671176e+00 1.96612656e-01 2.20997900e-01 -1.87014505e-01 -1.08418429e+00 -1.60698622e-01 5.32372952e-01 -5.93011916e-01 -6.56479478e-01 -7.58747101e-01 -1.24905360e+00 7.45429993e-01 -4.43531543e-01 1.49816120e+00 -5.41698813e-01 6.66780233e-01 4.46562618e-01 -7.31653273e-01 -8.04964602e-01 -6.68896496e-01 3.42496097e-01 3.72887462e-01 4.90419380e-02 9.12034750e-01 -2.21623674e-01 -1.29512295e-01 2.30437845e-01 -1.00652802e+00 -4.36777711e-01 4.27706301e-01 5.18194556e-01 2.79003382e-01 1.76347047e-02 7.73932219e-01 -1.29823422e+00 1.03371918e+00 -4.99028563e-01 -3.43557596e-01 5.34558713e-01 -6.67211175e-01 1.60704389e-01 5.86176395e-01 -2.17797473e-01 -1.38494921e+00 -1.70321882e-01 -3.33753347e-01 5.52587807e-01 -2.91457683e-01 8.68957043e-01 -9.91774201e-02 4.28848952e-01 7.02646971e-01 -2.54956931e-01 -4.91735846e-01 -4.08365071e-01 3.06209266e-01 1.08690250e+00 -1.70383174e-02 -7.49146402e-01 2.60926455e-01 2.34100208e-01 -6.35507405e-01 -7.63015211e-01 -1.27016938e+00 -5.25880098e-01 -1.03928995e+00 -2.40280136e-01 1.18646157e+00 -9.13678229e-01 -1.15819722e-01 3.00554037e-01 -1.41096890e+00 -8.97714123e-03 -1.34879634e-01 6.11466050e-01 -1.80195615e-01 3.24104995e-01 -6.50617778e-01 -1.01878524e+00 -3.12719941e-01 -1.07556117e+00 1.21165848e+00 8.70143697e-02 -6.81361437e-01 -1.84225023e+00 6.04598343e-01 4.35284048e-01 2.57729199e-02 -3.57658833e-01 6.93145692e-01 -1.00949705e+00 5.30801117e-01 -4.15461630e-01 6.91125840e-02 1.12238228e+00 1.77557215e-01 2.58554965e-01 -1.00296199e+00 -5.76183051e-02 3.05577576e-01 -7.43229508e-01 6.21640027e-01 1.48795247e-01 4.84096944e-01 -1.22514099e-01 -8.53686929e-02 -2.57140666e-01 1.24533117e+00 3.11844647e-01 3.61245185e-01 3.02343130e-01 3.61218482e-01 1.17707980e+00 7.91865826e-01 -9.23849568e-02 6.54133022e-01 3.39268833e-01 -4.96105194e-01 -3.40145677e-01 4.80624050e-01 8.15157294e-02 1.09974825e+00 1.81243408e+00 7.81426206e-02 -5.61898291e-01 -8.21866453e-01 8.40578794e-01 -1.76110804e+00 -5.21448374e-01 -2.32142642e-01 1.83889496e+00 1.35878205e+00 2.60541737e-01 -2.01553896e-01 -9.31063071e-02 5.70868611e-01 1.21901751e-01 3.11002117e-02 -1.14687777e+00 -4.05011415e-01 2.74469525e-01 2.02538639e-01 9.33083057e-01 -1.28563142e+00 1.03785825e+00 5.77824831e+00 8.50677669e-01 -1.22863805e+00 5.07479131e-01 8.35351467e-01 5.86508989e-01 -4.53344107e-01 1.18708916e-01 -8.72297645e-01 2.26352990e-01 1.13061643e+00 -1.71287119e-01 -4.07584012e-01 1.00095892e+00 2.00542733e-01 -3.19971889e-01 -9.92778718e-01 4.26691443e-01 5.37853539e-01 -6.15678489e-01 -5.91040328e-02 -2.90442973e-01 1.14449430e+00 6.40117452e-02 -2.33969569e-01 5.08716881e-01 5.71037829e-01 -6.94579840e-01 7.65344381e-01 3.63912880e-01 5.76066077e-01 -6.01061761e-01 1.46615779e+00 3.40339810e-01 -9.44778621e-01 5.74807823e-01 -6.04097307e-01 -1.27350584e-01 3.87090534e-01 7.87847042e-01 -4.86149788e-01 6.30712450e-01 6.39079928e-01 9.81258392e-01 -6.78714037e-01 2.40021557e-01 -6.95978880e-01 5.35580993e-01 -2.29421034e-02 -2.37211004e-01 2.28953019e-01 -6.44893348e-01 9.83549878e-02 1.53730667e+00 2.74045914e-01 -2.90442169e-01 3.28955382e-01 2.20716536e-01 4.92170937e-02 9.29327905e-01 -8.43867898e-01 -1.19007556e-02 -1.77828535e-01 1.39863575e+00 -9.13548887e-01 -7.32642114e-01 -7.47415304e-01 1.04402328e+00 7.67372102e-02 4.23777193e-01 -3.88524026e-01 -2.17832923e-01 1.43252894e-01 -3.37266773e-01 5.15080243e-02 1.40534759e-01 -2.38747746e-01 -1.46301794e+00 1.95546910e-01 -8.42591524e-01 1.78719506e-01 -8.97534788e-01 -1.80453479e+00 1.09512365e+00 -1.01893991e-01 -1.05390882e+00 -2.37408251e-01 -1.04002273e+00 -3.25492144e-01 8.69774163e-01 -1.48935866e+00 -9.75754797e-01 4.08224374e-01 3.34280163e-01 8.21986556e-01 -2.08014578e-01 1.10924459e+00 2.70829111e-01 -3.72607112e-01 2.86379308e-01 6.28825501e-02 1.74546853e-01 1.26513469e+00 -1.60621548e+00 -1.87803134e-01 5.26982605e-01 7.88161755e-02 7.12369621e-01 8.26113582e-01 -5.18342078e-01 -4.92202520e-01 -8.64110589e-01 1.82228887e+00 -8.30461860e-01 1.11784804e+00 -3.04922014e-01 -6.20493114e-01 6.32007897e-01 8.91053021e-01 -5.33805013e-01 1.18991637e+00 4.44992244e-01 -2.16604799e-01 -1.52341589e-01 -8.33446324e-01 3.41463536e-01 2.98865229e-01 -7.58049607e-01 -8.04342806e-01 7.89377451e-01 6.09874964e-01 1.31231979e-01 -1.01371121e+00 3.92793417e-01 4.46526945e-01 -7.51499236e-01 4.21416849e-01 -6.95270240e-01 5.99653840e-01 -3.80147249e-01 -1.82034492e-01 -1.62254560e+00 2.93560952e-01 -3.97895090e-02 7.41412342e-01 1.48024690e+00 1.34836125e+00 -6.76365495e-01 3.06975812e-01 5.28260410e-01 -5.45032062e-02 -4.15314525e-01 -4.75294888e-01 -5.04405260e-01 6.46484196e-01 -7.37503529e-01 -1.52274445e-01 1.24986172e+00 4.38575894e-01 1.03984451e+00 -1.82752490e-01 5.09166997e-03 1.03436232e-01 -5.96705340e-02 4.98932034e-01 -1.28467393e+00 -1.65459029e-02 -1.57563329e-01 -2.91630656e-01 -9.10260558e-01 6.75868511e-01 -8.27560186e-01 1.83249474e-01 -1.54562104e+00 7.99808502e-02 -1.03667982e-01 -1.19368173e-01 2.97841907e-01 -2.33618513e-01 4.61288035e-01 -3.24765325e-01 -1.27533615e-01 -9.56956267e-01 3.55712444e-01 9.33033884e-01 -1.84031144e-01 -9.00184456e-03 1.18967436e-01 -6.80744350e-01 1.21443522e+00 8.00980985e-01 -5.03137887e-01 -4.64052200e-01 -6.15549505e-01 8.69777679e-01 -2.85442442e-01 -5.11486948e-01 -5.62514663e-01 6.62083700e-02 -3.01964954e-02 1.29011959e-01 -3.48347932e-01 2.26234868e-02 -8.70076120e-01 -4.27673310e-01 -1.81190670e-03 -4.53543693e-01 4.07642365e-01 4.15821224e-02 3.02412480e-01 -8.44426274e-01 -6.37055814e-01 6.27340674e-01 -2.56250948e-01 -5.91700912e-01 -2.60370225e-01 -7.27403402e-01 5.51230237e-02 8.87024403e-01 2.46729314e-01 -4.52930741e-02 -6.07763767e-01 -1.04643035e+00 3.31801683e-01 5.11638701e-01 4.98816073e-01 1.18580461e-02 -1.10633421e+00 -7.65712559e-01 -7.81928282e-03 2.87685752e-01 -3.03693384e-01 -1.61411136e-01 8.22008431e-01 -4.38756496e-01 6.55418694e-01 1.90579087e-01 -5.11529863e-01 -1.22225809e+00 4.47325289e-01 2.05921724e-01 -4.26765084e-01 6.37535751e-01 9.60569739e-01 2.48679921e-01 -1.09587622e+00 1.55268252e-01 -3.27476621e-01 -7.82328367e-01 7.98090756e-01 2.40247250e-01 -1.44768372e-01 2.43541151e-01 -9.96377110e-01 -3.10682297e-01 7.82136440e-01 -1.06276400e-01 -6.05364442e-01 1.29308116e+00 -5.54138362e-01 -8.41225863e-01 1.14567578e+00 1.33087063e+00 5.62670767e-01 -4.45333898e-01 -2.08385929e-01 4.37640697e-01 2.83157080e-01 -1.31938756e-01 -7.18885064e-01 -9.34347391e-01 8.28106761e-01 4.53363717e-01 7.72410810e-01 9.64963555e-01 3.01746339e-01 3.39108318e-01 5.16229570e-01 3.77905041e-01 -1.60440433e+00 -2.09414482e-01 8.01422715e-01 7.24054635e-01 -1.59081602e+00 -1.48217008e-01 -3.83128315e-01 -1.12636340e+00 1.10740674e+00 3.06299776e-01 1.01761915e-01 1.02869701e+00 7.12047219e-02 7.37160861e-01 -3.50415111e-01 -8.72135162e-01 -2.11345747e-01 3.73574138e-01 5.48101246e-01 1.15194035e+00 1.59465834e-01 -9.49355602e-01 7.23433554e-01 -2.97403634e-01 -1.13913208e-01 3.15260470e-01 8.78969431e-01 -5.03549099e-01 -1.81365097e+00 -1.99028969e-01 -6.78261044e-03 -7.58902729e-01 -3.45495224e-01 -8.32799733e-01 5.43614209e-01 5.36099255e-01 1.33634365e+00 -4.52099517e-02 -5.72975911e-02 2.75830030e-01 5.28202236e-01 -9.46630910e-03 -1.05810428e+00 -5.27560413e-01 2.34420359e-01 4.88676667e-01 1.13209918e-01 -1.33458710e+00 -5.64907849e-01 -8.05793464e-01 5.53860180e-02 -3.51427883e-01 9.10821021e-01 8.19316030e-01 1.24245632e+00 -1.88427240e-01 1.70133129e-01 6.21376932e-01 -2.74620891e-01 -1.76710133e-02 -1.30698562e+00 -6.97202921e-01 4.01439130e-01 7.44028986e-02 -1.66551158e-01 -7.13055789e-01 7.74129391e-01]
[11.29684066772461, 6.927677631378174]
1d0a1a43-e1ef-487d-9635-8fc4105f2be1
biometric-recognition-why-not-massively
2203.03719
null
https://arxiv.org/abs/2203.03719v2
https://arxiv.org/pdf/2203.03719v2.pdf
Biometric recognition: why not massively adopted yet?
Although there has been a dramatically reduction on the prices of capturing devices and an increase on computing power in the last decade, it seems that biometric systems are still far from massive adoption for civilian applications. This paper deals with the causes of this phenomenon, as well as some misconceptions regarding biometric identification.
['Marcos Faundez-Zanuy']
2022-02-23
null
null
null
null
['misconceptions']
['miscellaneous']
[ 1.99268341e-01 -2.30724737e-01 -2.77933050e-02 -6.37729168e-01 8.60931128e-02 -5.12055159e-01 5.86415172e-01 1.44160688e-01 -5.83879113e-01 7.83096611e-01 -8.21593106e-02 -6.25932455e-01 -5.12680411e-02 -6.79702878e-01 8.49392042e-02 -5.33431768e-01 5.04490316e-01 1.73322693e-01 -1.70889914e-01 -1.16597384e-01 6.10209107e-01 8.09962511e-01 -1.28776014e+00 -2.42328972e-01 4.93474156e-01 8.58457744e-01 -5.15132189e-01 3.70671272e-01 -2.17082918e-01 7.38478899e-02 -8.48090291e-01 -1.09701777e+00 5.73411025e-02 -3.90641809e-01 -6.93871975e-01 3.47067602e-03 2.35446960e-01 -5.19977272e-01 -1.86010823e-01 9.46510732e-01 5.86984515e-01 -3.75989676e-01 4.49538946e-01 -1.08760428e+00 -7.18209147e-01 2.51982391e-01 -7.30960369e-01 2.33189076e-01 7.79142618e-01 -3.24377790e-02 4.65813667e-01 -3.98906708e-01 1.39566869e-01 1.12053084e+00 6.90159798e-01 6.97970390e-01 -8.91365051e-01 -6.18082404e-01 -2.73335755e-01 -1.46881342e-01 -1.69487953e+00 -5.77106774e-01 4.95553583e-01 -4.10246909e-01 7.35355258e-01 4.12620723e-01 7.26527393e-01 6.13596797e-01 4.83987600e-01 2.11124897e-01 1.17002141e+00 -5.80039024e-01 6.94422647e-02 6.34792745e-01 3.66525277e-02 3.17194074e-01 9.39427376e-01 -1.60448015e-01 -3.39556634e-01 -5.53111315e-01 9.38756645e-01 1.18704066e-01 1.30447131e-02 6.19496480e-02 -5.43259501e-01 5.31866312e-01 -2.16016427e-01 8.19861114e-01 -5.35377443e-01 -2.24072248e-01 3.79156172e-01 -1.87299922e-02 1.38594478e-01 3.42824280e-01 -2.56836414e-01 -6.21990621e-01 -6.49149656e-01 6.03679717e-02 8.44860673e-01 5.23259044e-01 3.88089567e-01 1.01325959e-01 8.21009457e-01 4.99366730e-01 1.39490396e-01 6.12570226e-01 1.46439925e-01 -4.82265353e-01 -2.04897299e-02 7.26602733e-01 2.89002717e-01 -1.34244490e+00 -1.78824425e-01 1.80538833e-01 -7.76268244e-01 -1.78159431e-01 6.53670073e-01 -3.28214586e-01 -4.89997268e-01 1.31944370e+00 -1.46946937e-01 -3.59360129e-01 -2.26386279e-01 4.32517320e-01 3.69798928e-01 1.55281797e-01 1.27830416e-01 -1.96349606e-01 1.24607110e+00 2.62304842e-01 -8.24977040e-01 -4.56171017e-03 -1.90318264e-02 -8.67944837e-01 5.87544799e-01 7.45199859e-01 -9.41148281e-01 -3.83651018e-01 -6.62696958e-01 4.20717597e-01 -3.43837261e-01 -2.56100893e-01 9.31126833e-01 1.92733908e+00 -6.68577671e-01 3.59193087e-01 -7.01689661e-01 -1.04495192e+00 2.58056402e-01 6.64256871e-01 -2.57700205e-01 1.90792754e-01 -8.08409810e-01 1.07403040e+00 1.66505277e-01 3.29561472e-01 4.04366910e-01 -1.84810042e-01 -5.10242462e-01 -2.46586084e-01 -6.18851632e-02 -1.92593336e-01 6.42238259e-01 -7.03010440e-01 -1.37667775e+00 1.12331545e+00 -1.71428770e-01 -1.47571504e-01 2.55595654e-01 9.80802551e-02 -9.67945993e-01 -2.27954268e-01 -2.80010104e-01 3.25820029e-01 4.28810060e-01 -8.79376411e-01 -6.82865322e-01 -7.70730019e-01 -1.10431939e-01 -1.22654013e-01 -3.60610366e-01 3.36611271e-01 -7.01581687e-02 -1.79904297e-01 9.21228454e-02 -9.96280432e-01 -1.72753856e-02 -4.19986308e-01 9.47759002e-02 -2.50855029e-01 6.24589264e-01 -5.09806752e-01 1.13384080e+00 -2.38449383e+00 -5.68168402e-01 4.28970367e-01 -2.99283952e-01 4.28480178e-01 3.93352091e-01 7.10618258e-01 9.57337245e-02 2.44932026e-01 3.12018216e-01 2.85925001e-01 -1.67406201e-01 5.37598848e-01 -3.85603607e-01 7.13266909e-01 -4.34205718e-02 4.98204768e-01 -8.13062549e-01 -2.14144692e-01 4.70410109e-01 5.96421540e-01 5.96459880e-02 -1.61424324e-01 7.71485150e-01 5.11172712e-01 -4.49244469e-01 1.15239024e+00 8.73787999e-01 6.94688708e-02 3.91956091e-01 1.89360976e-01 -2.86883712e-01 2.45056421e-01 -9.44665670e-01 1.01300633e+00 1.36363819e-01 3.40929240e-01 -2.96056330e-01 -5.58297217e-01 8.95146430e-01 6.21705174e-01 7.33599484e-01 -5.15216053e-01 4.07707632e-01 5.07833362e-01 3.43660176e-01 -5.03144860e-01 6.02517009e-01 -5.19710183e-01 -1.49929121e-01 1.53838500e-01 -4.69777048e-01 -1.01225689e-01 -1.39702737e-01 -2.33262539e-01 4.67682183e-01 -2.43605971e-01 5.65678120e-01 -2.10014373e-01 5.63396335e-01 -5.97603917e-02 4.33466852e-01 2.90199518e-01 -3.03842157e-01 1.70961663e-01 1.55573070e-01 -7.48781204e-01 -5.76931357e-01 -7.33716846e-01 -4.45455223e-01 3.46704543e-01 3.46205801e-01 -2.47236669e-01 -8.81660402e-01 -2.30046123e-01 2.59111732e-01 3.83380502e-01 -2.73817837e-01 1.42823532e-01 -2.45892584e-01 -6.77682281e-01 9.22203779e-01 4.85773742e-01 6.61972284e-01 -6.58282161e-01 -1.04309773e+00 1.94704741e-01 1.17752895e-01 -1.10271645e+00 -5.43881729e-02 -2.46999711e-01 -1.02189708e+00 -8.23423386e-01 -6.94662690e-01 -1.75506830e-01 7.61028171e-01 4.22460228e-01 6.90318942e-01 3.82409692e-01 -5.42460680e-01 7.53152132e-01 -1.27667546e-01 -8.47818792e-01 -1.59550235e-02 9.92381424e-02 3.28583211e-01 7.77778476e-02 1.16722298e+00 -3.06324899e-01 -5.86401999e-01 2.25978643e-01 -6.12471700e-01 -5.66421390e-01 4.17602718e-01 3.34826916e-01 -2.75343686e-01 1.75544053e-01 5.48266172e-01 -7.50695467e-01 7.87038445e-01 -1.23705626e-01 -3.21516335e-01 4.86142039e-02 -1.06249154e+00 -3.92818063e-01 8.56253058e-02 -3.36483605e-02 -1.03548241e+00 -1.73180461e-01 -3.47553492e-01 5.12584686e-01 -5.29606879e-01 1.95846617e-01 1.96824223e-02 -5.50726891e-01 4.49131191e-01 7.43348449e-02 2.23943293e-01 -4.56153154e-01 -3.85179073e-01 1.07090271e+00 3.83958608e-01 -3.89627844e-01 6.16319895e-01 5.45123577e-01 -6.42267466e-02 -1.31717384e+00 -3.33992928e-01 -4.75386202e-01 -3.81140888e-01 -3.73527646e-01 4.99513447e-01 -3.52441847e-01 -1.14866388e+00 7.05868483e-01 -9.00508583e-01 6.20634496e-01 6.81626052e-02 5.01517534e-01 -7.09079430e-02 7.78034210e-01 -5.19979835e-01 -1.21631885e+00 -2.23337859e-01 -7.91243672e-01 5.05957246e-01 5.67552984e-01 -6.87478304e-01 -9.40872431e-01 -7.54482225e-02 4.44414079e-01 7.88572788e-01 2.12101117e-01 6.57343388e-01 -3.79624605e-01 -4.91587967e-02 -8.59981656e-01 -2.32004151e-01 1.17413148e-01 8.03369164e-01 3.44819799e-02 -1.08760142e+00 -3.82482171e-01 6.08024374e-02 -3.98452803e-02 -1.29156724e-01 3.40330541e-01 8.24000955e-01 4.28208485e-02 -4.48460340e-01 8.04736838e-02 1.38968945e+00 1.12410235e+00 1.01583934e+00 2.59153277e-01 -6.10090457e-02 6.85019016e-01 4.53845978e-01 3.66238356e-01 2.33393356e-01 5.52471817e-01 -4.49310392e-02 -7.08557516e-02 5.32888293e-01 -2.31916636e-01 -9.01899114e-02 6.33200645e-01 -7.48674810e-01 8.74413326e-02 -1.06758320e+00 3.45490932e-01 -1.34342909e+00 -8.48788440e-01 -4.52484004e-03 2.45139170e+00 3.45441222e-01 8.71550292e-02 3.59117031e-01 2.55793810e-01 6.63985491e-01 -1.60075650e-01 -4.34075177e-01 -7.77660370e-01 1.05754994e-01 1.43315449e-01 5.43003619e-01 1.41621724e-01 -6.28316522e-01 5.55039167e-01 8.52098560e+00 2.41662964e-01 -1.54681945e+00 -2.08874047e-01 6.99062943e-01 6.25131845e-01 3.62900421e-02 6.10437393e-02 -6.94023967e-01 5.37038445e-01 8.23898435e-01 -3.68530631e-01 5.14526740e-02 5.68145752e-01 -1.25362009e-01 -6.36268735e-01 -6.52489007e-01 1.29950392e+00 1.85933262e-01 -8.03236485e-01 2.92226493e-01 6.20157897e-01 2.31286615e-01 -6.84857249e-01 8.03789347e-02 -1.17833883e-01 -4.88886863e-01 -1.00394356e+00 1.71486005e-01 4.16745335e-01 6.75380230e-01 -9.03515935e-01 1.17514682e+00 5.59022464e-02 -7.76472211e-01 2.59520531e-01 -5.31128109e-01 -5.53373218e-01 3.29159200e-04 3.01494062e-01 -6.19064808e-01 3.65822315e-01 6.18909895e-01 -7.56320357e-02 -5.22533894e-01 9.61510718e-01 1.74003199e-01 5.34559011e-01 -1.83338463e-01 -2.71118045e-01 2.80111525e-02 -5.81660867e-01 -4.70397696e-02 1.12788320e+00 2.46377975e-01 2.99484670e-01 -5.00147820e-01 2.48212084e-01 2.93481648e-01 -1.30438348e-02 -8.35674584e-01 -4.06561404e-01 4.23492253e-01 9.75812674e-01 -7.99214721e-01 -2.21400380e-01 -8.45903337e-01 8.52081954e-01 -6.51890039e-01 6.28447905e-03 -6.60899818e-01 -3.98360550e-01 4.31174040e-01 4.96656060e-01 -4.00007784e-01 -2.25366831e-01 -5.25933206e-01 -9.21220958e-01 -5.24451137e-02 -9.39203501e-01 2.39834294e-01 -2.81094104e-01 -9.33487296e-01 3.51553261e-01 -1.50046259e-01 -1.00449789e+00 -2.27013454e-01 -6.52652204e-01 -3.40766251e-01 1.06607056e+00 -7.25077212e-01 -8.33723545e-01 3.48340571e-02 2.43237704e-01 7.16122985e-03 -1.96877554e-01 1.25522745e+00 2.32101738e-01 -2.77252793e-01 6.17306113e-01 -6.15345687e-02 2.68577725e-01 6.52051806e-01 -9.11620140e-01 4.04285312e-01 4.67756152e-01 -2.62439013e-01 1.18186879e+00 7.35763550e-01 -4.16993469e-01 -1.50775599e+00 4.29208018e-02 1.24735999e+00 -4.45439905e-01 4.11568522e-01 -1.24421105e-01 -6.02457047e-01 5.74459314e-01 4.02631938e-01 -8.10020864e-01 1.17464256e+00 2.13396892e-01 -1.34741277e-01 -1.43927559e-01 -1.42470860e+00 5.30640721e-01 5.11165500e-01 -5.23175955e-01 -4.96914834e-01 -1.89423412e-01 -5.02892852e-01 -1.10658720e-01 -1.03030539e+00 2.83455402e-01 1.24685812e+00 -8.90373230e-01 7.18526006e-01 -3.03084254e-01 -2.99703389e-01 1.42320711e-02 -4.32135388e-02 -5.71452558e-01 -2.68994391e-01 -7.00685859e-01 5.73677301e-01 1.37800920e+00 2.07227066e-01 -1.04574287e+00 1.05782771e+00 1.57308269e+00 7.76859999e-01 -7.09539354e-01 -8.07402253e-01 -7.43787885e-01 -1.61546052e-01 -6.01549447e-03 5.49592972e-01 1.06819654e+00 4.90455359e-01 2.31294394e-01 -5.06249309e-01 -1.10979602e-01 6.63005471e-01 9.18751359e-02 6.32031739e-01 -1.42288184e+00 -9.21561569e-03 -4.35055971e-01 -1.10633838e+00 -6.91879570e-01 -5.86764097e-01 6.54443540e-03 -4.75633115e-01 -1.20273256e+00 4.00907040e-01 -2.58597910e-01 -3.17977309e-01 2.13855416e-01 7.02548176e-02 3.59126031e-01 -4.33731638e-02 6.28274828e-02 1.95620388e-01 -4.17365432e-01 6.24368966e-01 8.81949961e-02 1.93465557e-02 2.92649865e-01 -1.15779185e+00 7.54248798e-01 1.02114284e+00 -1.50705099e-01 -1.85942352e-01 -1.80516049e-01 4.57520813e-01 1.97995335e-01 -1.65748924e-01 -1.03534985e+00 3.04560252e-02 -2.23514825e-01 7.04542458e-01 -5.20710826e-01 2.87621140e-01 -9.09457505e-01 5.98807633e-01 8.18391860e-01 3.24693739e-01 3.53721201e-01 3.76398504e-01 1.91618472e-01 -9.04697627e-02 -3.46191168e-01 7.73254454e-01 1.05407313e-01 -3.56855750e-01 -9.54279080e-02 -6.79917634e-01 -7.09266663e-01 8.15893829e-01 -9.36735034e-01 -7.91477337e-02 -3.65696728e-01 -4.89656270e-01 -2.56193817e-01 9.20589149e-01 2.83620775e-01 5.55027544e-01 -9.66807663e-01 -1.63071901e-01 1.44759625e-01 -1.10471711e-01 -6.96313620e-01 1.83981299e-01 9.00873303e-01 -6.89498067e-01 6.79359317e-01 -7.17262983e-01 -3.45929146e-01 -1.68356121e+00 3.58387768e-01 2.13176489e-01 3.01050007e-01 -5.03691852e-01 5.78959823e-01 -3.59142959e-01 1.00125626e-01 1.38344109e-01 5.66796251e-02 -2.36473858e-01 -1.89950451e-01 7.02618659e-01 6.04585111e-01 4.46621068e-02 -6.57074451e-01 -7.50766397e-01 6.55666947e-01 -1.91934705e-01 -2.01126784e-01 9.16331410e-01 -1.86631173e-01 -4.14352149e-01 8.20413530e-01 5.48438191e-01 2.61126757e-01 -3.12084407e-01 2.80597270e-01 2.18489915e-01 -1.14073157e+00 -4.71565127e-01 -7.13036120e-01 -8.00536513e-01 7.74258852e-01 6.07987821e-01 5.95158219e-01 1.09981346e+00 -3.45199764e-01 7.74156868e-01 5.32239079e-01 1.12910104e+00 -1.06654429e+00 -6.57650352e-01 1.38422787e-01 4.95335370e-01 -7.88402438e-01 3.55065376e-01 -4.22652841e-01 -5.24617553e-01 1.08631349e+00 1.19225793e-01 3.29266712e-02 7.31921792e-01 2.33092993e-01 1.82494134e-01 -1.19496346e-01 -5.99124171e-02 2.44280905e-01 -1.71503752e-01 7.58522630e-01 1.03338170e+00 2.27433830e-01 -9.64584470e-01 1.16401203e-01 -1.97388083e-01 2.19769880e-01 6.16366208e-01 1.22427726e+00 -3.35701346e-01 -1.50405121e+00 -7.04777241e-01 6.78378820e-01 -1.03771448e+00 4.40148890e-01 -9.27827001e-01 7.97103286e-01 3.10637474e-01 1.20931971e+00 3.49098258e-02 -5.49747944e-01 4.03284311e-01 2.16857523e-01 7.84980416e-01 -3.45982641e-01 -6.06139839e-01 6.60226867e-03 1.87459081e-01 -1.54086858e-01 -6.73163056e-01 -1.02541637e+00 -8.28934550e-01 -1.05772614e+00 -5.88832378e-01 5.40757656e-01 8.05332065e-01 8.91823709e-01 1.08723015e-01 -2.11062208e-01 2.78736144e-01 -2.78876603e-01 -4.51752961e-01 -9.41391945e-01 -1.06842637e+00 2.56074548e-01 -1.76125780e-01 -3.42408359e-01 6.05480075e-02 -5.17472476e-02]
[13.048223495483398, 1.075966477394104]
0ba809b1-11b7-4319-b326-71439f2dc296
robust-homography-estimation-via-dual
null
null
http://openaccess.thecvf.com/content_CVPR_2020/html/Ding_Robust_Homography_Estimation_via_Dual_Principal_Component_Pursuit_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Ding_Robust_Homography_Estimation_via_Dual_Principal_Component_Pursuit_CVPR_2020_paper.pdf
Robust Homography Estimation via Dual Principal Component Pursuit
We revisit robust estimation of homographies over point correspondences between two or three views, a fundamental problem in geometric vision. The analysis serves as a platform to support a rigorous investigation of Dual Principal Component Pursuit (DPCP) as a valid and powerful alternative to RANSAC for robust model fitting in multiple-view geometry. Homography fitting is cast as a robust nullspace estimation problem over either homographic or epipolar/trifocal embeddings. We prove that the nullspace of epipolar or trifocal embeddings in the homographic scenario, of dimension 3 and 6 for two and three views respectively, is defined by unique, computable homographies. Experiments show that DPCP performs on par with USAC with local optimization, while requiring an order of magnitude less computing time, and it also outperforms a recent deep learning implementation for homography estimation.
[' Manolis C. Tsakiris', ' Laurent Kneip', ' Rene Vidal', ' Daniel P. Robinson', ' Zhihui Zhu', ' Yunchen Yang', 'Tianjiao Ding']
2020-06-01
null
null
null
cvpr-2020-6
['homography-estimation']
['computer-vision']
[-8.29591528e-02 -2.41090983e-01 -1.63838919e-02 5.37742786e-02 -7.38126934e-01 -9.24724638e-01 8.67296100e-01 -2.93142885e-01 -1.93855852e-01 1.71764150e-01 2.22592160e-01 -2.36449882e-01 -3.17219824e-01 -4.29493904e-01 -9.21903551e-01 -7.25765646e-01 3.46118689e-01 7.69019246e-01 -4.25981909e-01 -1.49144664e-01 5.15366256e-01 9.36193109e-01 -1.09880447e+00 -3.32894683e-01 2.49567971e-01 6.09146893e-01 -2.73531824e-01 8.44841123e-01 5.81906736e-01 4.76261936e-02 1.31213367e-01 -5.59741139e-01 7.27402925e-01 3.04525763e-01 -6.65484965e-01 6.65191531e-01 1.24060845e+00 -3.61157835e-01 -6.95017397e-01 9.63436007e-01 4.74555016e-01 6.60287514e-02 4.52289045e-01 -1.29217398e+00 -7.50220060e-01 -8.29178318e-02 -7.84706235e-01 -1.00063503e-01 7.58625150e-01 -8.20537880e-02 1.10392785e+00 -1.24640787e+00 7.01690197e-01 1.22441983e+00 6.38012409e-01 -1.10670142e-02 -1.50781250e+00 -5.59020340e-02 -3.30195427e-01 -1.60509139e-01 -1.47002423e+00 -5.85285425e-01 8.56444061e-01 -7.30869174e-01 1.03116739e+00 3.82509917e-01 7.10195363e-01 8.55633616e-01 3.87325913e-01 3.85253847e-01 8.27824116e-01 -5.69986343e-01 -1.08944029e-01 -2.09613219e-02 1.63975284e-01 8.14181328e-01 3.55772763e-01 1.33233845e-01 -3.25849503e-01 -4.59721208e-01 1.44128227e+00 1.98629111e-01 -4.96895373e-01 -1.25171208e+00 -1.68081141e+00 1.00058877e+00 -8.54916200e-02 -1.36301473e-01 -3.31191868e-01 7.75826871e-02 -2.39061993e-02 1.26437515e-01 2.73242593e-01 5.71400642e-01 2.66213734e-02 8.87971185e-03 -6.37745559e-01 2.05741569e-01 9.73923445e-01 1.37707782e+00 8.03612292e-01 1.21887580e-01 8.28723133e-01 6.11850977e-01 2.94886202e-01 8.93228769e-01 3.53082605e-02 -1.37148201e+00 3.99954498e-01 4.07004595e-01 2.12834522e-01 -1.52036691e+00 -5.02242565e-01 -2.50540972e-01 -9.60966587e-01 4.04543206e-02 4.65535969e-01 4.55523394e-02 -4.30923194e-01 1.20120692e+00 2.39871666e-01 5.51149137e-02 -2.56313205e-01 9.14674461e-01 2.76161253e-01 2.35598534e-01 -1.05493772e+00 -1.41437992e-01 1.30785215e+00 -7.41613030e-01 -3.46169502e-01 -1.78957909e-01 2.90360808e-01 -1.32600880e+00 4.12024558e-01 4.17634636e-01 -1.18649077e+00 -3.17385584e-01 -1.26899266e+00 -5.32038629e-01 4.42824373e-03 1.73154905e-01 6.46790087e-01 5.10845244e-01 -1.15137303e+00 6.21182859e-01 -7.46520698e-01 -3.93937945e-01 -1.58087835e-01 3.31611902e-01 -9.32052791e-01 -9.60072652e-02 -3.01466674e-01 8.83836091e-01 -6.27279654e-02 3.54082659e-02 -3.14052612e-01 -7.14204609e-01 -1.19307768e+00 -1.41021252e-01 3.04302275e-01 -1.03087676e+00 7.00505674e-01 -3.59623879e-02 -1.34248292e+00 1.38683188e+00 -1.89306557e-01 -3.25837940e-01 7.00474799e-01 -4.14017588e-01 -9.16359276e-02 3.23198020e-01 3.78052220e-02 4.34042931e-01 1.07440960e+00 -1.20884573e+00 -9.86105502e-02 -8.72233868e-01 2.18487233e-01 3.95093769e-01 2.67434597e-01 -1.51729912e-01 -6.84097826e-01 -2.75079489e-01 1.07392001e+00 -1.37377417e+00 -2.64030457e-01 -1.35314632e-02 -6.28135204e-01 1.52464211e-01 4.99009877e-01 -7.05560863e-01 7.29686081e-01 -1.72572577e+00 5.38224101e-01 3.57087970e-01 5.01489460e-01 -2.49757975e-01 1.53805584e-01 6.07188761e-01 -4.67367738e-01 -2.10183918e-01 5.05219735e-02 -4.02193457e-01 -7.44460523e-02 7.48894662e-02 -4.55401659e-01 1.40908659e+00 -2.43689373e-01 7.54015565e-01 -6.05554461e-01 4.95653935e-02 7.21604407e-01 6.21197402e-01 -4.25466299e-01 -2.84513328e-02 3.90521497e-01 4.07171458e-01 -1.48964629e-01 3.93967658e-01 1.06193101e+00 -2.32326776e-01 2.11100169e-02 -6.35033786e-01 -3.62311184e-01 -1.45198479e-01 -1.63756800e+00 1.76022589e+00 -3.87272567e-01 6.88627779e-01 1.39238700e-01 -8.57726157e-01 8.41806948e-01 3.32081020e-01 8.90919149e-01 -1.69993177e-01 4.03927207e-01 1.69999689e-01 -4.76223618e-01 -2.71392524e-01 4.47508901e-01 -1.73246711e-02 2.09378675e-01 3.20673645e-01 1.74452007e-01 -2.80687481e-01 -9.94150862e-02 7.35784620e-02 5.85560322e-01 1.35052532e-01 5.26055932e-01 -4.73593950e-01 5.96860588e-01 -2.67304808e-01 3.50051522e-01 3.89405638e-01 4.15735543e-02 1.04729486e+00 3.72782797e-01 -7.97516942e-01 -1.45147884e+00 -1.24046803e+00 -3.63998324e-01 1.83420449e-01 2.23963976e-01 -3.86588335e-01 -5.05218148e-01 -1.17499687e-04 2.39618704e-01 2.97098607e-01 -3.00271481e-01 9.43104252e-02 -8.19296062e-01 -3.42571437e-01 1.37613267e-01 4.97691423e-01 2.98811257e-01 -1.27445683e-02 -4.58696395e-01 -1.18678868e-01 -1.42073721e-01 -1.52852213e+00 -8.38385701e-01 -2.13744804e-01 -9.88175511e-01 -1.51296675e+00 -9.61412966e-01 -5.88248968e-01 6.71977222e-01 9.90606964e-01 1.03913569e+00 -3.37667853e-01 -1.80913135e-01 1.13509881e+00 1.84845850e-01 2.47792214e-01 -1.31493032e-01 -6.27704561e-01 4.41320807e-01 1.70512289e-01 4.74830806e-01 -8.62746119e-01 -5.52545667e-01 7.24500179e-01 -4.91645187e-01 -2.43545976e-02 6.68002442e-02 7.52976000e-01 7.10733056e-01 -4.07923609e-01 -4.87059832e-01 -5.03422379e-01 2.29888171e-01 -2.08817422e-01 -1.27065253e+00 1.23856127e-01 -4.33263808e-01 -5.38007170e-02 4.48705345e-01 5.00379801e-02 -5.74205339e-01 3.19379896e-01 2.91203767e-01 -1.11414897e+00 -8.42310712e-02 3.23562562e-01 -1.26818083e-02 -7.13953137e-01 5.68322599e-01 1.98611408e-01 1.70598492e-01 -4.96441573e-01 5.90219498e-01 8.14906508e-02 7.27586746e-01 -4.21510577e-01 1.17852354e+00 1.18182576e+00 5.62006891e-01 -1.25002849e+00 -3.43173295e-01 -1.28123820e+00 -1.42198122e+00 1.40226483e-01 9.70609486e-01 -1.01220667e+00 -9.53812480e-01 1.77932560e-01 -1.34836626e+00 4.60466534e-01 1.90040275e-01 8.62540901e-01 -1.18477154e+00 1.06591058e+00 -1.60712421e-01 -4.74346310e-01 -9.46367010e-02 -1.28960800e+00 1.49014950e+00 -3.29745710e-01 -6.18459880e-02 -1.32046163e+00 2.55175382e-01 8.56109142e-01 -2.06748083e-01 3.49707246e-01 5.22004902e-01 -3.04545939e-01 -9.63814974e-01 -5.52714288e-01 -1.34642124e-01 1.55099779e-01 -8.81463513e-02 1.83236897e-01 -9.10383224e-01 -6.03606582e-01 5.06012082e-01 1.89469367e-01 5.01187027e-01 5.24647236e-01 4.97196436e-01 -1.83290914e-01 -1.84891462e-01 1.41996205e+00 1.73827553e+00 -2.28145421e-02 4.96510863e-01 4.74711061e-01 1.09232020e+00 5.45849681e-01 1.80382311e-01 5.04144490e-01 3.15518945e-01 8.98004353e-01 4.17126596e-01 2.69110560e-01 2.73882776e-01 -2.18799233e-01 -8.06941912e-02 8.68429542e-01 -2.39665553e-01 2.25946456e-01 -9.05947983e-01 5.51429451e-01 -1.78287625e+00 -8.17387819e-01 -3.10204476e-01 2.73556042e+00 3.77262942e-02 -2.83961535e-01 2.62096571e-03 -2.70130094e-02 5.64196587e-01 2.48458341e-01 -3.25300723e-01 -1.19664408e-01 -4.11907703e-01 -2.64641225e-01 8.14930618e-01 8.44219625e-01 -1.28913939e+00 6.00412965e-01 6.37485647e+00 2.40577877e-01 -8.23650837e-01 -2.92775512e-01 -3.42753194e-02 2.55915821e-01 -2.72474349e-01 3.09221029e-01 -7.64425457e-01 -1.58985123e-01 3.41692090e-01 -2.19642594e-01 5.72631419e-01 8.12722445e-01 1.46439001e-02 8.78014863e-02 -1.44198334e+00 1.60651660e+00 5.35995185e-01 -1.66090739e+00 1.11626603e-01 5.21533668e-01 9.59071994e-01 2.32837766e-01 2.87687242e-01 -4.55423683e-01 3.00939474e-02 -7.35822856e-01 2.21023068e-01 3.26259315e-01 5.10880589e-01 -5.06796062e-01 3.95456731e-01 4.33559328e-01 -1.10582685e+00 2.83409923e-01 -9.11677361e-01 1.63029283e-01 3.60604048e-01 2.51445711e-01 -6.63751304e-01 7.56648719e-01 4.26227510e-01 9.24811780e-01 -2.44123936e-01 1.05920863e+00 2.55045593e-01 -1.67352259e-01 -4.49285448e-01 8.68117929e-01 2.88461447e-01 -1.04191446e+00 1.12979269e+00 7.43616223e-01 3.47551048e-01 1.78571597e-01 1.35500789e-01 8.25100005e-01 1.63553163e-01 2.07926352e-02 -1.14591062e+00 9.06200707e-02 2.34143272e-01 1.32428658e+00 -6.14806652e-01 -5.27396193e-03 -5.18465579e-01 8.78423095e-01 -6.80390224e-02 4.69817698e-01 -4.66417074e-01 4.91863303e-02 6.75391078e-01 6.74064681e-02 2.75633901e-01 -6.77044153e-01 -2.61477292e-01 -1.83254790e+00 2.50061393e-01 -8.83047462e-01 1.96289152e-01 -9.22768831e-01 -1.03104889e+00 2.42688283e-01 8.71199965e-02 -1.41190469e+00 -3.45064819e-01 -1.04439366e+00 -6.11972809e-01 9.29508328e-01 -1.13192749e+00 -1.34473610e+00 -3.12379509e-01 7.99760342e-01 4.04697180e-01 -2.83964396e-01 7.79697359e-01 -1.17051020e-01 -1.57611772e-01 2.58964002e-01 4.12915349e-01 1.31901950e-01 4.84658688e-01 -1.51210475e+00 6.92831278e-01 1.00374520e+00 5.17355859e-01 1.07986712e+00 7.61400461e-01 -3.49899650e-01 -2.24950767e+00 -5.54985583e-01 5.86722493e-01 -9.12878752e-01 8.70884001e-01 -2.34481856e-01 -6.50174439e-01 1.19435227e+00 2.34315664e-01 -2.18379442e-02 5.55853069e-01 3.15164477e-02 -6.35314047e-01 2.51588494e-01 -7.83521414e-01 6.61725521e-01 8.76489162e-01 -8.28488588e-01 -4.76082176e-01 7.12054253e-01 5.53733349e-01 -6.23417497e-01 -1.24641144e+00 1.83935314e-01 6.44053400e-01 -1.19688380e+00 1.50725150e+00 -4.76908237e-01 5.68604982e-03 -1.47447661e-01 -6.30269825e-01 -1.06071508e+00 -4.12118614e-01 -1.27443027e+00 9.03744102e-02 6.03674114e-01 -1.67366207e-01 -6.18461072e-01 9.80904281e-01 3.94812137e-01 -1.27358153e-01 -4.66652006e-01 -9.60449576e-01 -7.88625658e-01 3.17448452e-02 -3.06998253e-01 2.26735145e-01 1.14728916e+00 -1.15957953e-01 3.20824087e-01 -4.38907683e-01 5.14763951e-01 1.15759957e+00 4.21970785e-01 1.20621395e+00 -1.16015017e+00 -4.08745170e-01 -4.77742672e-01 -8.20402980e-01 -1.43072307e+00 1.73092842e-01 -7.86387742e-01 -4.21682388e-01 -1.11992764e+00 -9.69955884e-03 1.13881730e-01 3.58027548e-01 -3.27615261e-01 4.18923318e-01 1.62021831e-01 1.60493597e-01 5.27777612e-01 2.93395631e-02 3.43078107e-01 1.12154830e+00 -3.91601175e-02 -5.09230234e-03 2.49915957e-01 -3.71280581e-01 1.23134029e+00 3.45634878e-01 1.67507187e-01 -2.41919130e-01 -6.94005728e-01 4.74003613e-01 4.01955545e-01 5.63362420e-01 -8.45031023e-01 4.56200778e-01 -1.72653422e-01 1.98235497e-01 -8.85066152e-01 7.24403679e-01 -9.40157712e-01 2.84501970e-01 -1.61215752e-01 2.54353993e-02 5.45921504e-01 -9.90340672e-03 6.78556263e-01 -1.48233101e-01 -7.35740736e-02 7.11842537e-01 -3.20368826e-01 -5.12607157e-01 5.44070542e-01 2.25342825e-01 -1.26432851e-01 7.07013667e-01 -7.47528732e-01 -2.74010003e-01 -5.67509830e-01 -8.13227057e-01 7.75100589e-02 8.98948193e-01 1.95443183e-01 1.02429545e+00 -1.40792608e+00 -6.11330092e-01 6.39005125e-01 2.29603976e-01 -1.52642891e-01 1.43358052e-01 1.18671584e+00 -1.06934118e+00 8.75568449e-01 -1.95820078e-01 -1.00724125e+00 -1.44402230e+00 8.33859801e-01 4.65248108e-01 1.85910523e-01 -8.27088296e-01 6.61139846e-01 4.48643774e-01 -6.42848909e-01 1.81265905e-01 8.29472467e-02 -1.56981081e-01 -1.73910394e-01 2.43280768e-01 8.98169816e-01 -4.21925448e-03 -1.27430201e+00 -6.74716309e-02 1.36217594e+00 7.79534802e-02 -1.52572453e-01 1.28699636e+00 -4.21585411e-01 -4.66278762e-01 2.45493904e-01 1.78722072e+00 2.79523253e-01 -1.23301792e+00 -3.17639112e-01 -1.21190153e-01 -7.67900705e-01 -2.35338416e-02 1.67568862e-01 -7.18913794e-01 1.05720353e+00 1.47499204e-01 1.88681856e-01 6.05230927e-01 -6.44871220e-02 4.65577304e-01 6.41342342e-01 3.59169364e-01 -6.97800756e-01 -8.87544155e-02 5.07778823e-01 1.28495252e+00 -1.28735256e+00 3.68440300e-01 -6.48969650e-01 -3.31748545e-01 1.56071055e+00 2.57814437e-01 -6.28896713e-01 8.07945907e-01 -3.03574860e-01 -2.41512829e-03 -4.82376546e-01 -4.60821301e-01 2.19659016e-01 8.29516172e-01 7.06180334e-01 1.03899717e-01 -1.51884109e-01 2.50203937e-01 -5.44140697e-01 -4.32010055e-01 -5.99122286e-01 7.56432056e-01 5.33506930e-01 -3.39779645e-01 -8.30296218e-01 -7.23391771e-01 -1.32967755e-02 -6.02554642e-02 1.12939052e-01 -2.66196281e-01 9.68135715e-01 -3.41167092e-01 7.99996972e-01 1.14931241e-01 -2.99932986e-01 3.30678225e-01 -2.30017096e-01 8.69921803e-01 -4.57288951e-01 1.13007076e-01 6.88668609e-01 -7.63707384e-02 -9.03408170e-01 -4.38507587e-01 -7.34207988e-01 -5.59729576e-01 -5.78682244e-01 -3.21127683e-01 -1.40182197e-01 6.81126773e-01 1.01109147e+00 1.24181762e-01 -3.54193211e-01 7.85551429e-01 -9.71396208e-01 -9.40875232e-01 -4.15901244e-01 -7.10988581e-01 3.21579367e-01 7.30089903e-01 -7.07367301e-01 -6.46794140e-01 4.94626723e-02]
[8.061165809631348, -2.2950212955474854]
14da53e7-6f4c-47dc-89df-3782978f3fa1
multi-sensor-prognostics-using-an
1608.06154
null
http://arxiv.org/abs/1608.06154v1
http://arxiv.org/pdf/1608.06154v1.pdf
Multi-Sensor Prognostics using an Unsupervised Health Index based on LSTM Encoder-Decoder
Many approaches for estimation of Remaining Useful Life (RUL) of a machine, using its operational sensor data, make assumptions about how a system degrades or a fault evolves, e.g., exponential degradation. However, in many domains degradation may not follow a pattern. We propose a Long Short Term Memory based Encoder-Decoder (LSTM-ED) scheme to obtain an unsupervised health index (HI) for a system using multi-sensor time-series data. LSTM-ED is trained to reconstruct the time-series corresponding to healthy state of a system. The reconstruction error is used to compute HI which is then used for RUL estimation. We evaluate our approach on publicly available Turbofan Engine and Milling Machine datasets. We also present results on a real-world industry dataset from a pulverizer mill where we find significant correlation between LSTM-ED based HI and maintenance costs.
['Anusha Ramakrishnan', 'Vishnu Tv', 'Lovekesh Vig', 'Gaurangi Anand', 'Pankaj Malhotra', 'Gautam Shroff', 'Puneet Agarwal']
2016-08-22
null
null
null
null
['exponential-degradation']
['time-series']
[ 1.76034465e-01 -1.14718474e-01 -1.91125162e-02 -2.31336534e-01 -4.91954535e-01 6.24258891e-02 1.92771465e-01 2.44416088e-01 2.14094579e-01 7.39385843e-01 -1.63991258e-01 -3.45793754e-01 -1.83223143e-01 -8.01094413e-01 -9.35938120e-01 -8.21716130e-01 -2.13837400e-01 3.61797363e-01 -2.46861428e-02 4.01261374e-02 2.92120129e-01 5.32654107e-01 -1.52896166e+00 3.94351125e-01 3.15049320e-01 1.50648785e+00 4.33695734e-01 9.95331168e-01 4.92004216e-01 1.14327919e+00 -1.01410270e+00 2.39829108e-01 -2.13081509e-01 -2.72106886e-01 -7.71005988e-01 3.78912896e-01 -3.90276343e-01 -5.02040088e-01 -6.62537158e-01 6.25111818e-01 3.06459516e-01 6.15105266e-04 7.91906357e-01 -1.09150565e+00 -4.27903324e-01 3.56066465e-01 1.06228575e-01 3.00321341e-01 1.05497815e-01 2.43533939e-01 4.21810389e-01 -5.18943429e-01 1.90149188e-01 1.03970659e+00 6.61033750e-01 -6.78790882e-02 -7.65046716e-01 -1.06119579e-02 -2.07841501e-01 3.67395699e-01 -1.13151324e+00 -2.72810906e-01 7.16372550e-01 -4.06459957e-01 1.48788607e+00 -1.26281157e-01 3.92669797e-01 1.13119614e+00 1.55872929e+00 5.70248246e-01 9.10853267e-01 -2.18616292e-01 2.94336528e-01 -4.22290564e-01 -4.90534045e-02 6.59318388e-01 -1.44413169e-02 4.88412261e-01 -3.17003667e-01 3.12869102e-02 8.45473289e-01 3.21051955e-01 2.61491016e-02 1.85515270e-01 -1.13951600e+00 2.71733820e-01 8.16344377e-03 3.70315582e-01 -8.32832396e-01 3.46843779e-01 7.49832094e-01 1.11291516e+00 6.29883528e-01 4.28366721e-01 -8.43257129e-01 -3.39122415e-01 -1.04516876e+00 -1.94484800e-01 8.88419449e-01 6.37078881e-01 3.90746981e-01 5.61733067e-01 -3.94179970e-01 7.42872536e-01 8.79504085e-02 5.05997419e-01 7.30112791e-01 -1.00135517e+00 -7.61401430e-02 2.24476248e-01 2.78652251e-01 -5.64098775e-01 -3.11613500e-01 -5.27522743e-01 -8.39911520e-01 2.97443092e-01 -2.90913552e-01 -2.99788415e-01 -1.11406243e+00 1.16219199e+00 -2.89061189e-01 1.83688670e-01 1.90318748e-01 4.18201894e-01 -1.98644828e-02 7.99941599e-01 -3.13809484e-01 -5.12898326e-01 8.72527242e-01 -6.96526825e-01 -1.02842009e+00 3.84822488e-02 4.15594041e-01 -3.03424895e-01 7.01957822e-01 1.01757574e+00 -1.06171930e+00 -7.68894672e-01 -1.43174493e+00 4.04337376e-01 -3.21641773e-01 3.35939854e-01 5.49422316e-02 1.07513554e-01 -7.41299272e-01 1.43419969e+00 -1.31198192e+00 -3.69294137e-01 2.01471280e-02 3.24812382e-01 -1.77664012e-01 -1.91519782e-02 -1.07939672e+00 1.38652503e+00 5.24309635e-01 2.39201039e-01 -2.08221483e+00 -2.26799443e-01 -6.94667518e-01 -1.01166762e-01 2.53222406e-01 -4.76486921e-01 1.51445782e+00 -5.41101277e-01 -1.63562083e+00 1.16989203e-01 1.31819770e-01 -9.50165153e-01 1.24222934e-01 -3.53607863e-01 -1.05199397e+00 -3.57472412e-02 -4.72813606e-01 -9.74012539e-02 1.34210873e+00 -1.06986809e+00 -5.83290637e-01 -1.28825456e-01 -3.04258198e-01 -3.21430355e-01 -3.51460278e-01 -1.57507524e-01 1.42326280e-01 -4.27271068e-01 2.04761662e-02 -6.64460003e-01 2.54543405e-02 -4.14658695e-01 -4.59160864e-01 -1.85808465e-01 1.20533442e+00 -1.23616028e+00 1.41490233e+00 -1.86245477e+00 3.50150801e-02 -6.52488917e-02 -2.13315085e-01 5.77222742e-02 5.63511439e-02 9.69913423e-01 -1.27868578e-01 -4.01591241e-01 -4.11701530e-01 -2.59590268e-01 9.23042186e-04 7.46680319e-01 -3.88370186e-01 6.92710996e-01 5.32408059e-01 5.97222626e-01 -6.08104169e-01 -2.57698268e-01 3.88303757e-01 3.60408455e-01 3.72505754e-01 5.20774007e-01 -3.33701819e-01 1.93078488e-01 -1.10597171e-01 8.85786414e-01 9.24074203e-02 -2.01792945e-03 -8.15195292e-02 -2.26066276e-01 5.21374755e-02 -2.95053385e-02 -5.38403451e-01 1.69533193e+00 -9.39657271e-01 6.55357420e-01 -3.09093744e-01 -1.27311337e+00 1.01458013e+00 7.97425270e-01 5.75347960e-01 -8.62328112e-01 3.88833404e-01 1.88943490e-01 -1.80596337e-01 -7.89946496e-01 3.78932834e-01 -3.39616477e-01 -1.12008102e-01 1.90328240e-01 2.97973037e-01 -5.16232848e-02 -5.49920164e-02 -4.32907283e-01 1.75655186e+00 1.38385028e-01 -2.29108222e-02 -1.86316058e-01 3.18944007e-01 -2.31045380e-01 4.46202338e-01 2.31059879e-01 -2.81069577e-02 2.34097078e-01 4.22345430e-01 -3.38332415e-01 -1.34857023e+00 -1.03821743e+00 -7.25752264e-02 4.91853297e-01 -4.03257787e-01 -4.62985002e-02 -5.30349970e-01 -3.50099504e-01 1.82388693e-01 9.35042202e-01 -5.44900060e-01 -7.95220375e-01 -4.53414142e-01 -3.65427703e-01 4.65494633e-01 5.65314889e-01 1.36328757e-01 -1.20997334e+00 -1.08115458e+00 6.74788892e-01 2.52686769e-01 -8.66999447e-01 -1.55229136e-04 7.80323386e-01 -1.27779996e+00 -9.25476253e-01 -3.73988152e-01 -5.13696611e-01 5.44985533e-01 -4.55123484e-01 1.20252645e+00 -1.02675475e-01 -3.81129861e-01 2.09778130e-01 -1.27278820e-01 -3.47385138e-01 -7.22438037e-01 -3.77690017e-01 3.64047676e-01 -3.39212030e-01 -2.87250541e-02 -5.01047969e-01 -3.39248419e-01 2.01614127e-01 -9.15936172e-01 -3.59459996e-01 8.49287808e-01 8.77101123e-01 7.51707137e-01 1.06016600e+00 8.93961132e-01 -4.59030360e-01 7.91530013e-01 -7.23758042e-01 -3.57142359e-01 3.37029368e-01 -1.28204155e+00 4.41985667e-01 9.18326020e-01 -5.57828963e-01 -9.64122176e-01 -1.24600746e-01 3.45197599e-03 -1.00878477e+00 5.98157567e-05 9.28117871e-01 5.93917929e-02 4.29990619e-01 6.98973238e-02 2.49343917e-01 -7.65232220e-02 -7.59965599e-01 -2.00792581e-01 9.38897312e-01 1.02403283e+00 -5.74285626e-01 3.52486730e-01 2.01100707e-02 1.36820674e-01 -7.20311761e-01 -5.85796118e-01 -4.19548862e-02 -4.71313745e-01 -5.86406291e-01 4.59471434e-01 -8.03700566e-01 -6.84757411e-01 7.20656991e-01 -8.89692962e-01 -5.95522046e-01 -4.55427408e-01 3.94218415e-01 -7.29664385e-01 1.78669259e-01 -1.15999448e+00 -9.93349195e-01 -5.75638771e-01 -8.87358606e-01 1.14719331e+00 -2.81786829e-01 3.90960127e-02 -1.17334259e+00 4.15054671e-02 -9.64423046e-02 4.83198762e-01 8.15560639e-01 1.06665158e+00 -1.68360367e-01 7.58767128e-02 -8.79127860e-01 3.47524583e-01 1.22724903e+00 3.34080696e-01 3.96904498e-02 -7.75814116e-01 -8.05485070e-01 5.08494854e-01 -2.28480563e-01 7.40977883e-01 3.95097524e-01 1.38290417e+00 -3.79269004e-01 -2.01540262e-01 1.10685796e-01 1.49595571e+00 4.20871168e-01 7.40915716e-01 1.72267817e-02 5.04683018e-01 2.00605214e-01 1.13762057e+00 6.82095528e-01 -3.39007415e-02 1.62059397e-01 6.51937664e-01 2.38721833e-01 1.24222925e-02 -2.44265378e-01 1.12584448e+00 1.37693226e+00 2.20819443e-01 -5.13798296e-01 -8.45152557e-01 5.97933769e-01 -1.47460544e+00 -4.34030026e-01 -4.92063388e-02 2.38430882e+00 5.75238526e-01 6.02086782e-01 -2.39725530e-01 7.73790240e-01 5.72813511e-01 -1.73479974e-01 -1.04741573e+00 -6.28992677e-01 9.01427865e-02 2.32690647e-01 8.70345056e-01 1.71458602e-01 -8.22954893e-01 4.49792653e-01 6.92538071e+00 5.53355992e-01 -1.17603338e+00 3.78095865e-01 5.21996677e-01 -7.14701563e-02 2.59980470e-01 -1.21540844e-01 -2.32260704e-01 4.88267362e-01 2.11439323e+00 -1.33409813e-01 3.40502530e-01 5.17102420e-01 3.45162183e-01 -2.15225086e-01 -1.39222133e+00 7.70954728e-01 -3.37196738e-02 -6.68459356e-01 -3.26342762e-01 1.29142106e-01 4.52118933e-01 1.59895778e-01 1.37489336e-02 3.09157610e-01 2.37280771e-01 -1.00100791e+00 7.42051899e-01 1.15122211e+00 9.84651208e-01 -9.49261367e-01 1.00535047e+00 4.93925720e-01 -8.93863857e-01 -5.09346366e-01 -3.93319368e-01 -1.61752433e-01 3.42596501e-01 1.09535909e+00 -6.95253909e-01 6.78131282e-01 5.81737816e-01 7.85933852e-01 -4.50446010e-01 6.56048119e-01 1.31228462e-01 7.72937357e-01 -2.83474997e-02 4.67981219e-01 -9.51015502e-02 1.54090837e-01 4.12335306e-01 7.34308004e-01 7.53155828e-01 -4.78287607e-01 -2.89052017e-02 6.43541873e-01 1.11090101e-01 -8.55310857e-01 -5.61008990e-01 -6.37079954e-01 2.92829692e-01 8.15024197e-01 -4.58544075e-01 -3.96072298e-01 -1.45041689e-01 1.31193733e+00 -2.22181648e-01 3.05357605e-01 -8.77265632e-01 -4.14973676e-01 3.02379072e-01 1.13309570e-01 2.13872895e-01 -5.80888867e-01 1.15588225e-01 -5.50129235e-01 1.35253370e-02 -6.92908168e-01 1.47268876e-01 -9.80206370e-01 -1.16532409e+00 6.04706585e-01 -5.25958426e-02 -1.37221801e+00 -5.55621803e-01 -5.80119073e-01 -5.20049810e-01 6.10633194e-01 -1.31724799e+00 -8.54763925e-01 -5.99878691e-02 2.98508823e-01 9.45343196e-01 -2.04732582e-01 6.10108614e-01 3.83077264e-01 -9.78008270e-01 5.20980507e-02 4.16277766e-01 -5.06477714e-01 5.44415116e-01 -1.22674561e+00 4.51670438e-01 8.97273302e-01 -4.08809304e-01 6.15266338e-02 1.26255620e+00 -9.46366608e-01 -1.81636536e+00 -1.51150525e+00 5.17905653e-01 -4.29896533e-01 7.22645402e-01 6.10716641e-02 -8.98874998e-01 7.81775355e-01 3.73957664e-01 3.73556055e-02 2.02728465e-01 -5.99889398e-01 3.34634691e-01 -2.14788675e-01 -1.24662995e+00 -3.19226980e-01 4.79313076e-01 -8.39307249e-01 -4.31993186e-01 4.43397343e-01 7.05482125e-01 -2.25710392e-01 -1.76074529e+00 5.23055971e-01 3.46683025e-01 -5.70040166e-01 6.57434940e-01 -3.97216797e-01 4.66616422e-01 -2.68463761e-01 -2.63994604e-01 -1.56714749e+00 -2.71118730e-01 -2.89817810e-01 -1.06517100e+00 1.01516795e+00 1.06257692e-01 -5.02749026e-01 1.75542623e-01 2.04358101e-01 -8.38105619e-01 -1.02573681e+00 -9.30741906e-01 -1.31374657e+00 -3.50349173e-02 -3.83444577e-01 6.86039507e-01 4.24472392e-01 -1.46441907e-01 4.88240421e-02 -5.92968404e-01 4.83314812e-01 3.99193108e-01 -2.11547941e-01 4.96981330e-02 -1.29618502e+00 -3.85199279e-01 2.70924866e-01 -3.74138504e-01 -4.60423529e-01 2.66862065e-01 -3.33009869e-01 4.12546456e-01 -1.64321649e+00 -5.18086627e-02 6.48365319e-02 -1.05664980e+00 4.08098966e-01 4.72643882e-01 -2.47131005e-01 -4.90046144e-01 2.72631377e-01 -5.22403240e-01 7.46507585e-01 8.70226800e-01 -2.53228009e-01 2.02345103e-01 -4.40222211e-02 1.89454705e-01 2.56167114e-01 8.85874331e-01 -4.89870310e-01 -3.30826133e-01 -4.20946002e-01 -7.19027827e-03 8.35775852e-01 4.11067188e-01 -1.66376364e+00 9.69315842e-02 7.75481164e-02 5.24023712e-01 -8.22402537e-01 4.52329338e-01 -9.59069967e-01 5.88949502e-01 1.03666675e+00 -8.64916295e-02 5.65430045e-01 2.43299473e-02 7.45191038e-01 -4.52439368e-01 -3.68904248e-02 5.40651858e-01 1.45859748e-01 -4.87058759e-01 2.71440089e-01 -6.35081291e-01 -7.78228223e-01 8.45489502e-01 -4.93185408e-02 -2.08101362e-01 -2.10136205e-01 -6.77424788e-01 2.03227237e-01 4.08353060e-01 6.18812323e-01 8.57192576e-01 -1.31705391e+00 -5.98805547e-01 1.64395139e-01 9.92451459e-02 -4.63326484e-01 2.84857512e-01 8.41893375e-01 -3.57443601e-01 6.64851785e-01 -3.50047588e-01 -4.11163926e-01 -9.35610950e-01 9.20446455e-01 4.64490503e-01 -2.40495518e-01 -7.33600914e-01 3.09750229e-01 -6.73069835e-01 1.94496542e-01 1.31513193e-01 -8.77642155e-01 1.05115622e-01 -2.12641910e-01 1.23006128e-01 6.41598105e-01 6.59364641e-01 -4.54851419e-01 -8.57866183e-02 -2.07334384e-01 2.99601316e-01 1.29784137e-01 1.42221665e+00 -1.38050228e-01 -1.66517049e-01 1.39734244e+00 1.24846399e+00 -1.12393594e+00 -1.50216043e+00 -8.94664600e-02 2.70636737e-01 -4.59466055e-02 7.10708976e-01 -8.42904985e-01 -1.25816882e+00 7.61243880e-01 1.17308104e+00 4.08022821e-01 1.46074331e+00 -2.51368612e-01 1.29891944e+00 6.34797990e-01 5.85229397e-01 -1.46661603e+00 1.46889016e-01 3.87343079e-01 9.54823732e-01 -8.40873361e-01 -2.14188218e-01 5.52587092e-01 -4.13891107e-01 1.21660411e+00 2.34512195e-01 -2.03475639e-01 7.66410887e-01 4.96579379e-01 -3.79736394e-01 -4.04993802e-01 -1.35357356e+00 2.81445861e-01 -1.82902068e-01 3.46290469e-01 3.11221659e-01 3.17580253e-01 9.12736729e-02 4.07295704e-01 8.67120028e-02 2.91884869e-01 5.94056189e-01 1.32606792e+00 -4.64491785e-01 -9.33498859e-01 -3.22153747e-01 9.48670447e-01 -5.30701876e-01 3.54853004e-01 3.84117477e-02 2.34606549e-01 7.56972295e-05 1.18460870e+00 1.40406653e-01 -8.72033834e-01 4.96433228e-01 1.97072625e-01 5.63944280e-01 -4.66094673e-01 -3.48550647e-01 -2.42859632e-01 2.15022847e-01 -6.58184350e-01 -3.38338548e-03 -7.19435036e-01 -1.25744939e+00 -2.64736265e-01 -3.76128823e-01 -2.58450825e-02 8.61329079e-01 1.11947370e+00 3.41915011e-01 1.41236758e+00 9.17718470e-01 -7.25646317e-01 -8.07613611e-01 -1.39203227e+00 -1.13272727e+00 9.49755087e-02 6.98494494e-01 -8.64897490e-01 -2.22470179e-01 2.28068009e-01]
[6.803615093231201, 2.5282604694366455]
11782c4a-2a13-49d2-9799-370821e61030
the-environmental-discontinuity-hypothesis
2205.15931
null
https://arxiv.org/abs/2205.15931v1
https://arxiv.org/pdf/2205.15931v1.pdf
The Environmental Discontinuity Hypothesis for Down-Sampled Lexicase Selection
Down-sampling training data has long been shown to improve the generalization performance of a wide range of machine learning systems. Recently, down-sampling has proved effective in genetic programming (GP) runs that utilize the lexicase parent selection technique. Although this down-sampling procedure has been shown to significantly improve performance across a variety of problems, it does not seem to do so due to encouraging adaptability through environmental change. We hypothesize that the random sampling that is performed every generation causes discontinuities that result in the population being unable to adapt to the shifting environment. We investigate modifications to down-sampled lexicase selection in hopes of promoting incremental environmental change to scaffold evolution by reducing the amount of jarring discontinuities between the environments of successive generations. In our empirical studies, we find that forcing incremental environmental change is not significantly better for evolving solutions to program synthesis problems than simple random down-sampling. In response to this, we attempt to exacerbate the hypothesized prevalence of discontinuities by using only disjoint down-samples to see if it hinders performance. We find that this also does not significantly differ from the performance of regular random down-sampling. These negative results raise new questions about the ways in which the composition of sub-samples, which may include synonymous cases, may be expected to influence the performance of machine learning systems that use down-sampling.
['Lee Spector', 'Thomas Helmuth', 'Ryan Boldi']
2022-05-31
null
null
null
null
['program-synthesis']
['computer-code']
[ 7.96236575e-01 8.31519365e-02 -1.79388955e-01 -2.72932470e-01 -1.07909657e-01 -4.74366218e-01 4.79484022e-01 2.82801598e-01 -4.85932320e-01 9.85593438e-01 1.71011224e-01 -6.72394514e-01 -8.40657502e-02 -1.14364183e+00 -7.80146420e-01 -6.82475984e-01 1.20192751e-01 1.86415300e-01 2.16195017e-01 -3.63283068e-01 6.59288108e-01 3.20112407e-01 -2.03456426e+00 1.65359676e-01 1.06995893e+00 -1.81368008e-01 1.30465940e-01 3.67424101e-01 -1.29032567e-01 1.69777721e-01 -9.97209907e-01 -5.54535575e-02 1.58460006e-01 -9.99781787e-01 -6.07736468e-01 4.22816426e-02 2.12706730e-01 6.31264672e-02 3.74893069e-01 8.07266414e-01 5.23101151e-01 9.51836631e-02 4.99667048e-01 -9.57525969e-01 -5.93425095e-01 9.31800008e-01 -6.51459575e-01 3.35531443e-01 2.75046051e-01 3.26789618e-01 7.06562877e-01 -1.35845602e-01 6.89492226e-01 1.30770314e+00 8.23726773e-01 4.20160502e-01 -1.60148537e+00 -5.05937874e-01 1.64193869e-01 -4.55465049e-01 -1.17086995e+00 -3.42357785e-01 5.57196617e-01 -2.20007256e-01 1.13333726e+00 4.56756890e-01 1.16145706e+00 6.50090754e-01 4.19181973e-01 2.53451407e-01 1.06333864e+00 -7.70476520e-01 4.01822299e-01 1.96647123e-01 -2.95509756e-01 5.21340728e-01 9.28323865e-01 1.94101110e-01 -5.04100502e-01 -4.57519829e-01 3.72602493e-01 -6.18336082e-01 -3.11608464e-01 -4.85103756e-01 -8.24171543e-01 8.55602264e-01 1.72258139e-01 4.67579365e-01 -1.45985886e-01 8.48202184e-02 4.02100533e-01 4.14976001e-01 3.80868942e-01 1.51267266e+00 -7.53816903e-01 -4.21180040e-01 -7.55691111e-01 3.65418047e-01 6.55503869e-01 4.60814595e-01 5.47251999e-01 3.33731204e-01 2.16107696e-01 8.32929075e-01 -1.90488957e-02 -5.05821034e-02 7.82222450e-01 -7.55640805e-01 2.08268017e-01 7.50687480e-01 -5.90357147e-02 -1.00850916e+00 -2.48677254e-01 -6.50685728e-01 -8.29323158e-02 2.99066305e-01 4.40986246e-01 -4.13256675e-01 -6.99877858e-01 2.07868838e+00 2.90421188e-01 -3.28548372e-01 -2.38157753e-02 3.15270185e-01 1.82742793e-02 6.50230110e-01 1.00842714e-01 -4.14810508e-01 6.88142836e-01 -5.99508166e-01 -2.43303910e-01 -4.52364117e-01 1.16431403e+00 -5.12264848e-01 1.32220531e+00 2.40973100e-01 -9.06433940e-01 -3.65571588e-01 -1.41972816e+00 4.84185249e-01 -4.09001857e-01 -4.58023578e-01 7.58764803e-01 1.36031485e+00 -8.16044867e-01 8.50026727e-01 -7.46547937e-01 -6.45218790e-01 2.03299448e-01 2.12964445e-01 -8.16342384e-02 1.30947828e-01 -8.74553502e-01 9.99987364e-01 4.66248751e-01 -1.48011431e-01 -4.34859246e-02 -6.54712677e-01 -6.86858654e-01 1.40746579e-01 4.10176784e-01 -7.97944546e-01 6.67378962e-01 -1.28179538e+00 -1.37436855e+00 4.26994115e-01 -1.56653643e-01 -1.97124943e-01 3.30645770e-01 2.43005857e-01 -1.33909002e-01 -4.97279227e-01 8.19533225e-03 7.73729980e-01 5.48958242e-01 -1.06212008e+00 -5.04278898e-01 -3.88219208e-01 -9.27624553e-02 3.94270658e-01 -5.23381174e-01 -1.35071361e-02 1.29704237e-01 -7.23242521e-01 2.17604060e-02 -1.24208009e+00 -2.64284611e-01 -6.15305722e-01 -2.97634155e-02 -1.11916810e-01 4.66699153e-01 -1.03250407e-01 1.25480795e+00 -2.12487221e+00 -4.25326712e-02 1.97949052e-01 -4.30133581e-01 2.02440992e-01 -3.26452583e-01 5.32036364e-01 -1.59349024e-01 5.67008853e-01 -3.84686679e-01 2.68237531e-01 -1.65781975e-01 1.40175819e-01 -8.61877799e-02 2.82883823e-01 5.06308198e-01 5.70402503e-01 -8.77206802e-01 -1.04605220e-01 -3.53692830e-01 2.73597956e-01 -1.08467853e+00 -4.44556445e-01 -3.89345944e-01 9.10681859e-02 -2.13491902e-01 4.47751999e-01 2.44361475e-01 2.66919062e-02 3.93216938e-01 6.68554008e-01 -2.52111524e-01 6.09403789e-01 -1.01007426e+00 1.30179524e+00 -1.82023436e-01 7.55586445e-01 -5.51841199e-01 -7.71703124e-01 9.85786855e-01 -3.65749747e-02 4.83905384e-03 -6.96355581e-01 -3.33300494e-02 3.13606501e-01 7.24425554e-01 -3.87388527e-01 7.46630907e-01 -3.54889482e-01 4.03959714e-02 6.69625759e-01 -5.89475811e-01 -5.90497792e-01 5.47972858e-01 -3.40768218e-01 1.10895240e+00 3.56131971e-01 1.88754290e-01 -4.05362129e-01 1.22460648e-01 4.00397867e-01 1.00914991e+00 8.82434249e-01 9.57993343e-02 3.43593746e-01 6.91764176e-01 -4.43627313e-02 -1.04413342e+00 -5.32682121e-01 -2.19370261e-01 1.03433895e+00 -1.13747217e-01 -5.08054256e-01 -6.39682829e-01 -4.79469597e-01 1.24600947e-01 1.27748466e+00 -6.11928940e-01 -6.77489519e-01 -4.93464202e-01 -1.28671193e+00 7.48194754e-01 3.75200897e-01 2.24956885e-01 -8.78919721e-01 -1.31914806e+00 1.51543155e-01 4.00172472e-01 -1.69987470e-01 -1.25853166e-01 5.26033998e-01 -1.21812630e+00 -7.86814868e-01 -5.84694386e-01 -6.54841185e-01 9.54216659e-01 1.90864690e-02 1.13429594e+00 5.07794380e-01 -2.52092272e-01 3.15492675e-02 -3.45100701e-01 -5.27184606e-01 -8.27351272e-01 3.46790969e-01 -1.18884452e-01 -7.37141430e-01 3.46585006e-01 -6.54303551e-01 -9.02230442e-02 3.24859351e-01 -9.95263696e-01 -7.07097873e-02 4.29568052e-01 1.08575702e+00 1.13969818e-01 6.17970407e-01 8.87887180e-01 -1.10615385e+00 1.02666140e+00 -4.26951200e-01 -4.95370865e-01 2.96205968e-01 -9.93942380e-01 3.57978046e-01 4.81802702e-01 -7.17775524e-01 -8.77736986e-01 -3.94927859e-01 5.52468486e-02 3.64552259e-01 -4.83703129e-02 7.69947052e-01 4.62978147e-02 -8.13913867e-02 8.45225990e-01 -3.21354643e-02 1.71117812e-01 -1.99003369e-01 -6.64011613e-02 4.36065376e-01 -1.70239523e-01 -7.89407432e-01 6.32710159e-01 -1.00156501e-01 -7.01851770e-02 -7.88255990e-01 -1.38523012e-01 3.71021360e-01 -1.33258164e-01 8.20216164e-02 5.06329656e-01 -4.41509217e-01 -1.47336453e-01 2.02765524e-01 -4.85880464e-01 -6.62052810e-01 -4.56338972e-01 2.85002321e-01 -2.93934017e-01 -8.74829479e-03 -1.21135712e-01 -6.35758996e-01 2.11832136e-01 -1.30312288e+00 3.79013509e-01 4.97942358e-01 -7.56897390e-01 -7.41533756e-01 1.03054307e-01 1.29369467e-01 6.02476776e-01 3.42901289e-01 1.42050016e+00 -5.67066491e-01 -1.18189909e-01 -9.52775031e-02 4.84964311e-01 9.93323475e-02 4.63140279e-01 3.46858829e-01 -4.38694805e-01 -4.03763741e-01 -5.58896177e-02 -1.16659209e-01 6.82277918e-01 1.98655710e-01 7.46048093e-01 4.88150679e-02 -4.33816254e-01 4.47650492e-01 1.26432157e+00 7.98272789e-01 5.83925545e-01 8.98406923e-01 6.00359365e-02 8.72767031e-01 6.50539100e-01 2.20127478e-01 -6.91672936e-02 4.80996549e-01 -2.34426513e-01 1.17503315e-01 1.29307553e-01 -1.97929040e-01 3.83590996e-01 2.05712363e-01 2.42980897e-01 -3.33000541e-01 -1.17234612e+00 6.10032320e-01 -1.34692037e+00 -9.12605822e-01 1.96028158e-01 2.19842339e+00 9.91237581e-01 5.48629642e-01 2.78656572e-01 2.26767689e-01 7.09962606e-01 -3.93850505e-02 -5.98993540e-01 -9.21109200e-01 -1.88998744e-01 2.29010567e-01 4.25989717e-01 3.70933175e-01 -2.78321356e-01 6.20839715e-01 6.88894892e+00 2.43762881e-01 -1.32448041e+00 -4.87389803e-01 7.66875327e-01 -2.90299952e-01 -5.92016459e-01 2.92682558e-01 -6.43440187e-01 5.88192701e-01 9.81877148e-01 -4.84976619e-01 3.66370708e-01 4.88626301e-01 2.06983760e-01 -5.30769885e-01 -1.10517728e+00 -4.95813265e-02 1.08173350e-02 -1.26190090e+00 -4.01847474e-02 1.85593635e-01 9.20929849e-01 -3.74436349e-01 2.18716174e-01 2.07219929e-01 3.00280511e-01 -1.03665161e+00 6.81397557e-01 3.44740003e-02 2.30748087e-01 -1.02104247e+00 4.87984687e-01 4.22197163e-01 -3.76396865e-01 -2.91108161e-01 -1.53133452e-01 -4.23795581e-01 -1.62364781e-01 3.10243785e-01 -1.10035992e+00 1.09307192e-01 4.93514210e-01 -3.27505469e-02 -9.90873635e-01 1.08894932e+00 -8.08846503e-02 8.20475399e-01 -3.48118216e-01 -3.50988507e-01 1.31521806e-01 -9.69287753e-02 6.93046927e-01 9.42368627e-01 4.51060981e-01 -2.57571191e-01 -5.38983941e-01 7.81447053e-01 4.14771855e-01 -1.65146273e-02 -8.74362707e-01 -4.90250885e-01 6.89172506e-01 2.20459953e-01 -1.05486214e+00 -9.29164737e-02 -7.86045194e-02 6.14070535e-01 -1.22704037e-01 2.49094740e-01 -6.20125353e-01 -6.12050831e-01 4.95232016e-01 3.01764071e-01 2.91186064e-01 -8.96891728e-02 -5.54305792e-01 -6.52508795e-01 -8.84812325e-02 -1.35301840e+00 1.70346528e-01 -6.25266969e-01 -7.75495172e-01 2.98804581e-01 7.66566992e-02 -4.07706082e-01 -3.52560759e-01 4.98259887e-02 -5.79210997e-01 1.00088358e+00 -1.02979231e+00 -2.42570803e-01 2.24100530e-01 -2.63260871e-01 6.17170215e-01 -4.86450046e-02 6.49231851e-01 -1.36219949e-01 -7.63015568e-01 7.43842483e-01 1.31891698e-01 -5.06971717e-01 7.42530048e-01 -1.08827507e+00 6.29172742e-01 9.85169172e-01 -1.25761539e-01 1.25930834e+00 1.08480763e+00 -1.11512411e+00 -1.20256376e+00 -7.49398828e-01 6.63185954e-01 -1.71455488e-01 2.26061717e-01 -7.32304230e-02 -9.08074379e-01 5.03327906e-01 1.37251645e-01 -6.25122666e-01 6.48748040e-01 4.38356280e-01 -2.20267624e-01 8.78811851e-02 -1.35708737e+00 9.32884276e-01 1.00734735e+00 1.25086447e-02 -5.45472562e-01 -2.44721174e-01 7.22389400e-01 -1.16824433e-01 -6.80927157e-01 5.96346855e-01 4.16572124e-01 -9.47158396e-01 7.12596714e-01 -6.53906047e-01 6.77668214e-01 -2.74310917e-01 -8.86149704e-03 -1.59512126e+00 -4.36597168e-01 -5.19642889e-01 6.03439331e-01 1.42032826e+00 8.64328265e-01 -1.25585461e+00 9.83642578e-01 7.85446942e-01 -1.49850816e-01 -8.98358941e-01 -5.38554251e-01 -8.79943013e-01 4.71506745e-01 3.22948210e-02 9.56621528e-01 1.16166854e+00 1.44186914e-01 1.64784029e-01 1.56096637e-01 -9.49148312e-02 1.18398309e-01 -8.42318609e-02 7.00117290e-01 -9.67142224e-01 -7.04358280e-01 -7.68632591e-01 -3.29768598e-01 -4.47995991e-01 -1.72349259e-01 -5.86213887e-01 7.22445622e-02 -1.10727239e+00 1.57560110e-02 -8.35031927e-01 5.42003289e-02 3.13840300e-01 -4.46419030e-01 -7.76629448e-02 1.89414337e-01 -1.18257143e-01 1.91271618e-01 1.96309045e-01 1.02589631e+00 1.00158758e-01 -7.24520385e-01 -7.78623521e-02 -1.19363141e+00 5.53882301e-01 9.56479728e-01 -5.81400990e-01 -7.77367055e-01 -4.22936648e-01 7.03953743e-01 -2.62528718e-01 -2.88851738e-01 -9.37825918e-01 -1.95679039e-01 -3.98125440e-01 3.45305145e-01 -1.46621224e-02 -3.90035361e-02 -3.73517036e-01 5.22686005e-01 8.76179039e-01 -7.27605164e-01 5.25043786e-01 6.59912527e-01 4.07575369e-01 1.32845998e-01 -7.25660086e-01 5.98649204e-01 -1.69070244e-01 -2.85677642e-01 -6.58811867e-01 -8.64053011e-01 1.81255694e-02 1.04411018e+00 -7.36364365e-01 -2.86824465e-01 1.56683668e-01 -5.87909818e-02 5.19842356e-02 1.28013313e+00 4.12334234e-01 1.62038784e-02 -6.71468318e-01 -4.69177485e-01 2.97045320e-01 -8.02587196e-02 -1.83708534e-01 -2.15443358e-01 5.84479809e-01 -5.88911772e-01 2.65117556e-01 -2.66037494e-01 -3.42598647e-01 -1.44660962e+00 3.34834516e-01 3.46549183e-01 1.46987632e-01 -5.35551608e-01 1.11494160e+00 -3.50987822e-01 -3.27658296e-01 2.24902138e-01 -2.26114646e-01 6.91497028e-02 7.04896376e-02 1.36144802e-01 1.59588769e-01 1.42625391e-01 3.37611586e-02 -2.61375278e-01 1.22945309e-01 -2.19374597e-01 -2.75762975e-01 1.60800052e+00 2.06486762e-01 -1.28054529e-01 7.32870817e-01 6.77382171e-01 2.05848098e-01 -9.18788254e-01 4.30917799e-01 4.65337150e-02 -8.50660324e-01 -8.88723210e-02 -8.90725136e-01 -6.21324360e-01 3.41034144e-01 2.85019279e-01 4.48867768e-01 1.22786009e+00 -4.04716015e-01 4.28198397e-01 5.20322204e-01 4.65780318e-01 -1.05969238e+00 -2.32391804e-01 1.79232121e-01 4.57594156e-01 -7.32044995e-01 2.53060758e-01 -2.80736595e-01 -3.42389494e-01 8.74802232e-01 7.69722819e-01 1.68804392e-01 -3.93571630e-02 3.10013205e-01 -4.32470381e-01 4.77245301e-02 -1.16531348e+00 1.74691722e-01 -4.01418537e-01 5.22613764e-01 7.40131199e-01 -8.89953375e-02 -8.31009388e-01 -2.43142605e-01 -4.37783629e-01 -8.54632035e-02 9.09845769e-01 1.53640604e+00 -4.97724384e-01 -1.40416217e+00 -6.35234892e-01 6.40624106e-01 -2.05288291e-01 -8.57307240e-02 -8.31877768e-01 9.59841490e-01 4.86296237e-01 7.89135695e-01 2.91660726e-01 -1.84371859e-01 8.19808021e-02 3.23562086e-01 6.91227853e-01 -7.40641654e-01 -8.70914340e-01 -1.52627274e-01 5.42445600e-01 1.03945218e-01 -2.84648724e-02 -1.17902935e+00 -1.00390935e+00 -3.01198214e-01 -6.17355943e-01 3.58957797e-01 6.95084870e-01 6.74494028e-01 4.51527476e-01 7.57581830e-01 2.73810714e-01 -2.95187235e-01 -5.30000508e-01 -5.69052219e-01 -1.71225697e-01 1.03673764e-01 -3.19713727e-02 -6.39006674e-01 -3.99022549e-01 -2.78670132e-01]
[8.038142204284668, 7.168479919433594]
c001806d-59b6-4f05-b4a6-ece480731182
icon-implicit-clothed-humans-obtained-from
2112.09127
null
https://arxiv.org/abs/2112.09127v2
https://arxiv.org/pdf/2112.09127v2.pdf
ICON: Implicit Clothed humans Obtained from Normals
Current methods for learning realistic and animatable 3D clothed avatars need either posed 3D scans or 2D images with carefully controlled user poses. In contrast, our goal is to learn an avatar from only 2D images of people in unconstrained poses. Given a set of images, our method estimates a detailed 3D surface from each image and then combines these into an animatable avatar. Implicit functions are well suited to the first task, as they can capture details like hair and clothes. Current methods, however, are not robust to varied human poses and often produce 3D surfaces with broken or disembodied limbs, missing details, or non-human shapes. The problem is that these methods use global feature encoders that are sensitive to global pose. To address this, we propose ICON ("Implicit Clothed humans Obtained from Normals"), which, instead, uses local features. ICON has two main modules, both of which exploit the SMPL(-X) body model. First, ICON infers detailed clothed-human normals (front/back) conditioned on the SMPL(-X) normals. Second, a visibility-aware implicit surface regressor produces an iso-surface of a human occupancy field. Importantly, at inference time, a feedback loop alternates between refining the SMPL(-X) mesh using the inferred clothed normals and then refining the normals. Given multiple reconstructed frames of a subject in varied poses, we use SCANimate to produce an animatable avatar from them. Evaluation on the AGORA and CAPE datasets shows that ICON outperforms the state of the art in reconstruction, even with heavily limited training data. Additionally, it is much more robust to out-of-distribution samples, e.g., in-the-wild poses/images and out-of-frame cropping. ICON takes a step towards robust 3D clothed human reconstruction from in-the-wild images. This enables creating avatars directly from video with personalized and natural pose-dependent cloth deformation.
['Michael J. Black', 'Dimitrios Tzionas', 'Jinlong Yang', 'Yuliang Xiu']
2021-12-16
null
http://openaccess.thecvf.com//content/CVPR2022/html/Xiu_ICON_Implicit_Clothed_Humans_Obtained_From_Normals_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Xiu_ICON_Implicit_Clothed_Humans_Obtained_From_Normals_CVPR_2022_paper.pdf
cvpr-2022-1
['monocular-3d-human-pose-estimation', '3d-human-reconstruction']
['computer-vision', 'computer-vision']
[ 2.22010121e-01 1.15891322e-01 3.78483087e-01 -2.93250769e-01 -6.68364406e-01 -5.42454302e-01 4.36858892e-01 -5.30703068e-01 -1.62233174e-01 5.80717564e-01 1.39280707e-01 3.22708249e-01 3.62691790e-01 -7.27789223e-01 -9.64801192e-01 -5.68907738e-01 7.38038123e-02 8.78341079e-01 3.12713832e-01 -4.28450465e-01 -3.33577663e-01 5.37864685e-01 -1.56183136e+00 1.96649618e-02 3.89983982e-01 9.16616857e-01 9.83490348e-02 9.17221129e-01 2.10833818e-01 3.00197065e-01 -4.53349203e-01 -4.81167436e-01 6.21825874e-01 -5.96068203e-01 -4.64726210e-01 5.00149786e-01 8.65867734e-01 -6.74017310e-01 -1.69354215e-01 7.12733567e-01 4.60776776e-01 1.40738800e-01 5.70651233e-01 -1.01269472e+00 -3.25452387e-01 -3.57639641e-02 -6.80017352e-01 -6.66715145e-01 9.17495668e-01 5.85875690e-01 7.28980422e-01 -9.89930689e-01 1.08785141e+00 1.45244992e+00 1.17806852e+00 9.50869560e-01 -1.68598318e+00 -5.35647094e-01 5.09790182e-02 -3.69222760e-01 -1.33077216e+00 -4.10390407e-01 8.33074152e-01 -5.36777854e-01 4.95551974e-01 5.30967295e-01 1.44389725e+00 1.43847632e+00 2.18954399e-01 6.39331758e-01 1.24497819e+00 -2.55202174e-01 1.41289830e-01 -1.01953819e-01 -4.64978099e-01 1.12986088e+00 -8.44633430e-02 1.80400357e-01 -5.63779116e-01 -3.82242620e-01 1.34831738e+00 2.82312557e-02 -4.88493860e-01 -8.53541672e-01 -1.29155815e+00 5.76756060e-01 2.76141942e-01 -4.00025636e-01 -5.86133897e-01 3.72749954e-01 4.44590449e-02 6.11802898e-02 3.55766922e-01 2.90760666e-01 -5.20957828e-01 -7.58248568e-02 -8.81545603e-01 8.98267329e-01 8.47815931e-01 9.36618626e-01 8.58667552e-01 5.03802449e-02 3.03377546e-02 4.75681365e-01 3.30138862e-01 9.00384724e-01 -2.84597218e-01 -1.28057361e+00 6.05696999e-02 2.92529076e-01 2.37540692e-01 -8.95098209e-01 -3.11171085e-01 5.69071062e-02 -6.61498010e-01 7.31158793e-01 6.33505225e-01 -2.04219252e-01 -1.30112147e+00 1.72513914e+00 8.32066417e-01 5.24839237e-02 -4.09334779e-01 1.43506682e+00 7.72975564e-01 4.81852949e-01 -3.12221143e-02 1.06508657e-01 1.44599354e+00 -7.55316854e-01 -5.01362562e-01 -1.72967210e-01 -2.67081112e-01 -7.05559671e-01 1.22382271e+00 4.12140101e-01 -1.44064915e+00 -4.91524488e-01 -6.80137992e-01 -1.67141587e-01 1.59735680e-01 -4.04049426e-01 4.20905769e-01 4.94410753e-01 -9.45500374e-01 7.06988752e-01 -9.54819679e-01 -4.80956733e-01 1.78652018e-01 2.15854630e-01 -5.21137536e-01 2.72909813e-02 -9.17591810e-01 8.93238425e-01 -3.25495720e-01 1.72100723e-01 -1.17240584e+00 -8.24645996e-01 -1.20325148e+00 -4.56207275e-01 5.30349433e-01 -1.06756055e+00 1.05476284e+00 -1.15324926e+00 -1.76427329e+00 9.93928850e-01 7.13979453e-02 1.04597295e-02 1.08773923e+00 -3.23388815e-01 6.51335493e-02 1.78623050e-01 -9.46531743e-02 8.49410772e-01 1.29991770e+00 -1.70750630e+00 -3.69185060e-02 -4.50986624e-01 -7.35673234e-02 4.07396525e-01 2.81196296e-01 -2.61266470e-01 -7.36739218e-01 -7.33500242e-01 1.93488777e-01 -1.14846122e+00 -2.28147373e-01 8.21876228e-01 -4.04424340e-01 2.77565151e-01 8.92478049e-01 -1.05924630e+00 4.68820274e-01 -1.79993713e+00 7.34373391e-01 4.30494606e-01 2.98455119e-01 -1.81052640e-01 -4.86349836e-02 4.90303487e-02 3.44265491e-01 -2.32348830e-01 -3.78487766e-01 -6.14695251e-01 9.36800614e-02 3.53602022e-01 1.61997586e-01 7.68985450e-01 1.02960132e-01 8.45318258e-01 -7.97095716e-01 -5.88577271e-01 3.83695632e-01 9.10794199e-01 -6.04652524e-01 5.91124535e-01 -4.69807088e-01 1.00248432e+00 -2.06534132e-01 8.37321341e-01 6.18367493e-01 1.22140467e-01 3.01340520e-01 -4.09883827e-01 8.31167847e-02 -2.32473448e-01 -1.21526253e+00 2.07679081e+00 -1.93413153e-01 1.81474343e-01 5.87695301e-01 -3.46587658e-01 7.50335217e-01 4.03968066e-01 6.12494528e-01 -4.00063455e-01 2.00715706e-01 1.05487660e-01 -4.71862495e-01 -4.66321439e-01 2.03241095e-01 -4.15094167e-01 -8.41460302e-02 2.05529377e-01 -1.37964515e-02 -6.41001821e-01 -4.26467240e-01 -1.60329804e-01 9.74043548e-01 1.00001395e+00 5.09858057e-02 -1.54815078e-01 1.97520554e-02 -4.24961075e-02 4.39963877e-01 3.86616915e-01 -3.26087177e-02 1.17942035e+00 3.39332789e-01 -8.04642379e-01 -1.51167655e+00 -1.44252992e+00 7.40294810e-03 7.93273628e-01 1.70494899e-01 -3.47669393e-01 -9.73764241e-01 -6.46530330e-01 1.61041588e-01 4.00386006e-01 -9.19475973e-01 2.29646429e-01 -8.24983716e-01 -1.53899625e-01 4.05235022e-01 2.84372061e-01 2.65882015e-01 -1.10135365e+00 -1.14975309e+00 1.48660481e-01 -2.57500589e-01 -9.54029024e-01 -7.72075117e-01 -1.06921270e-01 -5.58490992e-01 -1.10488355e+00 -1.03781891e+00 -4.68910843e-01 7.47831762e-01 -1.24840312e-01 1.31023777e+00 1.15866646e-01 -4.80478227e-01 6.57470703e-01 -2.38545239e-01 -9.40763801e-02 -3.97046536e-01 -4.22104537e-01 1.02224201e-01 1.17519476e-01 -3.33303809e-01 -6.47123456e-01 -7.15836406e-01 5.60795784e-01 -5.11954188e-01 4.90222782e-01 1.86219692e-01 8.42522264e-01 8.05688739e-01 -4.28053230e-01 -7.32859373e-02 -7.34910071e-01 1.93938807e-01 -8.06031823e-02 -4.48144168e-01 9.83845070e-02 -1.36546195e-01 -7.88351148e-02 3.70815456e-01 -7.27329791e-01 -9.16738689e-01 4.65597272e-01 -2.95786053e-01 -9.52907801e-01 -2.19712019e-01 -8.86602029e-02 -2.18532369e-01 4.97203320e-02 7.87633061e-01 -4.77219522e-02 4.54926193e-01 -5.19360006e-01 3.45095426e-01 1.17860340e-01 7.15020955e-01 -9.87804472e-01 1.01883399e+00 5.78696072e-01 6.17669662e-03 -1.00260532e+00 -5.12767375e-01 6.99048340e-02 -8.15351546e-01 -5.68985164e-01 1.23578858e+00 -8.86669993e-01 -8.22577000e-01 5.23157895e-01 -1.03189778e+00 -8.44472587e-01 -5.63020587e-01 1.46029815e-01 -7.77298093e-01 2.52052933e-01 -7.18330264e-01 -8.83332491e-01 -2.83409417e-01 -1.20514119e+00 1.59086168e+00 -1.06892303e-01 -7.92652249e-01 -8.08893919e-01 6.69134641e-03 4.65411097e-01 1.25952765e-01 1.11352003e+00 5.44288874e-01 3.37930262e-01 -4.81292754e-01 -1.13528892e-01 2.81725496e-01 2.05157474e-01 4.94727073e-03 -9.53369588e-02 -1.02119732e+00 -4.00854558e-01 -2.70578444e-01 -5.21414340e-01 2.65310049e-01 3.43357146e-01 8.54179680e-01 -5.54513812e-01 1.62999500e-02 7.95894265e-01 1.10388565e+00 -4.17760074e-01 6.00155950e-01 -1.59130901e-01 1.07397628e+00 7.32428789e-01 4.83606309e-01 5.12082100e-01 4.06028688e-01 1.02054167e+00 4.99754280e-01 -3.01753074e-01 -4.66954559e-01 -5.46276748e-01 5.80646753e-01 3.52406144e-01 -7.14664221e-01 2.01926112e-01 -5.98307610e-01 9.40159932e-02 -1.57198513e+00 -8.16707551e-01 -1.30795434e-01 2.28058887e+00 9.10200059e-01 -1.82845425e-02 5.48513830e-01 -1.81094617e-01 2.90882498e-01 2.56008971e-02 -6.75286353e-01 -1.60079420e-01 1.35562886e-02 4.05785590e-01 2.79460460e-01 6.72136605e-01 -7.50247002e-01 8.67483199e-01 5.63863611e+00 1.93753481e-01 -1.00835776e+00 1.63813591e-01 3.13305527e-01 -1.43507168e-01 -5.46181023e-01 -5.90364374e-02 -2.87637144e-01 2.37065345e-01 2.70104498e-01 5.99021554e-01 7.39538848e-01 6.67453766e-01 2.14229181e-01 -1.22677900e-01 -1.07542992e+00 1.05289555e+00 2.29755238e-01 -1.03281546e+00 -3.99676114e-02 1.76506773e-01 7.16143012e-01 -3.39089096e-01 -1.64826304e-01 1.88072249e-02 3.00748736e-01 -1.07804143e+00 1.41529214e+00 8.26719522e-01 1.20064092e+00 -5.18119514e-01 3.85105401e-01 3.60695153e-01 -1.18176973e+00 4.67166603e-01 3.25046331e-02 -4.77227531e-02 4.33022290e-01 1.60789236e-01 -5.19949079e-01 2.27384090e-01 9.20487106e-01 4.04426694e-01 -3.15017015e-01 4.17657405e-01 -2.08428919e-01 2.47743651e-01 -4.92269278e-01 2.94098616e-01 -2.34030262e-01 -2.92166233e-01 6.84210896e-01 1.01109958e+00 2.21414328e-01 2.00647086e-01 5.55951655e-01 1.08260787e+00 1.86750337e-01 -1.96443900e-01 -4.58826989e-01 4.26424354e-01 1.35068418e-02 1.22609401e+00 -6.04166806e-01 -2.10067958e-01 -1.03007466e-01 1.46120405e+00 2.36605316e-01 4.05993998e-01 -9.91700172e-01 2.92424887e-01 8.29965532e-01 8.44255507e-01 1.64495423e-01 -3.84701252e-01 -1.73433945e-01 -1.24406564e+00 -7.22234994e-02 -1.12602544e+00 -5.31891407e-03 -1.02227139e+00 -1.21207893e+00 5.20038486e-01 1.57277986e-01 -9.65646625e-01 -1.58157349e-01 -3.73641968e-01 -1.68253675e-01 8.31986308e-01 -7.37324357e-01 -1.57024038e+00 -7.16636837e-01 7.20845163e-01 6.84260368e-01 4.46431011e-01 1.08247375e+00 6.83208779e-02 -3.09731290e-02 3.77446830e-01 -6.68305814e-01 8.16607773e-02 6.84508443e-01 -1.26853704e+00 5.51107585e-01 4.43357974e-01 4.33134399e-02 2.71609217e-01 9.71400857e-01 -1.01327741e+00 -1.80730712e+00 -8.01952183e-01 2.54483372e-01 -9.51618373e-01 6.93170950e-02 -7.28019655e-01 -7.44212925e-01 7.57321179e-01 -1.31758861e-03 3.14672768e-01 1.10103838e-01 -5.44834994e-02 -3.93855900e-01 4.69719544e-02 -1.45404100e+00 7.29475379e-01 1.35173202e+00 -2.70782679e-01 -3.04465413e-01 1.27936974e-01 4.23991352e-01 -1.03000355e+00 -9.69763160e-01 3.29218894e-01 1.24140739e+00 -1.00323403e+00 1.13920486e+00 -3.91452938e-01 2.80648768e-01 -4.57436800e-01 -3.07564229e-01 -1.22115636e+00 -2.33753532e-01 -8.70055735e-01 -3.40586185e-01 7.17267931e-01 -1.74445193e-02 -1.07422575e-01 9.25541282e-01 8.42900455e-01 6.34369329e-02 -7.90147066e-01 -7.04160631e-01 -5.39203107e-01 -2.41290808e-01 -3.78394336e-01 5.15667856e-01 8.28894615e-01 -5.60644507e-01 2.36598864e-01 -8.85707557e-01 4.74455245e-02 1.02588856e+00 -9.27203745e-02 1.41009879e+00 -1.19187701e+00 -4.84670401e-01 1.22698732e-01 -2.70019144e-01 -1.00476897e+00 -2.03095917e-02 -4.24114317e-01 3.62712413e-01 -1.39428866e+00 1.13387704e-01 -5.67835093e-01 4.69741106e-01 6.52224183e-01 -3.00718471e-02 6.85752571e-01 4.46717232e-01 1.19260386e-01 -2.42400974e-01 4.53205824e-01 1.73964500e+00 1.29472613e-01 -2.87118822e-01 -7.61977434e-02 -9.31881145e-02 1.05771267e+00 4.22279060e-01 -2.38363564e-01 -1.71993837e-01 -3.65472496e-01 1.08120114e-01 2.89696306e-01 9.90291417e-01 -9.57304239e-01 -2.47765616e-01 -2.90872276e-01 7.92701781e-01 -3.65435392e-01 9.39270437e-01 -9.85054135e-01 9.67831373e-01 4.66031700e-01 -1.54894665e-02 2.78264344e-01 -4.48092185e-02 5.33590734e-01 6.20304883e-01 2.32127577e-01 9.78022397e-01 -5.96231580e-01 -2.88693309e-01 4.69785303e-01 -8.28669295e-02 1.14318199e-01 8.31645608e-01 -4.70603913e-01 4.54900771e-01 -5.94696283e-01 -9.56892014e-01 -1.55739337e-01 1.12486029e+00 4.10887241e-01 8.30100119e-01 -1.22385800e+00 -8.36111844e-01 5.58060944e-01 -1.27884015e-01 5.40375471e-01 2.36098364e-01 7.01647520e-01 -8.84149671e-01 -5.55571496e-01 -2.57676512e-01 -9.52314734e-01 -1.45568609e+00 2.07494721e-01 5.62050045e-01 1.16868339e-01 -1.16667461e+00 6.92657948e-01 2.28156269e-01 -7.49504209e-01 2.14396134e-01 -1.58221364e-01 5.61859190e-01 -1.82123050e-01 2.23423123e-01 2.03702182e-01 -2.60275424e-01 -9.53710139e-01 -2.04593718e-01 9.16380465e-01 4.33323979e-01 -4.52248812e-01 1.23253214e+00 3.15761045e-02 5.82172386e-02 4.82068926e-01 9.56170499e-01 3.37154895e-01 -2.05289960e+00 1.02983467e-01 -7.06365168e-01 -7.06302226e-01 -4.39522922e-01 -8.05698335e-01 -1.10324109e+00 5.78244567e-01 5.51103771e-01 -1.87138274e-01 8.53674471e-01 8.63842964e-02 9.98969376e-01 -1.15908511e-01 6.97383940e-01 -8.76096487e-01 2.20405951e-01 1.36870787e-01 1.15100241e+00 -9.05837536e-01 2.75752991e-01 -4.15291756e-01 -7.31095016e-01 1.00594831e+00 5.98522842e-01 -2.23403782e-01 4.82271373e-01 4.30202067e-01 2.76793748e-01 -4.31392461e-01 -2.23626092e-01 1.12631395e-01 4.36930388e-01 7.19637752e-01 4.16121557e-02 3.00363272e-01 2.51181483e-01 7.61781782e-02 -4.41050529e-01 -2.70220667e-01 1.98192358e-01 8.31690013e-01 -9.83802527e-02 -9.26619232e-01 -8.41038764e-01 1.47665411e-01 -3.53155695e-02 3.73549670e-01 -3.91776532e-01 8.68322492e-01 4.10842210e-01 4.67847288e-01 -6.32264987e-02 -4.36918616e-01 6.57626212e-01 -1.07592000e-02 1.02787018e+00 -5.60730517e-01 -6.67448759e-01 3.92329663e-01 1.14125527e-01 -7.49747813e-01 -4.25196677e-01 -8.38620007e-01 -1.11069334e+00 -5.66008627e-01 -1.58708319e-01 -2.13577569e-01 6.39872849e-01 6.07756376e-01 8.26804489e-02 3.43828350e-01 1.83267280e-01 -1.69884074e+00 -3.55840355e-01 -6.29455507e-01 -5.29906988e-01 8.93173873e-01 4.77790803e-01 -8.53035033e-01 -1.50628701e-01 2.73263812e-01]
[7.15957498550415, -1.2473492622375488]
007a2603-3b42-4d8b-9e44-cfeaa1becfa6
quality-aware-network-for-human-parsing
2103.05997
null
https://arxiv.org/abs/2103.05997v1
https://arxiv.org/pdf/2103.05997v1.pdf
Quality-Aware Network for Human Parsing
How to estimate the quality of the network output is an important issue, and currently there is no effective solution in the field of human parsing. In order to solve this problem, this work proposes a statistical method based on the output probability map to calculate the pixel quality information, which is called pixel score. In addition, the Quality-Aware Module (QAM) is proposed to fuse the different quality information, the purpose of which is to estimate the quality of human parsing results. We combine QAM with a concise and effective network design to propose Quality-Aware Network (QANet) for human parsing. Benefiting from the superiority of QAM and QANet, we achieve the best performance on three multiple and one single human parsing benchmarks, including CIHP, MHP-v2, Pascal-Person-Part and LIP. Without increasing the training and inference time, QAM improves the AP$^\text{r}$ criterion by more than 10 points in the multiple human parsing task. QAM can be extended to other tasks with good quality estimation, e.g. instance segmentation. Specifically, QAM improves Mask R-CNN by ~1% mAP on COCO and LVISv1.0 datasets. Based on the proposed QAM and QANet, our overall system wins 1st place in CVPR2019 COCO DensePose Challenge, and 1st place in Track 1 & 2 of CVPR2020 LIP Challenge. Code and models are available at https://github.com/soeaver/QANet.
['Zhihao LI', 'Songcen Xu', 'Zhiwei Liu', 'Zhihui Wang', 'Qing Song', 'Lu Yang']
2021-03-10
null
null
null
null
['human-parsing']
['computer-vision']
[ 2.01951891e-01 2.89196312e-01 -1.10150844e-01 -6.43779337e-01 -1.22656357e+00 -2.43371993e-01 -1.34539872e-01 -5.95572814e-02 -5.65908313e-01 5.53515851e-01 3.53798009e-02 -1.02599315e-01 2.64451295e-01 -7.65668094e-01 -7.65828013e-01 -4.55340385e-01 5.50315976e-01 2.23708138e-01 4.50028539e-01 1.76596809e-02 1.12620331e-02 -4.84010158e-03 -1.20393455e+00 3.01035583e-01 1.12190735e+00 1.23026168e+00 3.77700448e-01 6.96305215e-01 -2.78186023e-01 3.68782580e-01 -6.27395213e-01 -7.89992869e-01 8.11361596e-02 -4.71558303e-01 -8.98729265e-01 -4.02442753e-01 6.62719429e-01 -3.96821141e-01 8.30240697e-02 1.33432746e+00 9.97983515e-01 -1.16651803e-01 3.23889822e-01 -1.14665222e+00 -4.34998900e-01 9.16251838e-01 -6.64819181e-01 6.86214343e-02 6.56920150e-02 3.79551262e-01 1.04016447e+00 -5.66552401e-01 3.93117011e-01 1.50417757e+00 7.58881271e-01 8.16434741e-01 -6.90022886e-01 -7.32985616e-01 -6.06597774e-03 2.81436980e-01 -1.01899219e+00 -2.75463879e-01 5.82855225e-01 -1.19711738e-02 6.74005926e-01 1.53748736e-01 4.53696370e-01 7.72960603e-01 -2.34844834e-02 1.26132417e+00 1.29486084e+00 -1.96872935e-01 1.23216622e-02 -4.12518889e-01 3.11505854e-01 8.99982035e-01 7.32606798e-02 -9.30667818e-02 -4.35543299e-01 4.07341003e-01 6.20062172e-01 -4.90480453e-01 -1.69317588e-01 3.30997199e-01 -8.38047683e-01 6.45305097e-01 6.60634875e-01 1.38281420e-01 -1.70466200e-01 2.84468144e-01 4.42935497e-01 -2.76466817e-01 4.06045020e-01 2.16335580e-01 -6.46824121e-01 -3.20891589e-01 -1.04762042e+00 7.70621374e-02 4.78245258e-01 8.56348753e-01 4.53291595e-01 -1.57402545e-01 -5.28982997e-01 1.11029780e+00 4.80052799e-01 6.68964565e-01 1.77590996e-01 -1.37261105e+00 8.50338876e-01 5.45941889e-01 -2.04717830e-01 -6.59966588e-01 -6.03718162e-01 -1.94109470e-01 -8.68547857e-01 2.68565506e-01 7.66987741e-01 -1.69380382e-01 -1.35879946e+00 1.86957598e+00 4.22739714e-01 -6.29479960e-02 -1.70718715e-01 9.84199226e-01 1.35694659e+00 7.16081858e-01 5.30374825e-01 3.69145311e-02 1.70271575e+00 -1.08600152e+00 -8.06128204e-01 -1.63995594e-01 3.01859170e-01 -8.38793039e-01 1.28129053e+00 5.57697058e-01 -1.41924524e+00 -8.71713161e-01 -9.86780107e-01 -4.20970142e-01 -1.73704177e-01 3.28183472e-01 5.33049762e-01 6.78743601e-01 -1.01721883e+00 8.24826419e-01 -8.42590332e-01 -6.98824674e-02 8.41752827e-01 3.79489720e-01 -2.28830054e-01 -2.95940191e-01 -1.20863986e+00 5.86783886e-01 4.49114352e-01 4.28216487e-01 -6.06752396e-01 -6.28769100e-01 -8.68545532e-01 -4.97798696e-02 3.70844036e-01 -6.66003227e-01 1.38448942e+00 -7.55247593e-01 -1.59932137e+00 8.69410098e-01 -6.02987371e-02 -3.62768441e-01 6.95523441e-01 -4.21099067e-01 -1.92326605e-01 2.07574695e-01 2.38945588e-01 1.28201973e+00 4.29184794e-01 -9.05488610e-01 -6.53430700e-01 -3.80171657e-01 -5.97641543e-02 1.47042945e-01 2.16758490e-01 1.28366202e-01 -1.06481600e+00 -3.79761368e-01 1.61304131e-01 -7.12005675e-01 -1.08773291e-01 4.46655527e-02 -7.58061767e-01 -4.95116174e-01 4.02794600e-01 -1.18809092e+00 1.06090331e+00 -2.04900742e+00 -3.66806500e-02 -1.04952678e-01 2.00149119e-01 5.45492172e-01 -3.08918893e-01 -4.10668105e-01 3.17518175e-01 4.69522685e-01 -5.02994835e-01 -5.64228237e-01 1.29314646e-01 8.06424916e-02 4.31994677e-01 9.54873487e-02 4.41365629e-01 1.25366855e+00 -6.88622534e-01 -1.03185701e+00 2.36219287e-01 5.95338047e-01 -4.03007656e-01 2.05888197e-01 -3.52955848e-01 5.37164330e-01 -4.47889209e-01 8.89778793e-01 1.08016384e+00 -2.05510303e-01 -1.07962243e-01 -5.54010272e-01 -2.92159105e-03 2.21162006e-01 -1.16419172e+00 1.97929502e+00 -3.60173374e-01 3.83736193e-01 2.58818358e-01 -6.26294672e-01 6.72611594e-01 2.68723607e-01 2.50827104e-01 -1.08265412e+00 3.17203343e-01 1.54759333e-01 7.91653171e-02 -3.95633548e-01 2.70104855e-01 1.63698658e-01 -4.75340113e-02 -9.86892134e-02 2.61130840e-01 -6.98646381e-02 3.57485890e-01 6.27897233e-02 1.01761448e+00 3.29270869e-01 1.86091125e-01 -1.25874177e-01 3.75023454e-01 -2.41361260e-01 1.09352517e+00 5.28579414e-01 -6.82831883e-01 9.64206457e-01 6.53329313e-01 -9.60954726e-02 -7.91663170e-01 -1.13514686e+00 -2.31398091e-01 9.22969759e-01 2.33252034e-01 -3.34004372e-01 -1.23122859e+00 -8.41437042e-01 -3.15262139e-01 3.66604954e-01 -5.52678287e-01 2.38068059e-01 -8.71142447e-01 -9.30670917e-01 7.67001092e-01 8.06558371e-01 1.14940393e+00 -1.48417974e+00 -4.63839263e-01 2.69588619e-01 -6.23521805e-01 -1.25382340e+00 -6.48146212e-01 -8.15936327e-02 -7.47690737e-01 -1.03317201e+00 -9.24576461e-01 -6.41935945e-01 2.63764113e-01 -2.97636867e-01 1.27493572e+00 8.74379501e-02 -3.92841160e-01 -3.61731872e-02 -4.25230384e-01 -5.54727316e-01 -3.12889725e-01 1.54489607e-01 -5.49589157e-01 -3.88493299e-01 1.29712045e-01 -1.54007211e-01 -9.45922077e-01 4.07830298e-01 -5.79478860e-01 2.77595907e-01 7.80330777e-01 5.49686253e-01 1.04164541e+00 -1.29563749e-01 5.80320895e-01 -8.02724361e-01 2.90167838e-01 -7.13496134e-02 -5.80491185e-01 3.60686302e-01 -4.88239110e-01 -6.04439564e-02 3.18060756e-01 -1.72000438e-01 -1.07447326e+00 7.41873085e-02 -8.77612829e-01 1.98343862e-02 -1.54163569e-01 3.21508832e-02 -9.48331237e-01 2.46791214e-01 2.94147313e-01 -2.69029677e-01 -2.70221382e-01 -6.59981906e-01 5.02739906e-01 4.44268614e-01 7.32555926e-01 -5.13784885e-01 1.98074296e-01 1.75050616e-01 5.21645043e-03 -3.32967967e-01 -9.16849136e-01 -3.50780845e-01 -4.91392970e-01 -2.62771308e-01 1.54227626e+00 -8.31934869e-01 -9.59814310e-01 8.52976561e-01 -1.37057018e+00 -5.19133627e-01 -1.45537630e-01 3.48683238e-01 -4.32176918e-01 4.42413419e-01 -8.75938892e-01 -6.26320481e-01 -8.62825036e-01 -1.52375007e+00 1.22092831e+00 7.25463212e-01 7.59806708e-02 -5.70980489e-01 -2.50309944e-01 9.31015372e-01 2.09095314e-01 2.80376375e-01 6.98522270e-01 -2.75789797e-01 -5.46006560e-01 1.67554274e-01 -6.24201596e-01 6.91742301e-01 -2.10690558e-01 -3.77737135e-02 -1.10574687e+00 1.30831048e-01 -3.87148649e-01 -3.55440974e-01 1.15739822e+00 9.47811842e-01 1.36805868e+00 7.37849623e-02 -8.48964974e-02 6.79439306e-01 1.23109508e+00 1.44342959e-01 1.00951219e+00 -6.44267201e-02 8.86648953e-01 5.46026766e-01 8.49623263e-01 -9.18528438e-03 5.61682224e-01 7.32336283e-01 5.72209001e-01 -2.83149034e-01 -7.59686291e-01 -3.15536022e-01 2.75935858e-01 7.05848098e-01 -3.81564535e-02 -3.69885117e-01 -7.71860778e-01 3.31933647e-01 -1.67122948e+00 -6.88605249e-01 -3.22473556e-01 1.87622726e+00 9.41948235e-01 3.08701247e-01 1.59273878e-01 7.20745027e-02 9.12458956e-01 4.29444723e-02 -6.47509515e-01 -4.60479170e-01 -1.43599361e-01 6.14154458e-01 5.26361883e-01 4.76318657e-01 -1.13545763e+00 1.19572818e+00 5.07996178e+00 1.12858307e+00 -8.58347058e-01 4.37556207e-01 1.07493734e+00 2.64597893e-01 -3.62566896e-02 -2.38810003e-01 -1.10657728e+00 6.04981661e-01 8.29638124e-01 6.87982798e-01 1.72133222e-01 7.38895178e-01 2.54496336e-01 -2.81464458e-01 -7.30878830e-01 1.01913106e+00 -1.64296597e-01 -1.00118017e+00 -3.46498579e-01 -1.15211315e-01 3.65069509e-01 8.42346251e-02 2.25185901e-02 4.69788134e-01 1.51724517e-01 -1.24801576e+00 6.27686024e-01 3.87985528e-01 8.95827174e-01 -6.89583004e-01 8.60210776e-01 1.21136568e-01 -1.40203631e+00 2.22362280e-01 -3.50361109e-01 4.89741951e-01 4.78610665e-01 8.54557693e-01 -5.00336468e-01 4.97441500e-01 1.10570335e+00 3.71412039e-01 -5.29087961e-01 1.16799557e+00 -6.65691495e-01 8.15981328e-01 -3.25788826e-01 8.32534283e-02 1.36018703e-02 -3.45996395e-02 1.80874795e-01 1.24061012e+00 2.45129690e-01 2.57304728e-01 -9.16226059e-02 9.81822133e-01 -5.04561484e-01 2.54599512e-01 1.91729754e-01 2.52683312e-01 3.03166002e-01 1.62777185e+00 -8.94490421e-01 -3.78803730e-01 -2.09661528e-01 1.05220211e+00 2.25428715e-01 -3.97670344e-02 -1.05196023e+00 -3.20742011e-01 3.12173396e-01 -4.46852557e-02 2.63570845e-01 8.90785530e-02 -5.56234121e-01 -8.04932892e-01 1.48607805e-01 -8.65217090e-01 3.52318645e-01 -8.63983810e-01 -1.12870502e+00 7.11721182e-01 -1.32270038e-01 -8.43731821e-01 1.18550889e-01 -7.47501552e-01 -6.08218729e-01 1.09276819e+00 -1.61129224e+00 -1.26677513e+00 -5.03026426e-01 3.37422758e-01 6.18120551e-01 3.00435185e-01 4.64203596e-01 5.69523513e-01 -7.54135013e-01 9.00549710e-01 -5.76284766e-01 4.73532677e-01 7.11571693e-01 -1.52110696e+00 7.67143250e-01 7.96380758e-01 -8.99667591e-02 8.72043893e-02 3.90149504e-01 -6.76916301e-01 -8.24602902e-01 -1.17351031e+00 7.13555813e-01 -4.14022416e-01 3.95709872e-02 -1.86995849e-01 -8.42280030e-01 9.62083414e-02 2.53179878e-01 1.77153260e-01 2.90009975e-01 -2.91075222e-02 -3.39343846e-01 -1.65773720e-01 -1.50614905e+00 2.91592211e-01 1.13594496e+00 -4.85403538e-02 -2.28390366e-01 1.20609500e-01 1.11383986e+00 -6.90187097e-01 -1.06684721e+00 8.85438383e-01 5.80674887e-01 -9.36953366e-01 9.16597486e-01 -1.34165540e-01 4.74803358e-01 -3.08968067e-01 -2.51248598e-01 -1.02617514e+00 -3.29694077e-02 -2.60317773e-01 1.35176675e-02 1.58798671e+00 6.78101540e-01 -3.11181515e-01 7.70067334e-01 5.06453037e-01 -3.47774118e-01 -8.74957681e-01 -1.24333501e+00 -4.78041977e-01 3.18368256e-01 -6.77169681e-01 6.32606626e-01 3.50892067e-01 -7.24801779e-01 3.10737163e-01 -2.96740413e-01 1.24171801e-01 7.47025192e-01 -2.65663087e-01 5.64190745e-01 -9.87702549e-01 -4.75938439e-01 -5.06654859e-01 -2.04323784e-01 -1.05822504e+00 1.71933249e-02 -6.90980315e-01 3.46145719e-01 -1.91621327e+00 3.14973950e-01 -4.62020725e-01 -3.27000618e-01 5.98014951e-01 -6.40684247e-01 3.38181168e-01 5.92291355e-01 -1.43580601e-01 -7.78457403e-01 2.21345052e-01 1.65049613e+00 -1.73780739e-01 -8.65430459e-02 5.61185107e-02 -6.57313645e-01 6.35635555e-01 9.82626617e-01 -3.18644524e-01 -1.35119513e-01 -6.13484442e-01 1.05951928e-01 8.61035287e-02 5.08276045e-01 -1.16242003e+00 -1.29378662e-01 9.29448679e-02 4.59563255e-01 -8.07584822e-01 3.89282912e-01 -3.56804818e-01 -2.31713787e-01 3.61550957e-01 -1.97185129e-01 8.09092671e-02 2.94618040e-01 1.44778192e-01 -1.86757475e-01 -2.33583644e-01 1.16101658e+00 -1.84463009e-01 -9.11571264e-01 4.90021437e-01 4.84800220e-01 3.87795717e-01 5.90099275e-01 -6.69869706e-02 -6.29510999e-01 -7.65841752e-02 -6.75107718e-01 4.62019712e-01 7.59918690e-02 2.37518355e-01 5.91371775e-01 -1.08873165e+00 -7.72548616e-01 -2.42380172e-01 -4.14822251e-02 3.45232695e-01 7.05523074e-01 8.05330276e-01 -6.39937401e-01 3.46626565e-02 -1.31079003e-01 -7.05097258e-01 -1.35818005e+00 5.57120703e-02 4.36279535e-01 -5.09848893e-01 -4.23499227e-01 1.13922656e+00 7.86416605e-02 -5.58649480e-01 2.93277770e-01 -5.96858144e-01 -2.01251954e-01 -1.98177278e-01 6.29713476e-01 4.35603082e-01 3.12119834e-02 -6.35325491e-01 -4.13794458e-01 8.18406105e-01 1.16345227e-01 -1.90865993e-01 1.04246259e+00 5.96620217e-02 7.53423013e-03 -3.43639180e-02 1.21698213e+00 -1.21274807e-01 -1.43443155e+00 2.78608520e-02 -2.02269629e-01 -9.51147303e-02 3.30548035e-03 -1.33619034e+00 -1.45046234e+00 1.33323216e+00 1.11971557e+00 -3.02545369e-01 1.33950925e+00 1.12317100e-01 1.22074091e+00 -3.13394755e-01 1.18869036e-01 -1.29764628e+00 4.45934338e-03 3.08582217e-01 6.53910697e-01 -1.56965780e+00 -2.34191701e-01 -6.47112429e-01 -7.69893229e-01 8.47172201e-01 8.67712080e-01 2.72368282e-01 4.05201405e-01 2.12061748e-01 2.07093924e-01 2.54963748e-02 -2.06745386e-01 -3.82120341e-01 5.96583188e-01 4.78829682e-01 5.63482463e-01 3.78193647e-01 -5.52618086e-01 9.31431055e-01 -1.67081520e-01 4.33818623e-02 -8.91841482e-03 5.10859787e-01 -5.03853798e-01 -1.17032242e+00 -2.05391020e-01 3.90816301e-01 -7.45060980e-01 -2.13602528e-01 -7.61287287e-02 7.71271646e-01 6.72511637e-01 1.08121812e+00 -1.23305984e-01 -3.22984904e-01 5.18319547e-01 -8.56065378e-02 4.82329011e-01 -3.68889689e-01 -5.66661894e-01 8.93050581e-02 1.74867779e-01 -8.32828939e-01 -4.56322193e-01 -4.34627533e-01 -1.63033938e+00 -2.13867471e-01 -3.29504043e-01 -1.52731806e-01 9.54042256e-01 7.91176498e-01 1.70084268e-01 7.40119815e-01 1.01114869e-01 -4.66155082e-01 -2.46684060e-01 -1.19270694e+00 -3.17000628e-01 3.97225976e-01 -2.47762367e-01 -4.90538687e-01 4.80534211e-02 -1.64028466e-01]
[8.904526710510254, 0.15616299211978912]
8d47af82-d6cc-4003-8333-799a9d5e0400
data-mining-using-unguided-symbolic
1309.5931
null
http://arxiv.org/abs/1309.5931v1
http://arxiv.org/pdf/1309.5931v1.pdf
Data Mining using Unguided Symbolic Regression on a Blast Furnace Dataset
In this paper a data mining approach for variable selection and knowledge extraction from datasets is presented. The approach is based on unguided symbolic regression (every variable present in the dataset is treated as the target variable in multiple regression runs) and a novel variable relevance metric for genetic programming. The relevance of each input variable is calculated and a model approximating the target variable is created. The genetic programming configurations with different target variables are executed multiple times to reduce stochastic effects and the aggregated results are displayed as a variable interaction network. This interaction network highlights important system components and implicit relations between the variables. The whole approach is tested on a blast furnace dataset, because of the complexity of the blast furnace and the many interrelations between the variables. Finally the achieved results are discussed with respect to existing knowledge about the blast furnace process.
['Michael Kommenda', 'Gabriel Kronberger', 'Michael Affenzeller', 'Christoph Feilmayr']
2013-09-23
null
null
null
null
['implicit-relations']
['natural-language-processing']
[ 5.39644539e-01 3.40505332e-01 -1.63011119e-01 -3.36638033e-01 1.29186347e-01 -9.15416852e-02 4.01154220e-01 6.21845305e-01 -1.59717470e-01 1.03982937e+00 -2.84667671e-01 -2.87426174e-01 -9.16549861e-01 -1.05510259e+00 -2.03661501e-01 -7.86290050e-01 -1.16541356e-01 1.01129377e+00 -8.49502236e-02 -3.67513686e-01 4.85216856e-01 7.55675554e-01 -1.87659109e+00 1.45940617e-01 5.77880085e-01 6.19395494e-01 4.69123513e-01 6.37300968e-01 -4.03465539e-01 7.23698676e-01 -1.03933287e+00 3.08874309e-01 3.59045528e-02 -9.25245285e-01 -8.25726330e-01 5.60635924e-02 -5.25550067e-01 5.62294126e-01 2.88011670e-01 6.65138781e-01 1.40753940e-01 4.70980972e-01 7.88776577e-01 -1.21906745e+00 -2.81497054e-02 9.15424109e-01 -2.26171702e-01 4.36384752e-02 3.69444340e-01 -4.44679894e-02 9.85783875e-01 -7.47362852e-01 7.24495411e-01 1.19523430e+00 4.96877618e-02 -1.54956756e-02 -1.60108161e+00 -4.05234456e-01 1.53493792e-01 5.17513394e-01 -1.47559202e+00 1.12990558e-01 7.15851605e-01 -6.75293922e-01 1.28221297e+00 5.91611683e-01 8.35712016e-01 3.19246978e-01 2.42072210e-01 -2.43759342e-02 9.39147174e-01 -9.76474881e-01 5.30399621e-01 4.35350478e-01 5.36323011e-01 3.61454636e-01 1.79810658e-01 4.76568043e-01 -3.94170642e-01 -3.68563607e-02 2.96202987e-01 -2.43849978e-01 3.20904478e-02 -2.37329572e-01 -6.74154401e-01 9.71846521e-01 1.83311999e-01 7.52133667e-01 -6.07353091e-01 -4.51557115e-02 3.60842079e-01 6.16689026e-01 6.64295629e-02 9.58755493e-01 -7.78105557e-01 4.14880216e-02 -6.94203794e-01 5.49354255e-01 1.04208744e+00 7.26447940e-01 1.02480400e+00 8.20063576e-02 -2.18234599e-01 5.85308313e-01 2.54866928e-01 3.80834863e-02 2.91525155e-01 -3.73039067e-01 1.64407581e-01 1.17817295e+00 -2.70876229e-01 -1.09992909e+00 -6.47389233e-01 2.13490129e-02 -3.11700642e-01 9.19723511e-01 2.31699392e-01 -2.73773015e-01 -8.19315195e-01 1.18741989e+00 3.64112705e-01 -2.85470754e-01 2.72597410e-02 5.30805886e-01 7.86588311e-01 5.44933915e-01 1.06955506e-01 -8.08869839e-01 9.97268140e-01 -4.89098161e-01 -8.63827527e-01 2.66860396e-01 3.19090873e-01 -4.11379129e-01 5.33471406e-01 8.40514839e-01 -8.00008297e-01 -6.12641990e-01 -9.17501271e-01 4.54770416e-01 -4.87030178e-01 -1.21589921e-01 5.35673141e-01 4.56393003e-01 -5.51722944e-01 7.94571280e-01 -5.10958016e-01 -9.30495858e-02 -2.34411106e-01 8.27316463e-01 -2.78772810e-03 4.01985556e-01 -1.14650118e+00 1.26368511e+00 8.73059213e-01 6.65700734e-02 -4.26309973e-01 -5.43113232e-01 -6.29379153e-01 1.23941846e-01 6.57130301e-01 -4.07268584e-01 6.88836932e-01 -1.00248384e+00 -1.55371702e+00 2.80465305e-01 -2.45500311e-01 -1.89233571e-01 4.54487681e-01 2.77840793e-01 -5.07150233e-01 -2.57862657e-01 -3.41834933e-01 -3.45307022e-01 4.97309327e-01 -1.28926909e+00 -7.42506802e-01 -4.16507542e-01 -4.12051916e-01 -1.27105666e-02 2.82857895e-01 2.02733397e-01 -2.41952583e-01 -4.15770829e-01 1.89715818e-01 -5.98558009e-01 -5.52856445e-01 -7.78835654e-01 -3.34333479e-01 -2.39871889e-01 6.70669675e-01 -5.29005110e-01 1.72364295e+00 -1.68285978e+00 8.50716293e-01 1.18363214e+00 4.83852252e-02 -2.60388464e-01 3.74465585e-01 6.09250844e-01 -6.70242369e-01 5.43573126e-02 -3.66172165e-01 3.74512762e-01 -3.09337795e-01 2.40141898e-01 1.45372838e-01 1.75096700e-03 1.20100372e-01 2.95917958e-01 -6.44260764e-01 -4.41062987e-01 6.61665738e-01 2.34123096e-01 -1.74519882e-01 2.08865270e-01 -5.78115940e-01 3.64286959e-01 -7.30252385e-01 5.70732236e-01 3.45094323e-01 5.09842396e-01 6.17798805e-01 1.38490286e-03 -2.81498134e-01 -8.87990072e-02 -1.44303608e+00 8.95592749e-01 -4.63991433e-01 4.49284166e-01 -2.83467293e-01 -1.16360152e+00 1.85629535e+00 4.30182904e-01 6.71525717e-01 -6.27601147e-01 4.89343047e-01 9.57936272e-02 8.54113549e-02 -8.03057849e-01 5.15077889e-01 -8.21352154e-02 1.89623669e-01 2.56533951e-01 -1.46290243e-01 -4.76320893e-01 5.60803950e-01 -3.71283114e-01 8.75092506e-01 2.02740610e-01 7.12980270e-01 -3.59343588e-01 8.53948414e-01 3.65986258e-01 5.15427649e-01 3.86459500e-01 5.96737683e-01 1.17293499e-01 9.96256411e-01 -4.90522683e-01 -8.80784035e-01 -4.79522645e-01 -2.29591131e-01 6.40961051e-01 -5.21069467e-02 -4.93717670e-01 -2.28896394e-01 -3.67406398e-01 1.24675207e-01 1.14564610e+00 -8.01590085e-01 -3.67149524e-02 -4.01657432e-01 -7.08515346e-01 -7.48177022e-02 -1.17218317e-02 -1.88527077e-01 -1.31079972e+00 -8.61831307e-01 5.69489539e-01 2.77160078e-01 -1.43907949e-01 9.13063169e-01 7.48822689e-01 -9.72011924e-01 -1.29841185e+00 2.28058338e-01 -5.44239581e-01 5.57508707e-01 -5.38954139e-01 1.04033279e+00 4.73031700e-01 -5.85640609e-01 -4.88557220e-01 -4.38835144e-01 -6.77873611e-01 -9.47570622e-01 -2.01417759e-01 -3.19585055e-01 -3.87238950e-01 4.69997585e-01 -3.89405817e-01 3.09160501e-01 3.10810506e-01 -5.33198953e-01 -2.34757572e-01 2.70831615e-01 1.02742290e+00 7.99405694e-01 8.78662646e-01 3.21766764e-01 -1.07401550e+00 7.04560816e-01 -6.91452146e-01 -9.45642471e-01 4.62931395e-01 -1.03079331e+00 3.33686739e-01 4.71871227e-01 -4.46373552e-01 -8.99825692e-01 2.02531487e-01 3.46396059e-01 -1.09361485e-01 -2.08986923e-01 9.86887217e-01 -2.80173689e-01 1.82669327e-01 8.58339965e-01 -2.96037227e-01 1.35931119e-01 -4.39578623e-01 -2.15046912e-01 2.92640537e-01 1.49071708e-01 -6.18964911e-01 5.68421483e-01 -3.33503485e-01 6.75400436e-01 -7.63022602e-01 1.73845574e-01 -3.78570184e-02 -7.95232534e-01 -4.52244997e-01 4.82918739e-01 -1.27859622e-01 -8.10985446e-01 9.09217745e-02 -8.49775970e-01 -1.06012307e-01 -5.70045888e-01 5.67721546e-01 -3.76087636e-01 -1.22893088e-01 2.19888374e-01 -1.22992516e+00 -6.73223883e-02 -1.47935009e+00 3.70680615e-02 1.86566174e-01 -7.63483942e-01 -7.52087235e-01 1.91643685e-01 -8.32210556e-02 -6.29389361e-02 6.33826971e-01 1.40680027e+00 -8.48246813e-01 -1.58894241e-01 -3.46756160e-01 3.58411282e-01 3.11050884e-04 2.14357451e-01 7.40553677e-01 -4.97138143e-01 1.36925906e-01 -5.70803229e-03 1.95194185e-01 4.75315571e-01 3.92823309e-01 9.01322901e-01 7.37778051e-03 -4.80852097e-01 2.48575568e-01 1.58711016e+00 8.45504582e-01 6.42454624e-01 6.89522803e-01 4.26585972e-01 9.51321185e-01 1.08264363e+00 6.32822871e-01 -4.23756719e-01 8.72770131e-01 3.61323088e-01 -3.88457216e-02 3.32631469e-01 2.46756643e-01 4.29384746e-02 -7.09211687e-03 -3.30517530e-01 -1.62948728e-01 -1.04266453e+00 2.63957143e-01 -1.68951547e+00 -1.01706028e+00 -8.42180312e-01 2.09412169e+00 5.44054925e-01 1.51474491e-01 1.68730497e-01 8.01216781e-01 6.96749091e-01 -3.58330280e-01 -2.89753377e-01 -1.07818484e+00 -1.62296548e-01 7.10222006e-01 3.69707972e-01 8.92205775e-01 -5.90179443e-01 5.63449502e-01 6.61577559e+00 3.26791972e-01 -9.61000979e-01 -5.74530721e-01 4.06264305e-01 -1.90140918e-01 -2.66707659e-01 2.19564319e-01 -5.01154780e-01 2.21431732e-01 7.97915637e-01 -6.15655661e-01 5.89874089e-01 7.70467520e-01 4.22959894e-01 -5.60067296e-01 -1.11086702e+00 4.60962027e-01 -1.66993156e-01 -9.50550616e-01 -1.73154473e-01 6.18262701e-02 5.50289690e-01 -4.72834975e-01 -3.86532515e-01 -4.16970626e-03 3.92222047e-01 -1.30853927e+00 5.13345838e-01 6.71458006e-01 4.43214685e-01 -1.32871139e+00 8.68543863e-01 5.52472249e-02 -9.65070724e-01 -4.90420878e-01 -1.16786189e-01 -3.66849959e-01 -1.64194889e-02 6.61930919e-01 -1.34860969e+00 1.07674778e+00 5.68994045e-01 3.27375948e-01 -5.92131913e-01 8.21847141e-01 -1.28785923e-01 3.06371301e-01 -4.78816479e-01 -3.69601697e-01 -2.88624346e-01 -6.47981524e-01 6.55670524e-01 9.79553699e-01 3.73622149e-01 -4.45831567e-03 -2.36646295e-01 1.16070259e+00 8.33872020e-01 4.40314740e-01 -5.16763449e-01 2.17653766e-01 5.38134992e-01 9.01851892e-01 -8.28139722e-01 -1.81373462e-01 5.00680543e-02 2.74167567e-01 -6.60142004e-02 3.40191275e-01 -5.02860546e-01 -3.78079683e-01 3.94725829e-01 -1.36679992e-01 2.16391459e-01 2.42030367e-01 -7.91412175e-01 -2.51676887e-01 -2.43361056e-01 -9.20650542e-01 5.70065737e-01 -5.09450793e-01 -5.52851677e-01 7.00487018e-01 3.54693532e-01 -7.73317516e-01 -7.29737341e-01 -4.47423100e-01 -6.41153216e-01 1.43940306e+00 -6.35285079e-01 -4.76301312e-01 -3.11661899e-01 4.31251049e-01 3.88245016e-01 -6.59538984e-01 1.13795865e+00 -1.64067864e-01 -7.05279708e-01 2.32088372e-01 1.01360060e-01 -5.64328194e-01 -3.35535221e-02 -1.10266507e+00 -4.36683651e-03 6.32821381e-01 -3.29703093e-01 4.41259772e-01 1.19701278e+00 -9.96468127e-01 -1.18291199e+00 -5.50940692e-01 1.20851421e+00 -2.10624531e-01 5.90155840e-01 -3.78565118e-02 -9.09846365e-01 2.89300412e-01 -7.23950267e-02 -6.70424879e-01 5.46762347e-01 2.29532376e-01 3.16217691e-01 1.08850807e-01 -1.20797825e+00 3.18650991e-01 2.81450450e-01 1.31762490e-01 -5.93197584e-01 9.72176120e-02 3.49697739e-01 -5.07211499e-02 -1.01883185e+00 2.53382921e-01 3.47746432e-01 -8.25895607e-01 7.72000492e-01 -5.33400416e-01 4.38064307e-01 -5.03439903e-01 1.03990816e-01 -1.18724883e+00 -2.89980650e-01 -3.90972406e-01 3.32971886e-02 1.17894769e+00 7.53700972e-01 -2.84509450e-01 5.44700265e-01 8.72253418e-01 2.74549216e-01 -7.56104410e-01 -8.73824716e-01 -5.18629134e-01 -3.13081890e-01 -3.15114975e-01 8.82378876e-01 9.51557875e-01 2.47245103e-01 3.47235441e-01 -8.11455920e-02 -5.38219288e-02 2.74403870e-01 2.27958724e-01 5.64779758e-01 -1.49748194e+00 -6.19335771e-01 -6.36821449e-01 -6.62093580e-01 3.47068876e-01 2.12020334e-03 -8.65607381e-01 -1.98113978e-01 -1.34576344e+00 -2.65071571e-01 -6.50808334e-01 -2.24882245e-01 3.78149897e-01 -6.14638627e-03 -5.17167628e-01 9.59955677e-02 -1.36461452e-01 4.97032076e-01 2.21336875e-02 8.30901325e-01 -7.14775398e-02 -9.26625729e-01 2.88294703e-01 -3.17320377e-01 2.73050189e-01 7.90682852e-01 -6.14253998e-01 -4.37271029e-01 2.58069068e-01 1.66956916e-01 4.21263129e-01 3.12458035e-02 -7.24581897e-01 -3.38533260e-02 -6.84220016e-01 3.33891183e-01 -7.40244389e-01 1.13468040e-02 -1.42603886e+00 1.09310222e+00 7.09799290e-01 -3.58454019e-01 1.74980775e-01 1.03500731e-01 1.26735747e-01 -2.94624150e-01 -6.62877738e-01 5.54623008e-01 1.46861494e-01 -7.34571278e-01 -3.87377381e-01 -4.08037066e-01 -6.81375980e-01 1.46659029e+00 -6.23229206e-01 4.74815696e-01 1.39667317e-01 -1.24019611e+00 2.99452782e-01 1.95597410e-01 3.02894145e-01 4.25450087e-01 -8.36480319e-01 -7.96396315e-01 4.05833066e-01 2.55867429e-02 -2.99661607e-03 1.89326555e-02 4.36794519e-01 -6.10454023e-01 -1.80834462e-03 -4.04483467e-01 -3.78842801e-01 -1.82974374e+00 8.44006777e-01 3.93505991e-01 -2.81735718e-01 -4.55688983e-01 6.53380334e-01 -3.63091588e-01 -8.94616097e-02 1.25476345e-01 -2.79626518e-01 -8.13512921e-01 4.95384485e-01 4.07540411e-01 7.61766136e-01 3.69850010e-01 -3.71813476e-01 -3.36590886e-01 5.41697502e-01 2.96400905e-01 -5.54952323e-02 1.72543108e+00 3.04864608e-02 -3.91649276e-01 6.79643631e-01 8.12939286e-01 -2.07565263e-01 -6.70653522e-01 -1.90549549e-02 3.22394818e-01 -4.34394926e-01 6.30015507e-02 -9.59315836e-01 -1.17932642e+00 2.20230013e-01 4.60573137e-01 2.20075116e-01 1.39782965e+00 -1.55986294e-01 -5.38721681e-01 5.39625823e-01 9.68515128e-02 -1.31968176e+00 -5.87461710e-01 5.08941710e-01 1.03849697e+00 -7.60434330e-01 5.67274690e-01 -4.61286366e-01 -8.13080668e-01 1.63444388e+00 4.41958725e-01 6.40793070e-02 5.96820951e-01 5.25594831e-01 -1.57552108e-01 -4.85960871e-01 -5.92277050e-01 4.22821231e-02 -3.91332656e-02 6.36941612e-01 4.74285543e-01 1.03851579e-01 -9.65315342e-01 5.36516666e-01 -4.25896168e-01 1.34459093e-01 4.29611623e-01 8.99132550e-01 -3.20255101e-01 -1.47611356e+00 -7.40897894e-01 5.33970118e-01 -6.77586272e-02 1.45489678e-01 -7.48629689e-01 1.23849690e+00 2.65356809e-01 1.07703841e+00 2.99297441e-02 -5.51123619e-01 7.35298038e-01 6.77968040e-02 2.47906134e-01 -5.62007844e-01 -1.25505126e+00 -1.08140670e-02 2.62848854e-01 -2.77612746e-01 -4.73490581e-02 -8.58760774e-01 -1.63494837e+00 -2.19254583e-01 -5.80440402e-01 3.69851798e-01 1.00260139e+00 7.40730762e-01 -1.94391206e-01 1.03835642e+00 8.84388864e-01 -6.38862491e-01 -7.35855773e-02 -7.94409871e-01 -7.72613049e-01 2.63940424e-01 -4.25274931e-02 -8.58321011e-01 -4.06527668e-01 6.61043599e-02]
[7.699939727783203, 4.668495178222656]
74c6b326-886a-4e58-92de-7e6b79979b1e
memory-augmented-recursive-neural-networks
1911.01545
null
https://arxiv.org/abs/1911.01545v5
https://arxiv.org/pdf/1911.01545v5.pdf
Compositional Generalization with Tree Stack Memory Units
We study compositional generalization, viz., the problem of zero-shot generalization to novel compositions of concepts in a domain. Standard neural networks fail to a large extent on compositional learning. We propose Tree Stack Memory Units (Tree-SMU) to enable strong compositional generalization. Tree-SMU is a recursive neural network with Stack Memory Units (\SMU s), a novel memory augmented neural network whose memory has a differentiable stack structure. Each SMU in the tree architecture learns to read from its stack and to write to it by combining the stacks and states of its children through gating. The stack helps capture long-range dependencies in the problem domain, thereby enabling compositional generalization. Additionally, the stack also preserves the ordering of each node's descendants, thereby retaining locality on the tree. We demonstrate strong empirical results on two mathematical reasoning benchmarks. We use four compositionality tests to assess the generalization performance of Tree-SMU and show that it enables accurate compositional generalization compared to strong baselines such as Transformers and Tree-LSTMs.
['Pranay Mundra', 'Animashree Anandkumar', 'Zhichu Lu', 'Sameer Singh', 'Forough Arabshahi']
2019-11-05
null
null
null
null
['mathematical-reasoning']
['natural-language-processing']
[ 6.86467767e-01 4.87735659e-01 -2.40808681e-01 -4.94046986e-01 -2.73561031e-01 -5.94764948e-01 6.15417182e-01 1.58945486e-01 -2.38102451e-01 7.12033093e-01 5.96545279e-01 -6.97777510e-01 1.48879051e-01 -1.44371402e+00 -1.13351285e+00 -5.70749879e-01 -4.38039601e-01 5.83874524e-01 5.20140350e-01 -3.58037025e-01 -2.89480332e-02 2.06800565e-01 -1.74546111e+00 8.73792768e-01 7.39593267e-01 7.85228848e-01 1.48280740e-01 6.85429335e-01 -5.68360627e-01 1.26739383e+00 -3.00180197e-01 -4.82824504e-01 1.38925701e-01 -4.32311565e-01 -1.20897532e+00 -5.93526304e-01 8.98610294e-01 -5.68585992e-01 -5.37675858e-01 9.89675641e-01 1.11979246e-01 6.21610105e-01 4.86796081e-01 -1.03875422e+00 -1.22675169e+00 1.59258687e+00 -1.29599525e-02 5.12225926e-01 1.93146840e-01 2.05311388e-01 1.66435778e+00 -6.02966726e-01 7.02916861e-01 1.47962248e+00 6.84398711e-01 9.00210977e-01 -1.53743601e+00 -7.38708973e-01 4.71690267e-01 1.89199783e-02 -8.46012115e-01 -1.81672588e-01 3.27908278e-01 -1.41715452e-01 1.38671255e+00 2.00273126e-01 3.36885780e-01 1.05112720e+00 1.05766878e-01 1.04303956e+00 9.68303680e-01 -1.96048245e-01 3.13657194e-01 -5.82461834e-01 7.88011968e-01 1.08525968e+00 6.73992385e-04 5.13856299e-02 -6.87727273e-01 -1.67936251e-01 6.16142511e-01 1.92253038e-01 -1.09628245e-01 -2.71449417e-01 -1.04414618e+00 7.01384604e-01 9.48006988e-01 3.81676793e-01 -6.90152422e-02 7.24698424e-01 7.43441045e-01 6.70599163e-01 8.70444924e-02 6.41863704e-01 -5.44098139e-01 1.85088277e-01 -5.63959718e-01 4.95946795e-01 6.18732095e-01 1.05221272e+00 7.74291873e-01 1.26570314e-01 -2.20139638e-01 3.71470839e-01 -2.65512347e-01 1.39356345e-01 8.45693350e-01 -1.07371545e+00 4.63249236e-01 9.06241775e-01 -5.92751265e-01 -1.59165591e-01 -4.47611123e-01 -5.73623419e-01 -7.39514410e-01 8.23206678e-02 5.09403609e-02 2.30545744e-01 -1.07551134e+00 2.31376767e+00 -7.51850158e-02 5.39154708e-01 1.81595385e-01 3.84936422e-01 1.10150123e+00 7.62060761e-01 3.63151371e-01 -1.57586660e-03 1.00743151e+00 -9.20231700e-01 -2.64534742e-01 -2.65211612e-01 1.27370620e+00 1.62093028e-01 1.45010972e+00 3.75420786e-02 -1.56041873e+00 -5.85560501e-01 -1.09884334e+00 -5.24081767e-01 -6.16744697e-01 -8.87251556e-01 7.97442675e-01 4.55888331e-01 -1.40407836e+00 9.90227878e-01 -8.80496085e-01 -1.69635460e-01 6.16684318e-01 5.40028572e-01 -3.71184081e-01 -1.81172609e-01 -1.32601368e+00 7.98237085e-01 8.02024662e-01 -3.17029089e-01 -9.62868929e-01 -1.18366838e+00 -1.17829871e+00 4.34886813e-01 2.93707341e-01 -1.25391257e+00 1.68510699e+00 -8.62095118e-01 -1.26754546e+00 9.48231936e-01 -3.51291597e-01 -9.02157724e-01 -8.47614259e-02 -7.05552101e-02 -1.92981631e-01 -3.35971415e-01 -3.95502895e-02 8.09380531e-01 4.58867371e-01 -9.85327959e-01 -8.83248270e-01 -5.09979844e-01 4.69861716e-01 -3.45021114e-02 -1.51165947e-01 -3.26618075e-01 4.10118431e-01 -6.95518374e-01 4.11146373e-01 -6.90791726e-01 -4.54850271e-02 -4.64508355e-01 -6.92440197e-02 -5.48662663e-01 7.60122716e-01 -4.72718865e-01 1.42497420e+00 -2.09291577e+00 5.10685503e-01 -7.97181502e-02 6.06126666e-01 3.51200923e-02 -2.13397145e-01 1.34818166e-01 -3.14569771e-01 2.66292661e-01 -6.06642425e-01 -1.34230211e-01 4.94471043e-01 6.85809791e-01 -8.48662853e-01 -1.15044691e-01 2.23780915e-01 1.61475408e+00 -1.16527534e+00 2.61719245e-02 -2.32519731e-01 -1.95060313e-01 -9.70782697e-01 1.30767018e-01 -8.15155447e-01 -8.53570178e-02 1.13186106e-01 4.38777626e-01 3.59334201e-01 -2.68717200e-01 2.74704188e-01 8.34137425e-02 2.53980786e-01 1.01155508e+00 -6.65824711e-01 1.80666506e+00 -6.04000926e-01 3.20866793e-01 -4.51033205e-01 -8.85195673e-01 7.14252472e-01 1.71287507e-01 -4.75429982e-01 -1.10856950e+00 -1.59651980e-01 3.62854213e-01 3.43403578e-01 -7.59962276e-02 7.44874477e-01 -6.65338457e-01 -2.92527646e-01 9.04113650e-01 2.99775332e-01 -6.44186065e-02 2.34838665e-01 4.40294057e-01 1.39560676e+00 3.12159210e-01 1.71294913e-01 -4.89014387e-01 5.33827603e-01 -4.87079263e-01 4.70888317e-01 1.03013217e+00 1.39137372e-01 -4.63569053e-02 4.79895651e-01 -8.15195978e-01 -1.01438069e+00 -1.80479622e+00 2.82940954e-01 2.02258110e+00 -2.70566016e-01 -6.23391807e-01 -6.07931197e-01 -5.82642496e-01 1.20570131e-01 1.21930873e+00 -8.41334283e-01 -7.26991415e-01 -1.18016446e+00 -3.56835753e-01 7.77762830e-01 1.19590986e+00 4.28834349e-01 -1.31353402e+00 -7.29473293e-01 1.43355757e-01 1.99079722e-01 -7.32336760e-01 -2.38989413e-01 6.65950954e-01 -1.40941894e+00 -6.23023510e-01 -1.20734394e-01 -1.05007374e+00 4.72278833e-01 -2.98884690e-01 1.57657719e+00 1.55453667e-01 1.45648703e-01 1.24047054e-02 1.55102611e-02 1.15138635e-01 -6.90594792e-01 5.89329481e-01 -3.78407426e-02 -6.27706826e-01 3.97829175e-01 -1.22175491e+00 -3.23581919e-02 -8.99731927e-03 -1.08618748e+00 1.03906557e-01 2.45414332e-01 9.01661754e-01 2.24351451e-01 -3.02408934e-02 5.95313489e-01 -1.22124183e+00 6.57505155e-01 -3.49397331e-01 -3.88134778e-01 2.42478162e-01 -2.29312956e-01 5.22556603e-01 7.24044323e-01 -3.98048013e-01 -1.06303787e+00 -5.13047934e-01 1.61802918e-01 -7.00359717e-02 7.74921551e-02 5.91964126e-01 -2.16340587e-01 3.89979005e-01 7.26960659e-01 4.06974673e-01 -1.87862590e-01 -3.90853226e-01 9.00323212e-01 1.76543407e-02 1.11146832e+00 -1.29430830e+00 5.60811996e-01 1.50960371e-01 1.83687642e-01 -4.64286506e-01 -8.96840692e-01 2.65365064e-01 -6.54293060e-01 6.23234808e-01 6.49445891e-01 -5.79324067e-01 -8.44504774e-01 4.14598703e-01 -1.11916697e+00 -9.12316263e-01 -6.36854827e-01 -2.31095269e-01 -5.69335222e-01 3.98893058e-02 -1.17999053e+00 -5.39742231e-01 -4.85174239e-01 -9.16152239e-01 7.60182738e-01 -4.37540980e-03 -6.99174762e-01 -1.27706146e+00 -1.27074927e-01 -1.24425128e-01 5.25057733e-01 -3.44589772e-03 2.12908792e+00 -7.75434911e-01 -8.89567852e-01 3.06757420e-01 -2.03715041e-01 3.96231771e-01 -1.96762517e-01 -4.69516814e-01 -1.09595156e+00 -2.27275193e-01 -1.68165565e-01 -4.28640276e-01 1.26088655e+00 1.77640840e-01 1.22329390e+00 -3.84414107e-01 -1.71311438e-01 6.78258598e-01 9.21329856e-01 1.57984626e-02 4.96647686e-01 3.22864830e-01 5.88321507e-01 2.03784704e-01 -3.26498121e-01 -2.10220031e-02 4.46613342e-01 4.73753139e-02 4.64216799e-01 4.85528916e-01 -3.01870584e-01 -5.55407703e-01 4.70077544e-01 7.04358876e-01 -1.12480111e-01 5.65665476e-02 -1.05699944e+00 5.04610419e-01 -1.61824131e+00 -7.99786031e-01 1.47019640e-01 2.21368790e+00 1.04390693e+00 4.80961412e-01 -9.06485170e-02 6.04683161e-02 4.74671423e-01 1.93891585e-01 -5.08832455e-01 -8.02776694e-01 -2.98993528e-01 1.07507730e+00 2.75620401e-01 6.49586260e-01 -7.73917317e-01 1.14755595e+00 6.50270700e+00 2.83074886e-01 -8.00766289e-01 2.69487172e-01 2.78492272e-01 -3.35405439e-01 -7.87617922e-01 -1.21959448e-01 -7.30181992e-01 7.25645497e-02 1.10306418e+00 -2.49192700e-01 6.34934962e-01 5.52798092e-01 -8.01175356e-01 2.78069794e-01 -1.93516684e+00 4.26666081e-01 -2.51965582e-01 -1.57149339e+00 5.75673699e-01 -2.44738027e-01 9.60416555e-01 1.99069485e-01 3.35369587e-01 8.62816095e-01 7.41483688e-01 -1.23728657e+00 6.82585180e-01 2.82923847e-01 4.32328910e-01 -6.69667304e-01 5.22223949e-01 2.39235669e-01 -1.13932014e+00 -4.56576735e-01 -2.98522413e-01 -6.13009989e-01 -3.98176424e-02 3.08982283e-02 -5.80797255e-01 2.73994267e-01 5.13396919e-01 6.70057118e-01 -5.78885674e-01 3.32965732e-01 -5.58061600e-01 5.15976906e-01 -1.55438602e-01 8.24453831e-02 5.91358304e-01 1.03716917e-01 2.93264389e-01 1.16250467e+00 1.86176047e-01 9.39692855e-02 9.21047926e-02 1.26209331e+00 -7.14172781e-01 -4.09063935e-01 -6.00577474e-01 -3.50797474e-01 5.37065029e-01 4.99827445e-01 -2.29718983e-01 -7.83802927e-01 -3.37318927e-01 8.50752413e-01 6.79682314e-01 4.58530843e-01 -5.52549064e-01 -1.29806489e-01 8.03093851e-01 2.95202821e-01 3.38035554e-01 -1.57098457e-01 -5.28975248e-01 -8.17011058e-01 -8.43350440e-02 -8.11884344e-01 9.49158490e-01 -8.83683324e-01 -1.16234171e+00 4.38377440e-01 7.01136217e-02 -2.15038136e-01 -2.37800255e-01 -8.54272664e-01 -7.14859366e-01 1.03785992e+00 -1.18468964e+00 -1.43018067e+00 1.27892286e-01 3.83501679e-01 4.01911288e-01 -2.13228017e-01 1.11904943e+00 -5.18118814e-02 -2.79758900e-01 7.04394281e-01 -3.82983744e-01 1.83722004e-01 6.15872480e-02 -1.41830981e+00 1.27640331e+00 6.80370510e-01 -6.06007241e-02 1.39773691e+00 4.98551458e-01 -6.11157596e-01 -1.26948726e+00 -1.16160357e+00 1.13401687e+00 -7.18443573e-01 1.12626708e+00 -4.24290985e-01 -1.47781050e+00 1.63426268e+00 1.50022551e-01 -1.43203977e-03 5.87756753e-01 4.75346774e-01 -1.19594371e+00 8.12031180e-02 -6.83878303e-01 8.10245216e-01 1.76723838e+00 -9.81481194e-01 -1.48505270e+00 -1.66035980e-01 1.38330078e+00 -4.42441076e-01 -7.49446630e-01 4.28134978e-01 6.44102871e-01 -9.15006280e-01 8.60827565e-01 -1.37018347e+00 8.27217042e-01 -3.35882902e-02 -4.82070267e-01 -1.31186461e+00 -6.76521063e-01 -4.10099566e-01 -6.24800205e-01 8.08611453e-01 5.34796715e-01 -9.91951883e-01 8.65834773e-01 6.04325056e-01 -4.66196567e-01 -7.84386456e-01 -8.34418118e-01 -8.64215195e-01 6.33004129e-01 -4.17563140e-01 1.14888191e+00 8.59341860e-01 1.99211150e-01 7.59792805e-01 3.36426735e-01 1.11816585e-01 4.44817632e-01 3.58485281e-01 4.82020915e-01 -1.21263063e+00 -3.72048736e-01 -9.75271344e-01 -4.42803323e-01 -1.20684123e+00 8.23360920e-01 -1.66813707e+00 -3.00807744e-01 -1.33589792e+00 3.97118479e-01 -1.62083879e-01 -5.47311902e-01 6.21059835e-01 -2.58180946e-01 -7.60079548e-02 -1.53534831e-02 -2.39786297e-01 -4.88325506e-01 4.30065125e-01 1.10197258e+00 -5.18322349e-01 -5.08527597e-03 -2.85047740e-01 -6.94361627e-01 6.98715687e-01 7.12624431e-01 -2.52245039e-01 -6.37982249e-01 -7.67835140e-01 5.50850689e-01 -3.64527911e-01 5.30927002e-01 -8.40219915e-01 3.14732820e-01 1.34342015e-01 3.14075984e-02 -5.27007401e-01 1.36489749e-01 -4.37095165e-01 -7.42778853e-02 7.07716942e-01 -8.25209498e-01 4.19546038e-01 3.28581214e-01 2.59337366e-01 3.01376320e-02 -2.07414076e-01 6.99345887e-01 -6.09332263e-01 -9.92360413e-01 2.32386217e-01 -1.13975264e-01 3.49417150e-01 3.71186674e-01 -5.84802665e-02 -6.10675812e-01 -1.29463887e-02 -8.93671930e-01 1.80847943e-01 5.12696683e-01 3.89051974e-01 4.59828168e-01 -1.33979857e+00 -4.31854695e-01 3.23807985e-01 1.60063535e-01 3.64074886e-01 1.57839939e-01 5.49514115e-01 -2.86223620e-01 3.84194553e-01 -2.37021714e-01 -2.40707293e-01 -1.03838217e+00 9.49396372e-01 5.60817361e-01 -1.98983118e-01 -8.32416534e-01 1.36284065e+00 8.47336769e-01 -1.03719950e+00 2.97189891e-01 -1.29741096e+00 5.30657351e-01 -3.78369242e-01 7.08506346e-01 4.35478747e-01 3.33576947e-02 -2.17140093e-01 -5.85856959e-02 5.73586076e-02 -2.46661007e-01 1.20805942e-01 1.23036921e+00 3.05951804e-01 -8.30841362e-01 8.35269749e-01 1.17415893e+00 -6.14013076e-01 -5.68051100e-01 -7.65180826e-01 5.33162475e-01 -5.04792482e-02 -1.74949110e-01 -7.21700490e-01 -5.51731348e-01 1.00438321e+00 9.17408522e-03 -1.91629417e-02 7.24010587e-01 2.15813801e-01 1.12732530e+00 9.23462331e-01 3.42261583e-01 -6.57698095e-01 4.72162813e-02 1.07075858e+00 6.93267584e-01 -5.55209994e-01 -3.04279774e-01 -7.23900571e-02 -8.35077465e-02 1.02916193e+00 6.97433710e-01 -1.39871106e-01 2.41637275e-01 3.17525417e-01 -4.64330375e-01 -6.86275363e-02 -1.10530114e+00 -4.39621769e-02 8.09311345e-02 3.81847233e-01 5.52255869e-01 3.50906223e-01 2.59930581e-01 8.36532116e-01 -5.74683487e-01 9.71175358e-02 3.11196953e-01 1.13518751e+00 -6.78191841e-01 -9.40269232e-01 2.61700395e-02 5.72166860e-01 -6.92909583e-02 -5.83571851e-01 -3.25594872e-01 6.86060846e-01 1.02382496e-01 1.51330292e-01 3.34080696e-01 -3.49112719e-01 3.18576038e-01 8.67715716e-01 1.00069666e+00 -1.07509708e+00 -6.90254807e-01 -7.71180332e-01 8.88425112e-02 -6.05107665e-01 -2.08839346e-02 -4.64940727e-01 -1.76130962e+00 -6.47365451e-01 2.10242212e-01 -2.60051489e-02 -3.06818150e-02 9.21358824e-01 7.42334351e-02 8.99500072e-01 -8.18773955e-02 -4.11476314e-01 -8.42932165e-01 -8.40586603e-01 -5.82737148e-01 5.73577285e-01 4.63729858e-01 -4.19992089e-01 -1.51003912e-01 -5.57042696e-02]
[9.476664543151855, 7.379059791564941]
d0423a91-9408-4eea-88fb-50f76c4ef541
on-the-importance-of-building-high-quality
2202.06649
null
https://arxiv.org/abs/2202.06649v1
https://arxiv.org/pdf/2202.06649v1.pdf
On the Importance of Building High-quality Training Datasets for Neural Code Search
The performance of neural code search is significantly influenced by the quality of the training data from which the neural models are derived. A large corpus of high-quality query and code pairs is demanded to establish a precise mapping from the natural language to the programming language. Due to the limited availability, most widely-used code search datasets are established with compromise, such as using code comments as a replacement of queries. Our empirical study on a famous code search dataset reveals that over one-third of its queries contain noises that make them deviate from natural user queries. Models trained through noisy data are faced with severe performance degradation when applied in real-world scenarios. To improve the dataset quality and make the queries of its samples semantically identical to real user queries is critical for the practical usability of neural code search. In this paper, we propose a data cleaning framework consisting of two subsequent filters: a rule-based syntactic filter and a model-based semantic filter. This is the first framework that applies semantic query cleaning to code search datasets. Experimentally, we evaluated the effectiveness of our framework on two widely-used code search models and three manually-annotated code retrieval benchmarks. Training the popular DeepCS model with the filtered dataset from our framework improves its performance by 19.2% MRR and 21.3% Answer@1, on average with the three validation benchmarks.
['Xiaoning Du', 'Yan Liu', 'Li Li', 'Zhensu Sun']
2022-02-14
null
null
null
null
['code-search', 'code-search']
['computer-code', 'computer-vision']
[-4.68407832e-02 -4.13737774e-01 -1.97102159e-01 -3.87204707e-01 -1.01851559e+00 -5.78357279e-01 3.21817130e-01 2.29019478e-01 -6.29800975e-01 2.20539793e-01 -1.31210431e-01 -3.21462005e-01 -3.77660133e-02 -7.17604935e-01 -9.23902154e-01 -1.91438228e-01 3.60611141e-01 2.18938306e-01 5.27875423e-01 -3.49340767e-01 4.32990402e-01 -1.91720530e-01 -1.94299269e+00 5.25727808e-01 1.42577565e+00 9.99226689e-01 6.76112533e-01 3.77408683e-01 -7.67741561e-01 7.99275696e-01 -7.58266747e-01 -5.31921625e-01 1.31193727e-01 -2.42983669e-01 -8.59365821e-01 -6.85472071e-01 3.04881036e-01 -1.71219222e-02 2.21155491e-03 1.56199336e+00 3.45215023e-01 -1.32848453e-02 1.89615175e-01 -1.09386230e+00 -9.71096039e-01 8.87798727e-01 -1.37682617e-01 1.84316151e-02 4.55797017e-01 -1.65077493e-01 9.69702125e-01 -1.12560558e+00 7.14859605e-01 9.64454234e-01 7.25249648e-01 7.00347900e-01 -1.13018548e+00 -9.23591673e-01 -1.20682836e-01 2.74034172e-01 -1.54891539e+00 -5.62138736e-01 5.43064773e-01 -6.16061628e-01 1.25807893e+00 3.20198894e-01 1.22227453e-01 1.06577849e+00 1.56376660e-02 5.87378561e-01 3.63327414e-01 -4.93719608e-01 2.91993946e-01 3.98731470e-01 2.69506782e-01 7.24710405e-01 2.36345828e-01 -7.10592046e-02 -2.80829012e-01 -4.69007969e-01 4.48721834e-02 6.05814531e-02 -5.10692775e-01 -3.69744480e-01 -7.41899490e-01 8.23412418e-01 5.26399136e-01 5.88237286e-01 -1.08367335e-02 2.96672910e-01 8.02397609e-01 4.53595340e-01 8.56337771e-02 7.30567634e-01 -6.02373183e-01 -3.53717297e-01 -9.86998618e-01 3.10891509e-01 8.15036654e-01 1.38553214e+00 7.73664474e-01 -1.81054264e-01 -1.69336513e-01 1.25810659e+00 4.00273800e-01 6.48930728e-01 9.37355876e-01 -7.02187896e-01 5.82179725e-01 1.06184554e+00 -1.80028677e-01 -1.19873023e+00 9.15071815e-02 -6.21756315e-01 -4.34875160e-01 -2.23668799e-01 1.87240615e-01 3.17991942e-01 -6.64931238e-01 1.73534822e+00 -2.21662655e-01 -2.41919339e-01 8.60687643e-02 6.40859544e-01 1.08302629e+00 4.17935282e-01 3.29291448e-02 2.70399958e-01 1.22022617e+00 -9.91130888e-01 -4.94252175e-01 -3.57130557e-01 9.53943551e-01 -8.56184423e-01 1.65352666e+00 3.12027991e-01 -6.40307844e-01 -6.58850610e-01 -9.80008483e-01 -1.02325857e-01 -5.67689598e-01 3.17588925e-01 2.67874300e-01 4.91810054e-01 -1.01352811e+00 2.54421622e-01 -6.05286181e-01 -5.03041029e-01 1.79882750e-01 1.36419281e-01 -2.11160451e-01 -2.05004305e-01 -1.11998963e+00 5.99166095e-01 4.27512825e-01 -3.04922342e-01 -8.13367009e-01 -7.71779954e-01 -7.50392497e-01 4.49728966e-01 4.58047092e-01 -2.98074931e-01 1.36555696e+00 -1.05225945e+00 -8.22916210e-01 9.79219794e-01 -1.59415901e-01 -3.47223729e-01 2.84562111e-01 -2.24018127e-01 -5.50558448e-01 -2.44621396e-01 2.94709772e-01 3.06491405e-01 6.27911031e-01 -1.33316875e+00 -5.83919525e-01 -1.49813831e-01 1.64865628e-01 -3.78137439e-01 -4.62395400e-01 1.68719202e-01 -9.51027632e-01 -6.54156506e-01 -2.16730267e-01 -8.61895621e-01 1.60800084e-01 -1.61005199e-01 -2.57863432e-01 -3.28644603e-01 5.38703322e-01 -6.26421690e-01 1.71914375e+00 -2.42527437e+00 -1.24410562e-01 1.07368261e-01 1.15924619e-01 4.14105684e-01 -4.30556953e-01 4.45661485e-01 2.29900721e-02 4.00224239e-01 -4.23466504e-01 -1.33607656e-01 1.08951680e-01 8.29146728e-02 -5.60299337e-01 -6.39500022e-02 5.02288435e-03 8.51583540e-01 -9.76829708e-01 -4.81222749e-01 -3.68773580e-01 1.56247228e-01 -9.68566656e-01 3.60157490e-01 -3.56274366e-01 -1.97200313e-01 -6.08732402e-01 5.95032632e-01 3.60855877e-01 -3.15015167e-01 -2.84410596e-01 7.19396165e-03 1.67583153e-01 2.51148880e-01 -7.94449627e-01 2.19258261e+00 -8.14010799e-01 5.75230598e-01 -3.62023637e-02 -7.32839763e-01 1.03309309e+00 7.00928271e-02 1.31007180e-01 -1.09905457e+00 1.48028294e-02 5.61219990e-01 -1.68734312e-01 -8.89190197e-01 6.36521399e-01 4.97064561e-01 -2.97390074e-01 3.31308812e-01 3.42860669e-02 -1.51556373e-01 3.53519171e-01 2.14364216e-01 1.48400843e+00 -6.60117120e-02 1.12534277e-02 -5.65333962e-01 8.23477089e-01 1.99268118e-01 6.19041562e-01 9.49459255e-01 -1.43181100e-01 5.30978978e-01 4.11362857e-01 -4.15535241e-01 -7.25485146e-01 -6.07944489e-01 -1.40060082e-01 1.39482236e+00 1.72673896e-01 -7.86089957e-01 -1.00335968e+00 -9.11947608e-01 7.16486350e-02 7.95732498e-01 -5.31781316e-01 -6.01734400e-01 -5.02414644e-01 -4.14644510e-01 8.48763108e-01 1.93387970e-01 4.31254983e-01 -1.10199845e+00 -7.56722093e-01 1.38471588e-01 -3.32638204e-01 -8.85494530e-01 -5.67475557e-01 2.36496598e-01 -4.10324901e-01 -1.51917124e+00 -1.36922583e-01 -9.18888092e-01 7.11732447e-01 1.40536442e-01 1.55048537e+00 9.50709164e-01 -3.18907574e-02 6.36863187e-02 -7.22829044e-01 -2.50085562e-01 -9.77471948e-01 4.61800158e-01 -4.16259766e-01 -3.53934914e-01 7.86787331e-01 -1.81190491e-01 -4.27692950e-01 3.60189527e-01 -1.34330082e+00 -2.75072396e-01 5.14615953e-01 7.90890157e-01 2.46556625e-01 -2.28746012e-02 4.88741726e-01 -9.46264982e-01 1.01843369e+00 -5.72907925e-01 -8.41956675e-01 5.14497638e-01 -8.21319580e-01 4.32512254e-01 8.52537036e-01 -3.13234866e-01 -9.15647626e-01 -9.26383492e-03 -1.96010888e-01 -2.99491584e-01 -6.64947778e-02 7.82245874e-01 5.45758121e-02 -1.14530288e-02 1.24557054e+00 3.31461668e-01 -2.99765587e-01 -7.42215872e-01 1.47301808e-01 1.02449572e+00 7.01754391e-01 -9.24790740e-01 6.57050669e-01 8.87980163e-02 -7.54845619e-01 -3.42273891e-01 -4.79392380e-01 -5.28064430e-01 -2.59398609e-01 1.45498514e-01 6.91043973e-01 -7.72051096e-01 -2.94627577e-01 2.28610724e-01 -1.28498983e+00 -2.10143149e-01 8.79680738e-02 1.20116450e-01 -1.59220412e-01 3.28664362e-01 -3.88280213e-01 -3.78577948e-01 -5.18703640e-01 -1.75277555e+00 1.08756578e+00 -4.00231630e-02 -2.68777817e-01 -4.59261000e-01 1.47932380e-01 2.51230776e-01 7.90251791e-01 -2.34563634e-01 1.35776043e+00 -9.88049567e-01 -4.53670532e-01 -1.91750512e-01 -2.09190488e-01 5.25041997e-01 4.98534776e-02 2.07806304e-01 -1.07638752e+00 -2.72070497e-01 1.42020434e-01 -3.53866965e-01 7.61396706e-01 -2.24080443e-01 1.41735077e+00 -8.70143175e-02 -3.80026877e-01 5.71137071e-01 1.77026010e+00 5.85224211e-01 3.99084419e-01 4.00008887e-01 5.36721885e-01 3.31818193e-01 4.76200521e-01 1.27591133e-01 1.98087677e-01 6.67617321e-01 5.25500357e-01 5.03539801e-01 7.37040937e-02 -3.47201705e-01 2.33232573e-01 1.30262947e+00 6.77963614e-01 3.85160148e-02 -1.38086092e+00 7.92194426e-01 -1.84964728e+00 -5.73616445e-01 -5.59969246e-02 2.20643449e+00 1.15577638e+00 1.54674321e-01 -5.12659669e-01 -1.26095220e-01 4.88375813e-01 -3.27825010e-01 -5.64200759e-01 -2.62641668e-01 2.80026078e-01 1.06539048e-01 2.87767529e-01 2.55972862e-01 -8.91171455e-01 7.79919446e-01 5.43502808e+00 1.05975914e+00 -1.02020025e+00 1.56144738e-01 1.73050061e-01 9.50631350e-02 -5.69153011e-01 8.39103386e-02 -6.14679158e-01 8.58791709e-01 8.83100152e-01 -3.05950135e-01 6.76985323e-01 1.33801317e+00 -1.42378181e-01 -6.51601404e-02 -1.30270660e+00 1.20419896e+00 1.27308190e-01 -1.12906134e+00 -2.72773746e-02 -2.98886150e-01 6.47130787e-01 3.44726831e-01 -3.50895703e-01 8.40382218e-01 3.44205767e-01 -9.25363719e-01 9.98477638e-01 5.99598646e-01 6.72309041e-01 -4.97770309e-01 8.10238719e-01 4.40931559e-01 -1.14373648e+00 -3.35748792e-01 -3.60764921e-01 3.99939775e-01 -4.98090953e-01 6.70960009e-01 -4.90216851e-01 2.82953829e-01 1.05800867e+00 5.49049497e-01 -1.08589303e+00 1.15589643e+00 -4.88051511e-02 2.95426369e-01 -1.13386117e-01 -1.90748602e-01 4.52770181e-02 1.71792626e-01 2.08653569e-01 1.32180858e+00 3.28160226e-01 -4.64439452e-01 1.10836951e-02 1.51611459e+00 -4.31794196e-01 2.37326875e-01 -6.62764192e-01 -8.05519819e-02 8.27136934e-01 9.29711401e-01 -4.23477978e-01 -3.83047819e-01 -6.28436029e-01 6.69805110e-01 4.42880750e-01 4.46653455e-01 -9.01862025e-01 -8.24507415e-01 7.85890162e-01 -1.85029089e-01 1.95621148e-01 2.78217494e-01 3.68839130e-02 -1.25979888e+00 5.95054388e-01 -1.34057260e+00 2.38936931e-01 -6.36026144e-01 -1.12267029e+00 8.90307367e-01 -1.34549797e-01 -1.00544059e+00 -1.93632707e-01 -2.63603270e-01 -3.40044111e-01 7.96987116e-01 -1.39999688e+00 -5.29017031e-01 -6.25320971e-01 3.24079782e-01 7.11182773e-01 -4.73785579e-01 8.86932850e-01 8.92474592e-01 -4.37618047e-01 7.25565732e-01 1.75666198e-01 5.12835264e-01 6.36530757e-01 -1.02395535e+00 6.30066454e-01 9.97123301e-01 1.80431589e-01 1.16161156e+00 5.59657216e-01 -5.57314634e-01 -1.46727788e+00 -1.04965544e+00 7.98421443e-01 -5.30680180e-01 5.12966156e-01 -5.34229338e-01 -1.40572214e+00 3.35805029e-01 1.15030715e-02 8.56265351e-02 4.51357037e-01 -2.02206701e-01 -5.53986371e-01 -2.73029387e-01 -9.51958299e-01 3.39978784e-01 1.02247286e+00 -1.04119897e+00 -7.55232632e-01 1.08433291e-01 9.76136625e-01 -3.28653067e-01 -8.12557101e-01 4.33546662e-01 3.59046757e-01 -9.83146131e-01 7.22772300e-01 -5.67612708e-01 3.98957193e-01 -4.61211234e-01 -3.83391559e-01 -1.18914974e+00 9.26297754e-02 -1.50415659e-01 2.87640899e-01 1.29273021e+00 4.75407302e-01 -3.59589130e-01 5.24883151e-01 7.51756370e-01 -1.13370813e-01 -6.34294510e-01 -7.40936518e-01 -7.52418041e-01 3.66543606e-02 -7.25438535e-01 8.95664632e-01 1.02470934e+00 -1.17952019e-01 1.30157009e-01 3.87053937e-01 -1.28078163e-01 1.48842379e-01 2.00868219e-01 7.69495010e-01 -1.27790093e+00 -3.33952457e-01 -8.13508272e-01 -1.74429670e-01 -8.70834351e-01 3.90824556e-01 -1.21055090e+00 2.58154571e-01 -1.19968164e+00 3.59918594e-01 -5.87452948e-01 -2.05788195e-01 4.14406776e-01 -1.56036347e-01 -3.00789088e-01 7.59810731e-02 3.29613209e-01 -5.68048298e-01 5.43724358e-01 5.81644952e-01 -4.24198002e-01 -8.68018493e-02 -2.39144221e-01 -6.45863533e-01 5.23289561e-01 5.49740911e-01 -9.27758098e-01 -7.06548452e-01 -9.99055088e-01 8.32376540e-01 -2.78131247e-01 1.21964447e-01 -1.08362436e+00 4.72776800e-01 1.45614535e-01 -3.92407417e-01 -1.73826680e-01 -3.01999331e-01 -1.15756226e+00 1.59883901e-01 5.74255049e-01 -6.78110540e-01 3.04771334e-01 2.21065387e-01 4.75269735e-01 -4.57304448e-01 -7.82367706e-01 6.79121554e-01 -2.49153510e-01 -9.04668868e-01 -9.39721987e-02 7.48512382e-03 5.90232253e-01 6.33233130e-01 1.20516166e-01 -5.73662758e-01 -2.63117417e-03 -1.85166702e-01 2.36578375e-01 6.79731905e-01 1.05034721e+00 5.21212995e-01 -1.23011923e+00 -4.14039046e-01 3.86679202e-01 7.50727236e-01 1.38099989e-04 -2.82145023e-01 4.69916552e-01 -6.21148884e-01 5.89892149e-01 1.98621407e-01 -6.28475547e-01 -1.10432410e+00 7.45329559e-01 5.22973776e-01 -2.87996102e-02 -1.94104970e-01 6.94401681e-01 -8.11717585e-02 -7.50777721e-01 6.35042787e-01 -9.33969676e-01 -4.65103053e-02 -2.76128143e-01 4.85076100e-01 5.69412000e-02 5.47711015e-01 -4.78950739e-01 -5.29173791e-01 4.63920832e-01 7.85932317e-02 4.23571289e-01 1.13737917e+00 2.10864261e-01 -4.75747675e-01 2.01891184e-01 1.60388243e+00 7.50486702e-02 -6.44380331e-01 -4.65806574e-01 6.15593672e-01 -5.29594004e-01 -7.66559020e-02 -7.66098380e-01 -1.29430556e+00 6.64314568e-01 5.79783022e-01 4.15502638e-01 1.05560625e+00 3.31906922e-04 7.91912973e-01 7.79898107e-01 5.70444524e-01 -1.07252371e+00 -1.95759922e-01 6.04530871e-01 8.38406324e-01 -1.32278872e+00 -5.10156214e-01 -1.46223456e-01 -1.50245383e-01 9.01285172e-01 8.19812477e-01 6.36037067e-02 6.43723845e-01 2.64696419e-01 1.81336403e-01 -4.90306586e-01 -8.15817356e-01 2.97068665e-03 4.97757286e-01 2.10229218e-01 4.47236687e-01 -2.81504542e-01 -2.15738699e-01 8.33617151e-01 -1.04762219e-01 2.56976375e-04 2.62591302e-01 9.61954892e-01 -4.95414734e-01 -1.13519394e+00 -4.43850122e-02 5.92685163e-01 -3.83563310e-01 -2.79535651e-01 -3.49555463e-01 5.86506844e-01 6.31026030e-02 9.63481069e-01 -1.59888119e-01 -5.52882373e-01 7.07832754e-01 3.90209168e-01 -2.77742028e-01 -7.88432837e-01 -9.58584905e-01 -5.35934448e-01 -1.74265236e-01 -9.92967844e-01 -1.61383346e-01 -2.70508379e-01 -1.37487042e+00 -8.45042802e-03 -4.62603956e-01 4.46310639e-01 8.70182395e-01 7.59150922e-01 7.12568939e-01 4.18212742e-01 2.08466560e-01 -4.59261378e-03 -7.38926351e-01 -8.00772548e-01 9.51740369e-02 8.59383941e-01 3.84838462e-01 -5.02875328e-01 -4.02516603e-01 1.55816987e-01]
[7.510554313659668, 8.042950630187988]
01d64a40-665f-4dc3-ad43-600ffe161a94
pointflowhop-green-and-interpretable-scene
2302.14193
null
https://arxiv.org/abs/2302.14193v1
https://arxiv.org/pdf/2302.14193v1.pdf
PointFlowHop: Green and Interpretable Scene Flow Estimation from Consecutive Point Clouds
An efficient 3D scene flow estimation method called PointFlowHop is proposed in this work. PointFlowHop takes two consecutive point clouds and determines the 3D flow vectors for every point in the first point cloud. PointFlowHop decomposes the scene flow estimation task into a set of subtasks, including ego-motion compensation, object association and object-wise motion estimation. It follows the green learning (GL) pipeline and adopts the feedforward data processing path. As a result, its underlying mechanism is more transparent than deep-learning (DL) solutions based on end-to-end optimization of network parameters. We conduct experiments on the stereoKITTI and the Argoverse LiDAR point cloud datasets and demonstrate that PointFlowHop outperforms deep-learning methods with a small model size and less training time. Furthermore, we compare the Floating Point Operations (FLOPs) required by PointFlowHop and other learning-based methods in inference, and show its big savings in computational complexity.
['C. -C. Jay Kuo', 'Shan Liu', 'Jiahao Gu', 'Pranav Kadam']
2023-02-27
null
null
null
null
['motion-compensation', 'scene-flow-estimation']
['computer-vision', 'computer-vision']
[-3.68087530e-01 -3.31603885e-01 -4.24157292e-01 -4.37887907e-01 -4.40787137e-01 -3.29727173e-01 4.12363201e-01 -1.31435648e-01 -4.53963101e-01 3.60308856e-01 -1.64126962e-01 -4.17991430e-01 -8.25262815e-02 -8.11686456e-01 -8.52487504e-01 -5.31908929e-01 -3.14204752e-01 5.51642179e-01 3.97649556e-01 2.81972110e-01 5.12327015e-01 9.02266502e-01 -1.24205649e+00 -5.71071506e-02 7.91833997e-01 1.03770900e+00 2.61677206e-01 9.11217690e-01 -3.49764884e-01 1.04335368e+00 -1.63077086e-01 -1.77454621e-01 4.65358526e-01 2.78275847e-01 -8.20914447e-01 -1.54197752e-01 1.02120423e+00 -8.77881944e-01 -6.25784516e-01 6.44269764e-01 5.17114222e-01 2.94563681e-01 4.19505060e-01 -1.60293484e+00 -6.06028214e-02 1.57019570e-01 -8.48115981e-01 3.93304527e-01 -2.01101750e-01 5.72725654e-01 8.54280770e-01 -1.28838897e+00 5.81745863e-01 1.50057983e+00 7.24849522e-01 2.49563724e-01 -1.10313654e+00 -7.58258581e-01 3.03509772e-01 3.96647006e-01 -1.35848737e+00 -3.66391093e-01 7.59905219e-01 -6.95982933e-01 1.24937880e+00 -1.68741375e-01 9.10835743e-01 7.53050685e-01 1.95637807e-01 9.85117078e-01 3.46581131e-01 -3.23686711e-02 4.67110693e-01 -4.36618954e-01 -1.51582947e-02 9.95199263e-01 3.24154496e-01 3.47180665e-01 -6.24328911e-01 1.55445322e-01 1.21419644e+00 1.40981466e-01 -1.40548721e-01 -6.97917998e-01 -1.36213207e+00 7.63536751e-01 9.11871135e-01 -2.62153715e-01 -3.03363025e-01 9.87229586e-01 3.12858224e-01 -2.46362179e-03 5.86038172e-01 -5.26871644e-02 -6.35890424e-01 -1.52062401e-01 -1.10537636e+00 4.30439472e-01 7.03366518e-01 1.19268048e+00 1.23249412e+00 2.77003050e-01 3.39419432e-02 3.53147715e-01 6.90535486e-01 6.60439014e-01 -1.19803078e-01 -1.79382277e+00 6.94766462e-01 5.34866393e-01 5.20805530e-02 -1.14579296e+00 -3.99038702e-01 -3.25693786e-01 -8.17584455e-01 6.06758595e-01 4.46348459e-01 -3.06665212e-01 -7.85423100e-01 1.29941499e+00 6.16298914e-01 8.24559808e-01 -3.10489625e-01 1.08121383e+00 8.46219361e-01 7.79733062e-01 -2.03038789e-02 1.73070580e-01 7.21936584e-01 -1.26292503e+00 -2.77256906e-01 -4.45639640e-01 8.03749144e-01 -3.90775084e-01 8.56406391e-01 1.81945741e-01 -1.16868532e+00 -7.44119823e-01 -9.76097584e-01 -5.73099136e-01 -6.61965981e-02 5.10511454e-03 1.01493800e+00 1.65665582e-01 -9.89291310e-01 8.38241041e-01 -1.25203991e+00 -5.23204133e-02 8.98195982e-01 3.96571964e-01 -1.40613824e-01 1.98801830e-02 -4.52273607e-01 6.30005658e-01 2.57221162e-01 3.29439580e-01 -1.02379704e+00 -1.37575054e+00 -1.03687322e+00 1.97126433e-01 3.07240337e-01 -1.21876907e+00 1.32243693e+00 -2.88931698e-01 -1.41674960e+00 7.26399958e-01 -3.87913465e-01 -5.33756435e-01 7.19751656e-01 -6.01508796e-01 3.59719932e-01 2.55843610e-01 1.67590708e-01 1.26111269e+00 7.86047399e-01 -1.03128970e+00 -1.05356765e+00 -3.90329093e-01 -7.80529752e-02 -6.97420165e-03 1.64293975e-01 -5.30600965e-01 -8.22096944e-01 -2.20440492e-01 1.66960061e-01 -9.55896199e-01 -3.90699625e-01 7.68553972e-01 -3.38905483e-01 -2.48650298e-01 9.45439398e-01 -2.08710104e-01 9.18673158e-01 -1.95876229e+00 -8.42894763e-02 2.63973251e-02 5.40081084e-01 1.99358627e-01 1.63224246e-02 1.42756868e-02 1.45406440e-01 -1.64495558e-02 -2.60591447e-01 -4.76364136e-01 5.99143319e-02 2.30717137e-01 -4.98118043e-01 6.02171838e-01 2.13017911e-01 1.02675545e+00 -9.83038604e-01 -7.42375255e-01 7.91580498e-01 5.75134099e-01 -1.01400054e+00 -8.58824626e-02 -2.43578330e-01 3.05272073e-01 -4.67485130e-01 6.48542762e-01 9.06313002e-01 -4.41387415e-01 -3.68725568e-01 -3.58452439e-01 -4.05575901e-01 2.09365442e-01 -1.19217467e+00 2.22199512e+00 -5.09447277e-01 1.01811278e+00 -1.09453052e-02 -5.10727525e-01 8.05032611e-01 -1.91336274e-01 8.38130653e-01 -4.04876322e-01 1.13545641e-01 1.21947385e-01 -2.81831205e-01 -5.12151599e-01 4.60407734e-01 2.67395169e-01 3.42001200e-01 1.14975870e-01 1.55018911e-01 -2.50347644e-01 1.12564057e-01 1.45277455e-01 1.15932512e+00 6.19358361e-01 -1.34572849e-01 -7.44278505e-02 5.10465503e-01 3.87900531e-01 7.11279869e-01 6.60949767e-01 -4.91936237e-01 3.68926913e-01 3.06855798e-01 -7.59899616e-01 -8.85666668e-01 -1.17922103e+00 -1.29894940e-02 7.65022099e-01 4.65151638e-01 -4.23785061e-01 -4.72118407e-01 -6.42612040e-01 5.30714512e-01 6.61199272e-01 -1.98629886e-01 1.22676224e-01 -9.86443281e-01 -3.12353402e-01 3.03570777e-01 7.77846813e-01 8.09369147e-01 -8.04993868e-01 -1.00243568e+00 1.15814112e-01 -4.67847064e-02 -1.32673991e+00 -4.85860288e-01 -9.06509385e-02 -1.28566861e+00 -1.12721443e+00 -3.54872793e-01 -5.95701635e-01 4.85652268e-01 6.04881585e-01 1.25069129e+00 1.85110271e-02 -2.65608251e-01 2.66013354e-01 1.22932285e-01 -4.42861021e-01 2.45926052e-01 2.32938528e-01 -3.01709235e-01 -3.35593939e-01 6.22495592e-01 -6.96934938e-01 -8.54359448e-01 1.51737064e-01 -3.14449012e-01 1.19365342e-01 4.15775776e-01 3.84349227e-01 7.25577354e-01 -2.55987197e-01 -9.62883886e-03 -4.26307917e-01 -5.23024574e-02 -3.08452457e-01 -1.25744939e+00 -3.61654907e-01 -4.51410592e-01 -7.29026180e-03 2.39320487e-01 1.10152423e-01 -9.46932673e-01 4.37548280e-01 7.02914968e-02 -1.23637652e+00 -9.27241296e-02 1.25847742e-01 4.99465615e-02 -2.63207972e-01 4.90370095e-01 -1.08575635e-01 -5.44764362e-02 -4.24525410e-01 5.40610075e-01 1.26420453e-01 8.28881264e-01 -2.41601929e-01 1.05378401e+00 9.29327488e-01 3.88125747e-01 -8.86285484e-01 -8.59052360e-01 -5.75623989e-01 -1.18697321e+00 -5.25532186e-01 1.00616086e+00 -1.21144474e+00 -1.37838078e+00 3.56018573e-01 -1.50470424e+00 -5.78026056e-01 -3.84988785e-01 7.36901343e-01 -8.63244355e-01 8.59893486e-02 -5.30160785e-01 -5.96955955e-01 -4.11607534e-01 -1.01952791e+00 1.33278346e+00 6.84169829e-02 4.61083055e-02 -1.20379758e+00 1.09402537e-01 1.20716594e-01 -4.52181660e-02 3.55676144e-01 6.31912172e-01 6.57070503e-02 -1.53372610e+00 1.66546166e-01 -4.50253308e-01 4.88527641e-02 -1.73938707e-01 3.61081868e-01 -1.09428728e+00 -1.53485224e-01 -1.54160544e-01 -6.18274994e-02 1.07984817e+00 8.94493878e-01 1.28658962e+00 -5.48428856e-02 -5.22941291e-01 1.43324184e+00 1.61309111e+00 3.19387950e-03 4.63015586e-01 3.45811546e-01 1.25958765e+00 4.52557236e-01 7.71259904e-01 3.06045890e-01 6.51078105e-01 4.69351113e-01 9.44252789e-01 -1.73183590e-01 -1.69350803e-01 -5.28026044e-01 2.04435468e-01 6.18384004e-01 5.12111001e-02 -8.83992836e-02 -1.09810185e+00 4.95150745e-01 -2.11758327e+00 -1.09239459e+00 -5.16217530e-01 2.01530576e+00 1.30917981e-01 1.63682789e-01 -7.21201301e-02 -2.20017042e-03 4.53250170e-01 3.88927370e-01 -9.85267818e-01 -1.31764531e-01 1.89684033e-01 -4.86204363e-02 7.78666914e-01 8.70426893e-01 -1.17859900e+00 1.42664385e+00 6.18034029e+00 3.48921806e-01 -1.26918769e+00 -1.56635091e-01 3.69198233e-01 -5.19652486e-01 8.67965817e-02 6.54974282e-02 -9.69773114e-01 4.63108420e-01 6.63372040e-01 -6.69448003e-02 2.72140205e-01 1.09919071e+00 5.21089852e-01 -1.31363317e-01 -1.34683108e+00 1.40696168e+00 -3.50210726e-01 -1.95644927e+00 -7.41821826e-02 1.44488439e-01 3.81529778e-01 8.13434124e-01 -3.17804694e-01 1.24974035e-01 5.27843237e-01 -7.05402553e-01 7.05633819e-01 4.90095288e-01 4.85482007e-01 -7.50908852e-01 3.62704039e-01 5.15374243e-01 -1.20968688e+00 -1.38856366e-01 -5.72600603e-01 -1.71904549e-01 5.55670142e-01 7.06910491e-01 -8.52790654e-01 2.92142659e-01 9.24027920e-01 1.27467632e+00 -3.10195804e-01 1.40010631e+00 -2.62096375e-01 4.50369626e-01 -4.02264565e-01 2.99838603e-01 3.98763686e-01 -3.48766565e-01 6.35957003e-01 1.20725822e+00 3.14038455e-01 -4.93391789e-02 3.13318282e-01 1.26092899e+00 -7.98431635e-02 -4.54160064e-01 -5.71916878e-01 5.40680230e-01 6.97365999e-01 1.36604071e+00 -5.14863908e-01 -4.55365777e-01 -2.90511161e-01 6.66287243e-01 2.82090932e-01 3.79755437e-01 -7.78628826e-01 -4.11117256e-01 1.18053901e+00 1.26377180e-01 5.38770199e-01 -7.03292370e-01 -5.01478672e-01 -1.11385787e+00 -8.26595426e-02 3.01688854e-02 1.28507912e-01 -1.06607151e+00 -9.42809105e-01 8.11892897e-02 8.60305429e-02 -1.42942846e+00 -2.09588677e-01 -7.10707545e-01 -6.74233496e-01 7.88107812e-01 -2.03731275e+00 -9.38425958e-01 -8.79305482e-01 6.79680169e-01 6.51371598e-01 2.17001081e-01 5.00446185e-02 4.83642727e-01 -7.32401013e-01 2.64467858e-02 -2.83701092e-01 3.21771890e-01 3.42497140e-01 -1.22897840e+00 1.01777804e+00 7.62883842e-01 5.91643974e-02 1.94898739e-01 2.44350761e-01 -6.10138714e-01 -1.59707642e+00 -1.50468254e+00 9.24503267e-01 -6.24771833e-01 5.70226133e-01 -2.58834928e-01 -8.42353523e-01 8.47182393e-01 -8.06657299e-02 5.52466512e-01 1.10730700e-01 -3.32349896e-01 -1.13757804e-01 -3.47087234e-01 -7.95946240e-01 5.38128734e-01 1.57044137e+00 -1.60543263e-01 -1.92603707e-01 2.93611169e-01 8.47068012e-01 -6.86950207e-01 -6.29463136e-01 3.20348084e-01 2.76468575e-01 -9.96137202e-01 1.33528686e+00 -5.39681792e-01 4.94403124e-01 -5.47804058e-01 5.29691726e-02 -1.05956221e+00 -5.59772015e-01 -7.46954560e-01 -6.08601570e-01 8.49768341e-01 7.84610882e-02 -3.10286999e-01 1.18257618e+00 3.91111493e-01 -3.91854763e-01 -6.09329462e-01 -8.53975117e-01 -6.21766090e-01 -4.01274860e-02 -8.02856922e-01 5.01971841e-01 7.73558497e-01 -6.02782845e-01 5.04810810e-01 -5.59686460e-02 3.15385729e-01 1.04386389e+00 2.32684240e-01 1.33998060e+00 -1.37839055e+00 7.80444890e-02 -4.92980957e-01 -4.25499350e-01 -1.52533519e+00 3.32856089e-01 -1.00234604e+00 1.28132543e-02 -1.58639228e+00 -4.44749087e-01 -5.17212451e-01 9.15162340e-02 4.63077277e-01 -6.44736663e-02 -6.18903823e-02 5.54682612e-01 4.44757044e-01 -5.31235158e-01 5.89203417e-01 1.38958144e+00 -1.92189202e-01 -3.95160705e-01 -2.85711531e-02 -1.99308708e-01 1.06645977e+00 5.68961680e-01 -3.53578150e-01 -5.02104819e-01 -1.06362462e+00 2.91066952e-02 1.92842614e-02 7.98928797e-01 -1.18532753e+00 5.10597169e-01 -1.92112565e-01 5.71618080e-01 -1.26382577e+00 4.45421696e-01 -8.55336249e-01 -2.04258114e-01 5.88315248e-01 -6.76068515e-02 2.26412565e-01 2.03393787e-01 4.37313616e-01 1.27759390e-02 2.01626331e-01 7.60003984e-01 -1.53763011e-01 -1.00309920e+00 8.53370607e-01 -7.29039125e-03 1.10256776e-01 9.71900344e-01 -5.12384832e-01 -3.07254344e-01 -1.13320723e-01 -4.27014500e-01 5.95005631e-01 2.50698566e-01 4.39981580e-01 7.63272524e-01 -1.32340562e+00 -5.75731337e-01 2.14960292e-01 -2.58614957e-01 8.75425816e-01 1.65248945e-01 8.11054051e-01 -1.16014814e+00 5.42658031e-01 -8.34916756e-02 -1.32619512e+00 -8.62972081e-01 3.81071717e-01 4.98354882e-01 6.60519972e-02 -1.01994133e+00 9.11147356e-01 2.68469453e-01 -4.70488071e-01 3.64227295e-01 -6.11806691e-01 1.62509546e-01 -1.17868640e-01 4.27915603e-01 1.02243865e+00 -3.16345096e-02 -4.04857159e-01 -3.76019239e-01 9.31728601e-01 1.52861953e-01 1.27199054e-01 1.29337776e+00 -6.90285787e-02 1.14243440e-01 3.62831026e-01 1.36546636e+00 -4.17527288e-01 -1.97468472e+00 -6.70832247e-02 2.70463247e-02 -8.58748913e-01 5.79499841e-01 -2.99170583e-01 -1.49785876e+00 1.34655106e+00 5.45192122e-01 -3.28730524e-01 7.65049577e-01 -2.17946649e-01 8.81585717e-01 5.43402195e-01 4.77508426e-01 -8.92909765e-01 -8.84003639e-02 5.85613728e-01 6.08052075e-01 -1.30579185e+00 9.13677961e-02 -6.01214349e-01 -1.86546251e-01 1.01539803e+00 9.79312360e-01 -4.02182609e-01 8.33325744e-01 3.23451400e-01 7.43660852e-02 -3.25098634e-01 -1.02394593e+00 -1.27923593e-01 -4.00492959e-02 5.25200069e-01 -7.16418698e-02 -4.71342832e-01 3.99468988e-01 -3.37327540e-01 -2.89396763e-01 4.82101738e-01 2.03166500e-01 9.06317413e-01 -5.54981411e-01 -5.08938134e-01 -1.69836864e-01 5.37251979e-02 1.60984024e-01 1.82692021e-01 -1.02982275e-01 1.05884314e+00 1.03195578e-01 6.40918732e-01 7.40144074e-01 -1.92946181e-01 3.93032998e-01 -4.30249184e-01 4.84724402e-01 -3.30054611e-01 -3.87290061e-01 -5.42936511e-02 -3.52375567e-01 -1.26094556e+00 -4.72275794e-01 -4.80161756e-01 -1.60004365e+00 -8.04897606e-01 -8.90474916e-02 -3.20243001e-01 7.77476072e-01 7.82466650e-01 7.06138432e-01 5.48729420e-01 6.34437978e-01 -1.50533688e+00 -1.26429662e-01 -4.66895670e-01 -1.98341623e-01 6.00562990e-02 6.98975027e-01 -6.04655385e-01 -3.11557770e-01 1.49834812e-01]
[8.516879081726074, -2.0730879306793213]
85aed4d0-52f5-4518-b83c-3fdc2e3364b4
unitor-combining-syntactic-and-semantic
null
null
https://aclanthology.org/S13-2060
https://aclanthology.org/S13-2060.pdf
UNITOR: Combining Syntactic and Semantic Kernels for Twitter Sentiment Analysis
null
['Roberto Basili', 'Simone Filice', 'Giuseppe Castellucci', 'Danilo Croce']
2013-06-01
null
null
null
semeval-2013-6
['twitter-sentiment-analysis']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.172413349151611, 3.6024231910705566]
89bb0ea1-4ea2-4bbd-9113-09505ba4d6c5
mars3d-a-plug-and-play-motion-aware-model-for
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Liu_MarS3D_A_Plug-and-Play_Motion-Aware_Model_for_Semantic_Segmentation_on_Multi-Scan_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Liu_MarS3D_A_Plug-and-Play_Motion-Aware_Model_for_Semantic_Segmentation_on_Multi-Scan_CVPR_2023_paper.pdf
MarS3D: A Plug-and-Play Motion-Aware Model for Semantic Segmentation on Multi-Scan 3D Point Clouds
3D semantic segmentation on multi-scan large-scale point clouds plays an important role in autonomous systems. Unlike the single-scan-based semantic segmentation task, this task requires distinguishing the motion states of points in addition to their semantic categories. However, methods designed for single-scan-based segmentation tasks perform poorly on the multi-scan task due to the lacking of an effective way to integrate temporal information. We propose MarS3D, a plug-and-play motion-aware model for semantic segmentation on multi-scan 3D point clouds. This module can be flexibly combined with single-scan models to allow them to have multi-scan perception abilities. The model encompasses two key designs: the Cross-Frame Feature Embedding module for enriching representation learning and the Motion-Aware Feature Learning module for enhancing motion awareness. Extensive experiments show that MarS3D can improve the performance of the baseline model by a large margin. The code is available at https://github.com/CVMI-Lab/MarS3D.
['Xiaojuan Qi', 'Lan Ma', 'Xiaoyang Wu', 'Jianhui Liu', 'Chirui Chang', 'Jiahui Liu']
2023-01-01
null
null
null
cvpr-2023-1
['3d-semantic-segmentation']
['computer-vision']
[-8.87439176e-02 -1.10630184e-01 -3.89719784e-01 -7.16078699e-01 -7.82381713e-01 -5.63326359e-01 6.00969076e-01 -1.63940012e-01 -3.66532505e-01 -1.25438705e-01 -1.78906798e-01 -2.70292044e-01 -9.41474107e-04 -8.01080108e-01 -6.55322552e-01 -5.42007446e-01 -6.13070428e-02 6.08225584e-01 9.06979024e-01 -3.26665908e-01 1.25946939e-01 6.08335674e-01 -1.65438581e+00 3.95621434e-02 5.98809361e-01 8.58474135e-01 6.18671119e-01 6.58336937e-01 -2.31152594e-01 8.05203766e-02 -8.74081552e-02 3.80014896e-01 4.60968018e-01 -9.82791707e-02 -7.56295204e-01 1.23506293e-01 5.55280685e-01 -2.71791250e-01 -4.15290684e-01 8.48112822e-01 2.37898201e-01 4.31690753e-01 4.05955970e-01 -1.56458592e+00 -1.51739586e-02 4.27535772e-02 -5.60809791e-01 2.65134662e-01 3.03375453e-01 3.49781096e-01 1.01274610e+00 -7.49918103e-01 5.29295146e-01 1.52272213e+00 5.18132389e-01 5.74339032e-01 -9.10172939e-01 -5.39828181e-01 5.28676450e-01 2.38025263e-01 -1.03068554e+00 -1.87273681e-01 8.07249546e-01 -5.39310634e-01 8.90093029e-01 1.47235617e-01 1.01777828e+00 9.48319852e-01 4.22677919e-02 9.96925116e-01 8.08352351e-01 2.21619308e-01 2.41053954e-01 -2.54887491e-01 2.55487829e-01 7.97067523e-01 1.16946578e-01 8.20738226e-02 -4.94767338e-01 -2.40362156e-03 1.04051936e+00 2.00671643e-01 6.76633716e-02 -8.52582872e-01 -1.45948660e+00 9.32592392e-01 6.49742544e-01 1.70017540e-01 -2.78184861e-01 4.75165009e-01 1.63628563e-01 5.49363270e-02 4.79482085e-01 2.25878149e-01 -6.20649517e-01 -1.52847841e-01 -9.35680330e-01 2.93001860e-01 3.80083978e-01 9.61972773e-01 1.01883066e+00 3.42391781e-03 1.99841127e-01 5.47730386e-01 5.25870919e-01 7.78453708e-01 3.90360892e-01 -1.24044275e+00 3.61751199e-01 7.04870164e-01 -1.76903941e-02 -5.69130659e-01 -7.02165484e-01 -1.61227688e-01 -4.74240035e-01 3.25705200e-01 2.20187262e-01 8.60655904e-02 -1.25008810e+00 1.58081448e+00 8.21224093e-01 5.10363221e-01 -1.73706457e-01 1.17739606e+00 7.14761257e-01 7.33749449e-01 5.94168492e-02 3.53483498e-01 1.48501444e+00 -1.21454096e+00 -2.68216789e-01 -6.87066913e-01 6.96997941e-01 -4.44424659e-01 1.12127304e+00 -4.11255956e-02 -8.12870622e-01 -6.54258966e-01 -1.07089829e+00 -2.82123893e-01 -4.58584458e-01 -5.37522361e-02 8.79189193e-01 2.80411243e-01 -1.05500770e+00 3.98602039e-01 -1.38490641e+00 -5.47396958e-01 4.31821465e-01 2.72966653e-01 -4.18250233e-01 -5.22583015e-02 -8.24398994e-01 8.36218417e-01 2.90589362e-01 -8.80760178e-02 -8.64726841e-01 -6.88830912e-01 -1.26653349e+00 -3.14356416e-01 4.87796307e-01 -9.29203272e-01 1.42866468e+00 -4.13734138e-01 -1.30804551e+00 7.66512275e-01 -4.25046027e-01 -2.40067214e-01 5.29940963e-01 -3.01903158e-01 1.83780417e-02 4.09896344e-01 5.03373623e-01 1.07001090e+00 7.41833627e-01 -1.22887695e+00 -8.38628292e-01 -5.47168016e-01 -1.27648097e-02 4.04622167e-01 4.21268344e-01 -5.07783055e-01 -9.58173037e-01 -2.73279816e-01 7.46963501e-01 -1.36907363e+00 -5.87350309e-01 2.55260617e-01 -2.08605081e-01 -2.55256027e-01 1.26742339e+00 -2.59610862e-01 4.63301510e-01 -2.12884951e+00 1.36066712e-02 5.80639094e-02 -6.20316863e-02 -4.16679587e-03 -2.22399831e-01 1.11694753e-01 1.33195773e-01 -2.68839207e-03 -3.61981630e-01 -4.09006089e-01 -2.89587630e-03 3.14038664e-01 -7.57036060e-02 5.92765212e-01 3.22677016e-01 1.05745053e+00 -1.02190518e+00 -3.42312545e-01 7.16671944e-01 3.28965276e-01 -3.30515474e-01 4.07232642e-02 -4.02628362e-01 6.59900606e-01 -9.35138404e-01 6.77616298e-01 6.51629031e-01 -4.31663960e-01 -2.54625589e-01 -1.44586578e-01 -1.68715239e-01 3.05708826e-01 -1.20739162e+00 2.40453553e+00 -3.58333051e-01 5.15966952e-01 2.80497000e-02 -6.57938063e-01 7.30803251e-01 9.30695683e-02 8.80560935e-01 -6.53144479e-01 1.52595416e-02 1.23750463e-01 -2.35877112e-01 -3.60038072e-01 6.92970812e-01 7.40863010e-02 -4.05764699e-01 4.03346002e-01 -1.73691243e-01 -7.67551959e-01 -7.25216046e-02 2.52806157e-01 1.06261075e+00 3.75144362e-01 9.49326381e-02 -5.00749685e-02 2.18413875e-01 4.22554374e-01 7.28249133e-01 5.27079225e-01 -4.15586710e-01 6.38594866e-01 -6.43386366e-03 -2.39109427e-01 -8.69943500e-01 -1.11163175e+00 -5.56277782e-02 9.23216522e-01 9.30222750e-01 -2.35747114e-01 -3.79487514e-01 -7.85392702e-01 3.49956900e-01 6.23155713e-01 -4.09582675e-01 -1.54920742e-01 -4.52353835e-01 -4.94676977e-01 1.95231393e-01 7.31896698e-01 6.20545030e-01 -6.13121748e-01 -1.14305067e+00 3.54683865e-03 -3.13195437e-01 -1.19363475e+00 -4.36117768e-01 1.42589957e-01 -1.06530547e+00 -1.12928259e+00 -3.96560520e-01 -5.69245756e-01 4.21838701e-01 9.52993572e-01 8.48549664e-01 -1.23554438e-01 -1.80441871e-01 7.62520075e-01 -4.86813575e-01 -3.91185790e-01 -1.03739249e-02 1.11936346e-01 -1.25776753e-01 -3.67981613e-01 3.84569705e-01 -4.58416969e-01 -8.69862556e-01 7.09278941e-01 -8.01942527e-01 3.17261696e-01 6.27210736e-01 3.74674618e-01 8.41132760e-01 -3.09003770e-01 1.31626651e-01 -5.77436328e-01 -2.02745184e-01 -5.80499172e-01 -5.64677179e-01 -2.08109900e-01 -3.33004117e-01 1.90387145e-02 -1.93525895e-01 -3.45895529e-01 -7.25009739e-01 4.04637694e-01 -2.91317612e-01 -6.52421474e-01 -3.27285260e-01 1.09845258e-01 -2.66141534e-01 -1.46175578e-01 2.42099628e-01 6.28423318e-02 2.48621628e-02 -5.19860208e-01 5.69147170e-01 3.36114973e-01 4.79802817e-01 -3.81997377e-01 9.97318208e-01 9.49939787e-01 6.88223988e-02 -9.14898098e-01 -8.54321301e-01 -1.09969306e+00 -9.36797023e-01 -1.82640910e-01 1.21224594e+00 -1.29241192e+00 -2.80421942e-01 5.61701298e-01 -9.94877636e-01 -7.22202122e-01 -3.00225616e-01 6.29835129e-01 -9.05101299e-01 3.31474453e-01 -3.51906985e-01 -5.52790701e-01 4.62791398e-02 -1.30809605e+00 1.63677692e+00 2.76547611e-01 -6.71848282e-02 -8.09533179e-01 -2.98364572e-02 4.97686833e-01 1.28755152e-01 2.16713607e-01 6.07445478e-01 -5.53814471e-01 -8.66126478e-01 -1.00272536e-01 -1.24330051e-01 4.13028300e-02 1.00924121e-02 -2.17955664e-01 -1.01105893e+00 -2.56751209e-01 3.45609672e-02 -9.28477123e-02 1.02502251e+00 6.19960189e-01 9.66226220e-01 3.26250553e-01 -6.03810966e-01 8.45832050e-01 1.25400198e+00 1.96198404e-01 3.58249962e-01 4.77741063e-01 9.96651292e-01 4.62841630e-01 1.11522961e+00 2.71932393e-01 9.40414131e-01 8.43592167e-01 7.10221887e-01 -1.45039052e-01 -9.40521061e-02 -2.49817610e-01 2.14849487e-01 5.83495855e-01 3.50732654e-01 -1.18113510e-01 -1.11878121e+00 6.51750863e-01 -2.12970448e+00 -7.72124231e-01 -3.84028912e-01 1.99594927e+00 1.53855205e-01 1.80420857e-02 1.88858181e-01 -1.64321318e-01 5.31808019e-01 5.85926652e-01 -9.63819206e-01 3.69945206e-02 3.08247432e-02 -2.38541126e-01 6.97659969e-01 4.55975443e-01 -1.47515380e+00 1.26478624e+00 5.43231821e+00 6.42420650e-01 -1.17557037e+00 3.79423618e-01 1.48459256e-01 -1.47313699e-01 -3.22444618e-01 2.10590467e-01 -8.73767018e-01 3.60416919e-01 7.65516937e-01 9.24370512e-02 -1.67709529e-01 9.27384853e-01 3.20173383e-01 -3.25070113e-01 -1.08055604e+00 1.02641845e+00 -1.92757592e-01 -1.16948164e+00 -1.13013275e-01 1.56478062e-01 5.67394555e-01 6.28641844e-01 -7.97531679e-02 2.28978872e-01 3.17266822e-01 -5.70084631e-01 9.33591545e-01 2.64130086e-01 5.09009659e-01 -3.91845673e-01 3.44085634e-01 4.45614696e-01 -1.55972838e+00 2.97116283e-02 -2.02811092e-01 1.67370111e-01 5.36069334e-01 4.14748013e-01 -6.06432974e-01 5.71924925e-01 8.19476843e-01 1.06007659e+00 -4.93301600e-01 1.12803364e+00 -1.85074657e-01 2.32698664e-01 -5.56302726e-01 2.19203591e-01 5.91966569e-01 -3.35105002e-01 8.81507933e-01 8.23327899e-01 5.02435803e-01 3.63835841e-02 4.92300153e-01 7.05718279e-01 3.85325342e-01 -3.83311331e-01 -7.31982350e-01 1.42965391e-01 5.42403936e-01 1.20745099e+00 -8.78633380e-01 -3.24019015e-01 -5.33246160e-01 1.04653108e+00 -1.44230321e-01 3.44853431e-01 -8.77968848e-01 -1.18800938e-01 1.14636874e+00 1.51693448e-01 6.07505202e-01 -9.07063663e-01 -2.28410587e-01 -1.12530756e+00 -5.78710325e-02 -2.94167578e-01 3.06628406e-01 -9.38523769e-01 -9.04277027e-01 3.01246494e-01 2.24434704e-01 -1.34407258e+00 -1.48889571e-01 -4.28609550e-01 -5.51131248e-01 5.19441366e-01 -1.73858500e+00 -1.40485096e+00 -6.47455394e-01 6.78351402e-01 9.52119350e-01 3.09421211e-01 3.26270998e-01 1.40936673e-02 -4.44692135e-01 -1.33996364e-02 -1.32771552e-01 7.29215052e-03 4.18280542e-01 -1.18079233e+00 8.21834326e-01 9.12485182e-01 1.59873039e-01 1.22597598e-01 5.40485382e-01 -7.55519032e-01 -1.60830069e+00 -1.39922905e+00 3.37917626e-01 -7.13303208e-01 5.39144337e-01 -3.61054957e-01 -1.01352775e+00 7.95668662e-01 -3.92868727e-01 3.35369073e-02 4.60448951e-01 -2.77107097e-02 -3.57288092e-01 1.69241875e-01 -1.03232253e+00 5.05900860e-01 1.40366983e+00 -4.87232476e-01 -6.33077919e-01 2.03287154e-01 1.00445831e+00 -6.25488579e-01 -8.61605346e-01 5.00077903e-01 2.50067323e-01 -7.37695396e-01 1.24670386e+00 -2.36789927e-01 1.63882300e-02 -6.23749673e-01 -2.73092687e-01 -1.17126310e+00 -3.23943973e-01 -3.35250556e-01 2.42922548e-02 6.30258203e-01 1.74959704e-01 -7.05467641e-01 9.62794065e-01 5.00068069e-01 -5.50132871e-01 -3.85746062e-01 -1.13510430e+00 -9.84292507e-01 -6.70644864e-02 -7.28977740e-01 6.98941886e-01 6.66731060e-01 -4.67334777e-01 2.24913210e-01 1.43316671e-01 6.74525142e-01 6.19673073e-01 4.74224299e-01 9.76649284e-01 -1.38818097e+00 1.00147678e-02 -5.00379145e-01 -7.31371343e-01 -1.47522521e+00 1.94142446e-01 -8.10911655e-01 3.54157716e-01 -1.87522280e+00 -9.17518437e-02 -6.73048913e-01 -1.02080099e-01 6.12019181e-01 -2.55176183e-02 1.01322189e-01 4.32859898e-01 4.13893491e-01 -8.52469385e-01 7.56078541e-01 1.32699943e+00 -2.71834671e-01 -3.46817970e-01 8.97188038e-02 -4.23050076e-01 7.14300752e-01 8.30254555e-01 -4.70733494e-01 -7.24758625e-01 -5.72042167e-01 -7.64241368e-02 1.07004508e-01 7.53695607e-01 -1.22158027e+00 2.72552282e-01 -3.44942063e-01 3.86020243e-02 -9.58216369e-01 9.69648302e-01 -8.18424344e-01 1.85059369e-01 3.53804350e-01 2.67719924e-02 -3.97277623e-02 3.94080162e-01 9.08776402e-01 -1.38779953e-01 8.57411548e-02 7.84318745e-01 -2.65327364e-01 -1.45192087e+00 8.22133839e-01 -2.76173770e-01 -6.79333881e-02 1.21111250e+00 -5.48494756e-01 -3.78727347e-01 -2.05125645e-01 -6.10293269e-01 6.38594389e-01 7.96571255e-01 8.30809176e-01 6.73783064e-01 -1.15492892e+00 -3.44880313e-01 3.51681113e-01 3.30015033e-01 7.55249619e-01 4.58095729e-01 1.02579820e+00 -3.79391521e-01 3.65479171e-01 -1.86534435e-01 -1.23233151e+00 -1.20169735e+00 6.88032210e-02 2.79557168e-01 1.79262578e-01 -9.42835331e-01 9.10571516e-01 4.69653249e-01 -6.94526911e-01 -1.61544561e-01 -6.04445875e-01 1.38204888e-01 -1.08533069e-01 1.76562771e-01 2.97316551e-01 -2.92727239e-02 -8.35482657e-01 -5.97223043e-01 1.02165377e+00 1.95739299e-01 -2.97988594e-01 1.35164213e+00 -4.38278407e-01 2.99317181e-01 7.76821315e-01 1.18893719e+00 -4.56494719e-01 -1.66666639e+00 -1.61509469e-01 -8.04852098e-02 -5.84133923e-01 3.14275444e-01 -3.86644781e-01 -1.04843760e+00 1.00361526e+00 6.14293635e-01 -6.03356771e-02 8.02648723e-01 3.93958032e-01 9.96739626e-01 3.69465351e-01 7.94507265e-01 -8.37088585e-01 1.47437394e-01 7.22530842e-01 5.21529555e-01 -1.50797892e+00 -1.80151761e-01 -5.77303112e-01 -8.65345538e-01 8.09324503e-01 6.67231977e-01 -6.49921298e-02 7.47277677e-01 -6.07951209e-02 2.65211225e-01 -4.77190554e-01 -7.09523082e-01 -4.85078514e-01 3.82517070e-01 6.80546343e-01 -1.69785276e-01 6.20902963e-02 3.24912757e-01 3.16617250e-01 -9.34937224e-02 -3.64472091e-01 3.55275512e-01 1.11556566e+00 -7.01413095e-01 -9.38334167e-01 -1.85875773e-01 3.46776724e-01 3.91730547e-01 3.93426448e-01 -6.96151182e-02 9.43158686e-01 2.91334242e-02 9.61632550e-01 4.25458759e-01 -5.91427386e-01 3.71539593e-01 4.27877977e-02 2.53351599e-01 -7.69394815e-01 -1.29362509e-01 5.45094982e-02 -8.31220075e-02 -1.26548874e+00 -5.03173590e-01 -9.94934201e-01 -1.64302182e+00 -6.05694726e-02 -2.93534212e-02 -2.49576885e-02 1.09698415e+00 8.48225176e-01 6.26474500e-01 3.86964649e-01 5.88136494e-01 -1.34149325e+00 -2.47016892e-01 -5.35981953e-01 -6.74370110e-01 3.78675342e-01 4.41387504e-01 -9.68798220e-01 -3.31160635e-01 -1.82490960e-01]
[8.037915229797363, -2.476175546646118]
b7987367-bd49-4fad-a498-fe86de2fcd46
check-it-again-progressive-visual-question-1
null
null
https://aclanthology.org/2021.acl-long.317
https://aclanthology.org/2021.acl-long.317.pdf
Check It Again:Progressive Visual Question Answering via Visual Entailment
While sophisticated neural-based models have achieved remarkable success in Visual Question Answering (VQA), these models tend to answer questions only according to superficial correlations between question and answer. Several recent approaches have been developed to address this language priors problem. However, most of them predict the correct answer according to one best output without checking the authenticity of answers. Besides, they only explore the interaction between image and question, ignoring the semantics of candidate answers. In this paper, we propose a select-and-rerank (SAR) progressive framework based on Visual Entailment. Specifically, we first select the candidate answers relevant to the question or the image, then we rerank the candidate answers by a visual entailment task, which verifies whether the image semantically entails the synthetic statement of the question and each candidate answer. Experimental results show the effectiveness of our proposed framework, which establishes a new state-of-the-art accuracy on VQA-CP v2 with a 7.55{\%} improvement.
['Weiping Wang', 'Peng Fu', 'Ming yu Zheng', 'Zheng Lin', 'Qingyi Si']
2021-08-01
null
null
null
acl-2021-5
['visual-entailment']
['reasoning']
[ 1.77128151e-01 7.85788894e-03 -1.48763759e-02 -4.12767619e-01 -1.08856416e+00 -6.47402287e-01 6.52973771e-01 3.36554796e-01 -4.54853445e-01 3.23707193e-01 3.96711677e-01 -4.63239312e-01 2.63749007e-02 -7.38509178e-01 -7.06739962e-01 -2.27821857e-01 5.94277143e-01 4.80220556e-01 7.20898509e-01 -1.57722712e-01 4.55667347e-01 7.74426311e-02 -1.51084864e+00 9.15239513e-01 8.90049517e-01 1.15013301e+00 3.79509300e-01 6.78523183e-01 -5.17547309e-01 1.37152064e+00 -7.24559307e-01 -8.71435761e-01 -8.42743218e-02 -7.00703859e-01 -1.13812935e+00 -1.76884625e-02 9.19032753e-01 -4.22842205e-01 -4.85492021e-01 1.03540874e+00 4.00579154e-01 -5.39650247e-02 6.19857490e-01 -1.15095317e+00 -1.04006135e+00 2.99861640e-01 -5.17449915e-01 3.91259193e-01 7.13684261e-01 2.18925238e-01 1.54098821e+00 -1.37705386e+00 6.23868287e-01 1.28161263e+00 1.95194826e-01 5.46571314e-01 -8.16455841e-01 -3.92059326e-01 1.60643861e-01 9.86280620e-01 -1.33895814e+00 -2.56768912e-01 9.25707519e-01 -1.82565242e-01 7.41967738e-01 4.32943612e-01 4.44737375e-01 6.25030339e-01 -1.64350390e-01 1.31104326e+00 1.10456347e+00 -3.92217338e-01 8.07328969e-02 2.51807213e-01 2.26149380e-01 7.88624287e-01 -3.23401809e-01 -3.75084549e-01 -5.32680809e-01 7.26397485e-02 2.15660363e-01 -2.96196014e-01 -5.83803415e-01 -1.96021020e-01 -1.02340662e+00 8.84140551e-01 7.05390036e-01 2.29549929e-01 -4.46683586e-01 3.09087951e-02 2.35295147e-01 1.59406841e-01 5.80396736e-03 2.60341913e-01 -2.32350424e-01 4.68171597e-01 -1.08238506e+00 3.70362639e-01 6.59540951e-01 7.67368734e-01 7.48191178e-01 -2.65693337e-01 -8.09384644e-01 5.60601652e-01 4.80891705e-01 3.43371421e-01 2.58166432e-01 -9.90929961e-01 8.27395797e-01 7.80670762e-01 1.14270627e-01 -1.46703136e+00 -1.08451292e-01 -3.80727381e-01 -6.19674087e-01 -1.53832912e-01 5.68100929e-01 3.76415730e-01 -7.67295063e-01 1.51793504e+00 4.67497081e-01 -5.94016984e-02 1.03765965e-01 1.15224969e+00 1.59889829e+00 9.26619649e-01 2.55620599e-01 -1.04293145e-01 1.64179850e+00 -1.16365409e+00 -7.04824746e-01 -2.88057715e-01 3.30562711e-01 -1.01031446e+00 1.42491472e+00 1.32972121e-01 -1.17628980e+00 -9.10969675e-01 -8.54195297e-01 -4.19085145e-01 -4.36114594e-02 4.49471116e-01 -4.08072807e-02 5.17941177e-01 -1.04489803e+00 -4.39655147e-02 1.53971398e-02 -8.73334259e-02 4.06238198e-01 -1.62172318e-02 -1.21326804e-01 -3.90576154e-01 -1.31755900e+00 9.88089859e-01 3.30613822e-01 1.42177969e-01 -9.98191357e-01 -5.04667997e-01 -5.45965552e-01 2.70523489e-01 5.57627738e-01 -7.91660309e-01 1.26899123e+00 -1.13231027e+00 -1.09710741e+00 1.10090780e+00 -4.36409146e-01 -4.81102109e-01 5.51519215e-01 -1.57148659e-01 -4.71282005e-01 8.20200145e-01 1.26033902e-01 9.04334307e-01 8.03811908e-01 -1.53121209e+00 -8.80261362e-01 -2.72558630e-01 5.95651031e-01 2.51820713e-01 -1.97781816e-01 -1.71883352e-04 -9.20719624e-01 -3.22047412e-01 2.22338706e-01 -5.00805080e-01 1.65465221e-01 2.21918985e-01 -3.24312121e-01 -6.39836490e-01 7.99769938e-01 -1.05218995e+00 1.23988986e+00 -1.95091212e+00 -4.31534499e-02 9.18781832e-02 4.49597210e-01 3.42357606e-01 -3.41248453e-01 3.46636623e-01 2.17527881e-01 1.64810568e-02 -1.55765608e-01 -1.00529172e-01 1.30846173e-01 1.19170077e-01 -6.42694175e-01 3.10116887e-01 2.86301404e-01 1.14643478e+00 -6.16427839e-01 -1.09306288e+00 -9.88740399e-02 1.97345749e-01 -4.68399346e-01 5.21583319e-01 -5.94670415e-01 2.65852034e-01 -3.95993173e-01 4.74931687e-01 8.57175410e-01 -5.97649634e-01 6.43667281e-02 -5.32810509e-01 2.28817895e-01 8.84968042e-02 -9.66234803e-01 1.26651347e+00 -2.19913766e-01 6.47305667e-01 -1.13092735e-01 -8.43441844e-01 9.93872881e-01 1.80278406e-01 -1.70457289e-02 -1.23839653e+00 3.88713367e-02 1.30897745e-01 -2.37205625e-01 -1.00022197e+00 5.23507714e-01 -1.50904790e-01 2.20421106e-01 3.19590032e-01 6.76971003e-02 -2.29758158e-01 2.77893484e-01 5.23619175e-01 7.04280138e-01 2.36711890e-01 3.33453536e-01 1.20500252e-01 1.18524063e+00 1.22886881e-01 8.65650475e-02 8.89067829e-01 -4.62799191e-01 8.45807552e-01 6.15158081e-01 -2.77139693e-01 -8.00344825e-01 -1.20537233e+00 3.88672262e-01 1.08767736e+00 6.10347629e-01 -2.70038068e-01 -6.87330365e-01 -9.52931166e-01 -3.57054979e-01 9.58410382e-01 -5.71329474e-01 8.52127820e-02 -7.29980648e-01 -1.04530707e-01 5.13878226e-01 3.33795965e-01 8.02493632e-01 -1.16332722e+00 -6.41579151e-01 -1.59242645e-01 -8.69914830e-01 -1.22397387e+00 -4.31476831e-01 -5.22078574e-01 -5.84661722e-01 -1.25709391e+00 -8.38860750e-01 -1.19426608e+00 5.65456152e-01 3.35396826e-01 1.44913864e+00 5.26537001e-01 1.76730379e-01 6.66109502e-01 -4.24698740e-01 2.03502197e-02 -2.48290792e-01 -1.79579407e-01 -6.40212774e-01 3.93855125e-01 4.13831800e-01 2.93254219e-02 -9.37993944e-01 2.71698594e-01 -9.72537398e-01 -4.65938449e-02 7.07077563e-01 5.25554121e-01 8.32400620e-01 -3.02998781e-01 6.36418819e-01 -5.54643750e-01 7.94476092e-01 -3.95099819e-01 -3.10612857e-01 8.39819670e-01 -3.87701482e-01 1.16629690e-01 6.34195268e-01 -1.85588613e-01 -1.30212462e+00 -6.79547414e-02 -4.56449777e-01 -4.07066822e-01 -1.23554587e-01 5.26215792e-01 -2.26666048e-01 1.39624238e-01 5.37790060e-01 5.49163759e-01 -4.76667464e-01 -1.71234652e-01 7.53888488e-01 4.55691934e-01 8.28970671e-01 -5.13228893e-01 7.03122079e-01 5.61496973e-01 -1.65495694e-01 -6.30397201e-01 -1.15283418e+00 -6.33689225e-01 -3.20322961e-01 -6.04043722e-01 1.17071474e+00 -6.84342325e-01 -9.51836526e-01 -2.50419136e-02 -1.66975558e+00 3.77335280e-01 -5.04294932e-02 2.27948889e-01 -3.36777866e-01 8.63712728e-01 -3.20764244e-01 -8.52569878e-01 -4.46683317e-01 -1.00277901e+00 7.45162129e-01 4.16038811e-01 -1.96841881e-01 -5.12271881e-01 -1.02454089e-01 9.26131606e-01 3.02034289e-01 -3.50177810e-02 1.39790082e+00 -6.25991225e-01 -8.28082204e-01 -9.13734064e-02 -8.04284632e-01 1.72907874e-01 -4.18572187e-01 -8.69990364e-02 -8.98204863e-01 6.97714463e-03 2.44547427e-01 -6.45766079e-01 1.03031778e+00 8.06461573e-02 1.16893756e+00 -3.34858567e-01 5.46362363e-02 1.72296446e-02 1.43452013e+00 -1.92142967e-02 7.92008221e-01 5.18392548e-02 6.76820874e-01 1.08347237e+00 5.71460426e-01 1.69510409e-01 7.13526309e-01 4.30679709e-01 7.62791693e-01 -1.56695861e-02 -4.21637177e-01 -4.07728344e-01 1.37632027e-01 7.39185035e-01 2.83497661e-01 -4.74541336e-01 -9.21345770e-01 8.59711289e-01 -1.69421864e+00 -1.02357161e+00 -7.06118882e-01 1.85769176e+00 7.69195855e-01 1.27084389e-01 -7.20828474e-02 8.39415416e-02 7.21368074e-01 4.08530563e-01 -3.55854630e-01 -1.50295958e-01 -2.98506439e-01 3.38461071e-01 -1.69966906e-01 5.21525919e-01 -8.25814366e-01 9.11353350e-01 5.65612078e+00 1.07516289e+00 -8.34393740e-01 1.14103138e-01 6.86412513e-01 2.57569849e-01 -6.88970506e-01 7.02348351e-02 -5.84186018e-01 2.85814498e-02 2.97781736e-01 5.77572323e-02 7.48331100e-02 6.05017841e-01 -6.70095682e-02 -2.82971025e-01 -9.43268180e-01 9.49050248e-01 7.33795702e-01 -1.36007977e+00 5.73338509e-01 -7.12801099e-01 6.17004335e-01 -5.39407313e-01 2.81042606e-01 3.71741295e-01 -2.50564158e-01 -1.02455389e+00 9.99096334e-01 6.30199552e-01 5.31596065e-01 -8.05193007e-01 7.87457824e-01 3.48110378e-01 -1.33198249e+00 -9.11152642e-03 -2.59725392e-01 -2.88067316e-03 1.57697260e-01 1.91460133e-01 -7.28928149e-01 4.76957232e-01 8.39870453e-01 1.51527405e-01 -1.11414015e+00 9.99656498e-01 -7.57181883e-01 6.49393916e-01 1.49754658e-01 -4.73567337e-01 3.49888980e-01 1.84301764e-01 3.02488774e-01 9.56971169e-01 1.49402022e-01 2.43860781e-01 3.16220820e-02 9.57238317e-01 -2.46888772e-01 4.52171355e-01 -9.56049636e-02 1.61619484e-01 2.36260772e-01 1.07626617e+00 -5.93259811e-01 -4.56087798e-01 -6.23141527e-01 1.02784371e+00 4.24428165e-01 4.87851441e-01 -9.18638766e-01 -4.45685923e-01 -1.19983822e-01 4.08934504e-02 5.92949867e-01 -2.36502253e-02 -1.48048565e-01 -9.65216637e-01 3.92893255e-01 -1.05800378e+00 7.89865196e-01 -1.26425159e+00 -1.26258886e+00 6.78865492e-01 -2.10477665e-01 -1.18062699e+00 2.69178860e-02 -6.04567349e-01 -4.99987304e-01 6.93105459e-01 -1.80733323e+00 -1.25867450e+00 -4.22047585e-01 7.76638567e-01 5.72668314e-01 1.21216513e-01 3.96743864e-01 2.34183356e-01 -3.12727168e-02 4.19891655e-01 -3.31813753e-01 1.38964668e-01 5.91879606e-01 -1.02391791e+00 3.13496925e-02 1.03713560e+00 5.52851737e-01 3.56270820e-01 8.00520957e-01 -4.65600520e-01 -1.11477506e+00 -7.69080222e-01 1.37260413e+00 -5.05440295e-01 5.01683295e-01 5.24146780e-02 -1.15786350e+00 2.87452728e-01 7.55085766e-01 -2.10116029e-01 4.21669394e-01 -3.76460701e-01 -6.36334062e-01 -1.70180902e-01 -9.46039677e-01 7.50215590e-01 6.47955954e-01 -8.71629775e-01 -1.03520215e+00 1.78571433e-01 8.43368828e-01 -2.05566332e-01 -3.73177350e-01 3.83619189e-01 4.29737300e-01 -1.20071077e+00 1.12648368e+00 -5.76103985e-01 8.11389267e-01 -7.88114190e-01 -3.00432801e-01 -6.31528378e-01 -1.53914228e-01 7.67019987e-02 -1.66070968e-01 1.25492287e+00 5.27957320e-01 -2.86943894e-02 7.89050043e-01 2.93244153e-01 7.09915832e-02 -6.40882373e-01 -1.05654204e+00 -1.76960334e-01 -1.70789972e-01 -4.44866210e-01 4.63538259e-01 7.20210791e-01 -3.25794756e-01 6.40688539e-01 -4.89053637e-01 2.00887874e-01 2.90031552e-01 5.78276992e-01 6.60892308e-01 -7.23690391e-01 -4.89379168e-01 -5.12254536e-01 -7.72122145e-02 -1.49472392e+00 7.10542202e-02 -7.90530503e-01 2.70876229e-01 -2.02645302e+00 3.74110252e-01 2.85440952e-01 -1.95308849e-01 2.22792178e-02 -6.34771466e-01 2.69882590e-01 3.81262124e-01 2.31376126e-01 -1.17131531e+00 5.97903371e-01 1.43514073e+00 -5.21603584e-01 1.31560102e-01 -2.40818009e-01 -5.45717359e-01 7.11730480e-01 6.44345820e-01 -3.04175466e-01 -4.53160018e-01 -7.34972000e-01 5.01670420e-01 2.66373545e-01 6.91134691e-01 -9.30569232e-01 4.19085115e-01 -1.03047580e-01 5.16758144e-01 -1.01470220e+00 2.16039166e-01 -7.45923102e-01 -4.49252188e-01 4.49756354e-01 -6.84399188e-01 3.59324098e-01 -1.42828301e-01 7.77034223e-01 -5.57790399e-01 -4.76266712e-01 6.85131788e-01 -2.10340872e-01 -1.09371316e+00 1.60043344e-01 -2.08125293e-01 5.11375487e-01 8.77690673e-01 9.79442429e-03 -5.76656342e-01 -7.44978189e-01 -4.61491585e-01 3.92082393e-01 -4.31664288e-02 2.55991042e-01 1.29584908e+00 -1.30816424e+00 -1.01855159e+00 -3.95886123e-01 5.60044587e-01 -3.74751955e-01 5.70966005e-01 6.47225976e-01 -6.60322368e-01 4.71929997e-01 1.51908267e-02 -6.05373085e-01 -1.44582951e+00 7.65653014e-01 2.85193056e-01 -3.18458378e-01 -2.77353704e-01 1.09822643e+00 3.55888486e-01 -2.79226601e-01 3.34166527e-01 -1.44550517e-01 -8.57722402e-01 2.62268573e-01 5.94463170e-01 1.18966430e-01 -1.73108995e-01 -9.35971498e-01 -4.78850633e-01 6.95394218e-01 1.00628480e-01 -2.76019722e-01 6.85607553e-01 -2.32174501e-01 -1.95270896e-01 5.19982576e-02 1.15417814e+00 3.31240036e-02 -9.67342913e-01 -5.36784887e-01 5.66365756e-02 -5.25728226e-01 -3.11586022e-01 -7.47047186e-01 -1.03099430e+00 1.32052898e+00 4.85000432e-01 1.74030662e-01 1.21580493e+00 3.31939787e-01 8.74136329e-01 3.87585878e-01 -1.55443072e-01 -9.72050309e-01 7.31111407e-01 4.26309288e-01 1.16688156e+00 -1.23617339e+00 -3.40080485e-02 -2.27267191e-01 -9.06260669e-01 1.04810834e+00 8.94160986e-01 -1.06550904e-03 1.51751056e-01 -6.52308047e-01 1.72303498e-01 -3.18794876e-01 -5.96335590e-01 -4.41765845e-01 7.70219088e-01 4.78970319e-01 2.04434752e-01 -2.12535217e-01 -5.05875230e-01 4.92993236e-01 2.37353612e-02 -3.03287536e-01 2.05933675e-01 5.15583158e-01 -7.03085899e-01 -6.01477742e-01 -4.59836304e-01 2.25060225e-01 -3.17393184e-01 -2.85151094e-01 -6.60470605e-01 5.94541550e-01 1.44493714e-01 1.42843330e+00 -1.46557912e-01 -3.20995420e-01 4.07791436e-01 1.19764090e-01 4.85061318e-01 -1.16212964e-01 -6.37070060e-01 -2.33211577e-01 2.79234145e-02 -3.52925271e-01 -4.80577201e-01 -1.91073030e-01 -1.29708278e+00 -1.28046602e-01 -1.81941912e-01 2.27721006e-01 3.07502180e-01 1.17730260e+00 1.20695896e-01 3.45730633e-01 4.00116503e-01 -4.06897813e-03 -4.60215688e-01 -6.06059253e-01 -5.94770722e-02 6.30498886e-01 2.61021763e-01 -1.13438159e-01 -2.80790329e-01 1.02269664e-01]
[10.843645095825195, 1.621757984161377]
071a5880-82a5-487e-92e5-680869a01519
contrastive-object-detection-using-knowledge
2112.11366
null
https://arxiv.org/abs/2112.11366v1
https://arxiv.org/pdf/2112.11366v1.pdf
Contrastive Object Detection Using Knowledge Graph Embeddings
Object recognition for the most part has been approached as a one-hot problem that treats classes to be discrete and unrelated. Each image region has to be assigned to one member of a set of objects, including a background class, disregarding any similarities in the object types. In this work, we compare the error statistics of the class embeddings learned from a one-hot approach with semantically structured embeddings from natural language processing or knowledge graphs that are widely applied in open world object detection. Extensive experimental results on multiple knowledge-embeddings as well as distance metrics indicate that knowledge-based class representations result in more semantically grounded misclassifications while performing on par compared to one-hot methods on the challenging COCO and Cityscapes object detection benchmarks. We generalize our findings to multiple object detection architectures by proposing a knowledge-embedded design for keypoint-based and transformer-based object detection architectures.
['Abhinav Valada', 'Alexander Braun', 'Christopher Lang']
2021-12-21
null
null
null
null
['open-world-object-detection', 'knowledge-graph-embeddings', 'knowledge-graph-embeddings']
['computer-vision', 'graphs', 'methodology']
[ 1.87975653e-02 2.03292251e-01 -2.69634277e-01 -4.68140632e-01 -6.39224648e-01 -7.37553835e-01 7.68426061e-01 6.81017697e-01 -4.60494995e-01 1.29307866e-01 1.92931555e-02 4.35543582e-02 -4.33620214e-01 -1.14636803e+00 -5.85747898e-01 -6.38746202e-01 -2.74186283e-01 7.74865627e-01 6.77465558e-01 -4.74636704e-02 4.93198559e-02 6.39345288e-01 -1.91067863e+00 4.80565816e-01 2.72582412e-01 1.23762739e+00 -1.06683329e-01 5.33990204e-01 -3.36193532e-01 7.12650478e-01 -4.96699005e-01 -6.28139973e-01 3.48408490e-01 2.23927170e-01 -1.02882040e+00 3.60196158e-02 1.03592718e+00 -2.15740525e-04 -3.89402539e-01 1.15907097e+00 2.21388608e-01 2.35405743e-01 9.75960255e-01 -1.43879521e+00 -1.05311954e+00 3.76635581e-01 -4.19678867e-01 4.87416357e-01 1.19882971e-01 -9.62613225e-02 1.28754985e+00 -1.14454401e+00 6.21531785e-01 1.48963332e+00 8.21101069e-01 1.19298935e-01 -1.29801643e+00 -3.03706229e-01 3.36710215e-01 5.66690683e-01 -1.78621078e+00 -1.84484467e-01 2.95092434e-01 -5.37415564e-01 1.12830448e+00 2.72072911e-01 4.45534438e-01 7.68903077e-01 5.95308952e-02 6.48026347e-01 5.76526344e-01 -4.48222786e-01 3.15775931e-01 4.85440493e-01 5.07573724e-01 7.17736959e-01 7.82979131e-01 -4.13311198e-02 -1.24118268e-01 -1.51494861e-01 3.25283319e-01 2.95736760e-01 -4.30994332e-02 -9.16113913e-01 -1.22136712e+00 9.77185011e-01 1.02150249e+00 3.24382812e-01 -1.29976347e-01 2.87631959e-01 4.84761119e-01 1.57122761e-01 4.25415248e-01 5.74546099e-01 -3.23734075e-01 3.81615758e-01 -6.77255332e-01 2.50983149e-01 8.57174993e-01 1.25982094e+00 1.12164891e+00 -1.54304922e-01 -2.95175403e-01 7.32013226e-01 2.75956720e-01 2.73949325e-01 4.09999400e-01 -4.80602115e-01 3.36434305e-01 1.02127850e+00 1.45124067e-02 -1.26157928e+00 -3.20898056e-01 -5.18582642e-01 -2.85437405e-01 1.67624459e-01 3.79606307e-01 4.83859956e-01 -1.11437416e+00 1.35140693e+00 4.92104232e-01 1.05033457e-01 7.88823441e-02 9.46773827e-01 9.99827743e-01 4.86106843e-01 1.76695243e-01 4.15196210e-01 1.69673169e+00 -9.83322859e-01 -3.57320398e-01 -1.19243428e-01 7.37752199e-01 -6.01411641e-01 8.08948755e-01 2.37660423e-01 -5.01350820e-01 -5.42742729e-01 -9.41184938e-01 -3.06308240e-01 -1.20522034e+00 -2.32836679e-02 5.37475646e-01 9.88636792e-01 -8.75967741e-01 2.94751257e-01 -6.42294645e-01 -7.06151307e-01 7.83587933e-01 1.49454415e-01 -3.93284261e-01 -2.81250298e-01 -7.97252595e-01 1.07122386e+00 7.88876653e-01 -4.09318924e-01 -1.05817282e+00 -8.43182325e-01 -8.96145046e-01 3.48105431e-01 4.92252141e-01 -5.18615484e-01 1.04169583e+00 -1.02786648e+00 -5.97383857e-01 1.16472530e+00 1.12112768e-01 -5.79565465e-01 1.47259831e-01 -1.70372248e-01 -5.85719585e-01 2.76588827e-01 3.26154470e-01 8.47687304e-01 7.17215300e-01 -1.36273134e+00 -8.71397555e-01 -4.52538282e-01 2.25151494e-01 8.63915756e-02 -5.83893359e-01 1.80385970e-02 -3.36123288e-01 -7.28423476e-01 1.62597522e-01 -7.36593306e-01 -7.14627802e-02 4.65212882e-01 -3.31354827e-01 -4.55488443e-01 1.15087163e+00 -2.54134059e-01 9.60794151e-01 -2.24321723e+00 1.51011590e-02 1.62848711e-01 3.94686818e-01 1.53755277e-01 -1.98668271e-01 1.50660768e-01 -3.49371016e-01 1.86594710e-01 -9.19103473e-02 -2.07149908e-01 1.60229877e-01 4.63888735e-01 -5.22006094e-01 5.85065782e-01 5.15823185e-01 1.01359546e+00 -1.02069640e+00 -5.62104762e-01 2.23831370e-01 3.78245026e-01 -4.36315686e-01 -9.54615995e-02 -1.97731465e-01 -5.76293945e-01 -3.84483606e-01 9.43599224e-01 5.03963172e-01 -2.87924051e-01 1.27683282e-02 -2.96602160e-01 1.39017865e-01 7.67429695e-02 -1.28709590e+00 1.41600835e+00 -2.19992861e-01 6.96251512e-01 -4.63414460e-01 -1.00593793e+00 7.27418721e-01 1.67680413e-01 3.13604653e-01 -5.26958287e-01 3.54356542e-02 3.76065336e-02 -3.25420201e-02 -3.17120701e-01 7.37221837e-01 -5.40432483e-02 4.07921225e-02 2.34210923e-01 4.42482978e-01 5.76463044e-02 5.41160107e-02 3.53193134e-01 1.07775867e+00 -1.29231304e-01 3.71596247e-01 -5.32424867e-01 1.46894187e-01 4.35477644e-01 4.30762291e-01 9.66699660e-01 -2.79143095e-01 8.65103185e-01 2.57198602e-01 -7.03885376e-01 -9.53713119e-01 -1.61566126e+00 -4.50838685e-01 1.25584924e+00 4.40790206e-01 -5.11529148e-01 -5.08293629e-01 -8.86346161e-01 3.92842621e-01 7.54508674e-01 -9.09103155e-01 -2.41924778e-01 -2.07727298e-01 -6.72679126e-01 6.61642790e-01 7.63909876e-01 -2.83245314e-02 -7.02693462e-01 -5.96466780e-01 1.13086537e-01 3.04007977e-01 -1.03602147e+00 -2.73734540e-01 4.78011906e-01 -6.59911633e-01 -1.37406993e+00 -6.34564757e-01 -1.01742363e+00 6.35618031e-01 5.10504365e-01 1.31657970e+00 -8.97543877e-02 -9.12572205e-01 8.24045539e-01 -5.94922185e-01 -6.68201745e-01 -1.23624705e-01 -6.79185912e-02 1.90921262e-01 2.62626380e-01 7.52278328e-01 -9.68468636e-02 -4.34324026e-01 4.43277210e-01 -9.49124277e-01 -5.78994691e-01 4.57400739e-01 5.13239503e-01 6.90804839e-01 1.63680017e-01 3.29014242e-01 -6.31660879e-01 9.92634967e-02 -7.64563203e-01 -5.37295341e-01 5.29189408e-01 -4.92351562e-01 3.07769165e-03 2.67224401e-01 -6.80319905e-01 -7.65195191e-01 1.35394946e-01 5.60225964e-01 -6.74913228e-01 -4.21151191e-01 3.70986536e-02 -1.55822322e-01 -1.05837353e-01 8.75422060e-01 1.61410734e-01 -2.68200159e-01 -4.13679749e-01 7.50805497e-01 4.98548895e-01 4.43714648e-01 -5.90836108e-01 8.61205280e-01 8.13934386e-01 -2.30856657e-01 -1.01577854e+00 -8.63947630e-01 -9.40042973e-01 -7.05101550e-01 6.04464412e-02 1.17171109e+00 -9.77354765e-01 -1.88546434e-01 -5.20461006e-03 -1.09598887e+00 2.47796372e-01 -8.01671624e-01 3.31352443e-01 -3.88031274e-01 1.46967828e-01 -1.57759860e-01 -5.25569320e-01 1.11619927e-01 -9.88724709e-01 1.20794594e+00 1.54919475e-01 -2.32393090e-02 -9.26246226e-01 4.02704775e-02 1.01502016e-01 2.60510862e-01 8.18229988e-02 1.08045173e+00 -1.06526601e+00 -8.67424071e-01 -4.61843193e-01 -5.50690532e-01 2.01540679e-01 2.88076460e-01 -7.45674148e-02 -1.14433873e+00 -2.11795956e-01 -4.38722432e-01 -2.71029770e-01 1.19822133e+00 1.91706400e-02 1.10989416e+00 -2.02332228e-01 -7.16993868e-01 3.66441876e-01 1.63773322e+00 -9.89659205e-02 4.70402658e-01 3.61578643e-01 8.61814022e-01 5.85844338e-01 3.50469649e-01 2.03600958e-01 2.96974063e-01 7.32728839e-01 6.68210506e-01 -6.31619692e-02 -3.15947443e-01 -9.87370834e-02 4.24106084e-02 9.73511189e-02 3.22098374e-01 -2.75025874e-01 -1.18684769e+00 1.13752842e+00 -1.75412560e+00 -1.03603208e+00 -3.50856245e-01 2.14230251e+00 5.24869144e-01 8.81140456e-02 2.54490644e-01 8.98102075e-02 7.74500430e-01 -9.23093408e-02 -3.42945993e-01 -2.63859719e-01 -2.50109315e-01 1.88258246e-01 6.57843053e-01 -5.22808097e-02 -1.37409556e+00 8.50766003e-01 6.39333963e+00 7.73066998e-01 -7.74347186e-01 4.10096973e-01 4.06287462e-01 -1.87135294e-01 -1.24837436e-01 2.60709021e-02 -9.41459537e-01 1.13375306e-01 6.05073810e-01 -7.72816613e-02 -9.45618004e-02 1.18788910e+00 -3.92775804e-01 -9.21784341e-02 -1.61258590e+00 9.09277260e-01 1.98250622e-01 -1.48413360e+00 4.10199195e-01 -5.17215133e-02 7.64308333e-01 1.94408119e-01 6.98159635e-02 3.01555336e-01 3.31669509e-01 -9.98663187e-01 9.75982666e-01 3.97426844e-01 3.81265402e-01 -4.91718650e-01 5.25158525e-01 4.83463630e-02 -1.49791205e+00 -3.06417823e-01 -5.41920543e-01 1.22539043e-01 -3.28664452e-01 4.48649377e-01 -9.95215833e-01 6.23729885e-01 1.16574037e+00 7.46314943e-01 -9.15950596e-01 1.41805518e+00 1.38551936e-01 4.76119906e-01 -3.59830201e-01 1.08299159e-01 4.31450367e-01 2.09885642e-01 5.25022566e-01 1.45247602e+00 -1.55753419e-02 1.85080105e-03 4.57762897e-01 1.09001160e+00 -3.14471275e-02 -2.30627600e-02 -7.87724555e-01 8.31945091e-02 3.62444013e-01 1.20341003e+00 -1.05432189e+00 -6.26949787e-01 -7.22326577e-01 6.77960992e-01 3.43069345e-01 2.98740625e-01 -7.86892295e-01 -6.29506528e-01 1.03615320e+00 2.05158606e-01 7.32388079e-01 -1.12789713e-01 -9.37662497e-02 -1.01148582e+00 3.00437212e-02 -2.68290639e-01 7.42570519e-01 -6.70968890e-01 -1.55139160e+00 4.52667266e-01 1.55239478e-01 -1.20712650e+00 2.68927395e-01 -9.98493254e-01 -3.88733774e-01 5.33994019e-01 -1.70404220e+00 -1.18742371e+00 -3.08838487e-01 4.93029058e-01 4.97520238e-01 -1.32741526e-01 8.23541403e-01 4.64596033e-01 -4.28675175e-01 5.49327970e-01 2.41489172e-01 5.26952088e-01 5.27538538e-01 -1.39270341e+00 1.71436891e-01 6.93368793e-01 7.30201304e-01 6.31862998e-01 3.64724636e-01 -3.54025543e-01 -1.17703700e+00 -1.58112645e+00 7.11801648e-01 -9.51943636e-01 7.45289087e-01 -4.01891559e-01 -1.09798992e+00 8.45298290e-01 1.97393708e-02 6.38669133e-01 5.46621919e-01 1.70326829e-01 -1.05379784e+00 -9.23192874e-02 -1.18965447e+00 1.80899411e-01 1.04422605e+00 -6.23278141e-01 -9.38382864e-01 5.08283556e-01 7.69338131e-01 -7.00451434e-02 -8.04066479e-01 1.78324550e-01 3.38297337e-01 -6.36018932e-01 1.25913274e+00 -9.88023579e-01 2.38847852e-01 -5.84894300e-01 -5.72170019e-01 -1.03534925e+00 -6.53611243e-01 1.88753814e-01 -7.40016997e-02 1.11312532e+00 2.28178829e-01 -5.93686044e-01 7.27451742e-01 3.06734949e-01 -9.15030688e-02 -3.42520475e-01 -1.10005832e+00 -1.25460804e+00 1.13641642e-01 -3.54735523e-01 5.55136502e-01 1.09841192e+00 -5.73481500e-01 8.92121717e-02 3.53841484e-01 6.61265194e-01 6.30535364e-01 2.92916149e-01 5.49439430e-01 -1.45913649e+00 -3.28245200e-02 -6.04707003e-01 -1.32152748e+00 -5.74066699e-01 -3.88006382e-02 -1.30554521e+00 -1.12378500e-01 -1.53612936e+00 4.20214385e-01 -7.26156473e-01 -5.62834918e-01 6.95219755e-01 1.71520174e-01 4.67000663e-01 1.42071739e-01 1.06831096e-01 -9.27697122e-01 4.45648998e-01 5.21312475e-01 -6.75965488e-01 2.52507716e-01 -4.00910705e-01 -5.74321151e-01 6.74349487e-01 3.28316391e-01 -7.60292113e-01 -3.51876646e-01 -7.38809705e-01 1.79287672e-01 -6.54619277e-01 9.28281307e-01 -1.18067420e+00 2.36892745e-01 -5.72290756e-02 4.07212436e-01 -4.34330791e-01 3.03005725e-01 -1.18154395e+00 -9.61452425e-02 3.18753958e-01 -2.13008925e-01 -1.28501251e-01 3.75432342e-01 1.04910934e+00 -2.55083203e-01 -7.98843801e-02 8.48363996e-01 -6.31649941e-02 -1.26068759e+00 2.43966728e-01 -1.67074919e-01 1.96661100e-01 1.39169705e+00 -7.56794274e-01 -3.86441171e-01 2.18566611e-01 -7.87965000e-01 1.52075246e-01 3.22622269e-01 8.97317827e-01 6.81683838e-01 -1.36814642e+00 -5.58371782e-01 5.78383096e-02 8.96155417e-01 -6.85055479e-02 -1.16200916e-01 5.02022803e-01 -3.63881826e-01 4.70313340e-01 -4.94415872e-02 -9.48385835e-01 -1.19962573e+00 9.70277965e-01 5.07744253e-01 2.57157356e-01 -6.48812175e-01 8.85277510e-01 5.62939286e-01 -6.40927255e-01 3.94922793e-01 -5.64798534e-01 -6.25530109e-02 3.05983603e-01 4.85486567e-01 4.73397642e-01 3.60176653e-01 -7.60812044e-01 -6.91986382e-01 5.45772731e-01 -2.26468191e-01 4.99689490e-01 1.26012182e+00 1.52506948e-01 7.49837086e-02 3.98697495e-01 1.37242579e+00 -4.25530225e-01 -7.32872725e-01 -5.54723799e-01 4.14464265e-01 -5.39799273e-01 1.94757670e-01 -8.08671832e-01 -1.03839624e+00 6.55821383e-01 9.82073069e-01 2.70140707e-01 7.75800705e-01 5.16827822e-01 2.44719982e-01 6.19218886e-01 5.30717313e-01 -9.91656005e-01 1.82317406e-01 3.54400456e-01 6.55738354e-01 -1.31724691e+00 2.81755216e-02 -5.14381468e-01 -3.47082496e-01 1.12471366e+00 5.45140147e-01 -2.18306825e-01 8.41363966e-01 -6.14717193e-02 -2.22752452e-01 -5.47389328e-01 -5.02891660e-01 -5.58184624e-01 4.74856019e-01 7.63679802e-01 6.45457655e-02 2.04974875e-01 3.54079068e-01 3.82575959e-01 2.25134417e-01 -3.54509652e-01 2.61280715e-01 9.35194194e-01 -6.79548025e-01 -7.61332452e-01 -5.84032118e-01 5.75754523e-01 -8.68016481e-02 -1.39716923e-01 -4.80506599e-01 1.02399909e+00 4.32768345e-01 7.08018243e-01 5.74048698e-01 -8.42496231e-02 3.86026323e-01 8.79860967e-02 4.24382776e-01 -1.14541507e+00 -5.76310337e-01 -5.36280751e-01 -9.92228836e-02 -6.31495059e-01 -2.20727131e-01 -5.03526390e-01 -1.27354228e+00 1.35945126e-01 -6.03057325e-01 -1.92500472e-01 4.56075549e-01 7.51662433e-01 5.10033727e-01 4.34339225e-01 1.93331227e-01 -6.67689443e-01 -5.35488427e-01 -4.84224558e-01 -6.22324049e-01 6.41416490e-01 3.23390961e-01 -9.83514309e-01 -2.26358235e-01 5.44548370e-02]
[9.507607460021973, 1.5354729890823364]
8d7bf8df-e73b-459d-a22d-70ddf08158d2
a-simple-decentralized-cross-entropy-method
2212.08235
null
https://arxiv.org/abs/2212.08235v1
https://arxiv.org/pdf/2212.08235v1.pdf
A Simple Decentralized Cross-Entropy Method
Cross-Entropy Method (CEM) is commonly used for planning in model-based reinforcement learning (MBRL) where a centralized approach is typically utilized to update the sampling distribution based on only the top-$k$ operation's results on samples. In this paper, we show that such a centralized approach makes CEM vulnerable to local optima, thus impairing its sample efficiency. To tackle this issue, we propose Decentralized CEM (DecentCEM), a simple but effective improvement over classical CEM, by using an ensemble of CEM instances running independently from one another, and each performing a local improvement of its own sampling distribution. We provide both theoretical and empirical analysis to demonstrate the effectiveness of this simple decentralized approach. We empirically show that, compared to the classical centralized approach using either a single or even a mixture of Gaussian distributions, our DecentCEM finds the global optimum much more consistently thus improves the sample efficiency. Furthermore, we plug in our DecentCEM in the planning problem of MBRL, and evaluate our approach in several continuous control environments, with comparison to the state-of-art CEM based MBRL approaches (PETS and POPLIN). Results show sample efficiency improvement by simply replacing the classical CEM module with our DecentCEM module, while only sacrificing a reasonable amount of computational cost. Lastly, we conduct ablation studies for more in-depth analysis. Code is available at https://github.com/vincentzhang/decentCEM
['Dale Schuurmans', 'Jun Luo', 'Martin Jagersand', 'Jun Jin', 'Zichen Zhang']
2022-12-16
null
null
null
null
['continuous-control']
['playing-games']
[-1.89719483e-01 1.60200283e-01 -1.74068272e-01 7.39791170e-02 -1.29677749e+00 -5.95790863e-01 7.17894077e-01 2.39771456e-01 -6.90512776e-01 1.27142358e+00 -1.69527799e-01 -3.56910229e-01 -3.32967252e-01 -8.70993137e-01 -9.27379906e-01 -9.82064724e-01 -2.08718866e-01 7.70403981e-01 1.32297292e-01 -2.45161623e-01 5.00338674e-01 3.36498410e-01 -1.21561575e+00 -2.25588933e-01 9.33522999e-01 6.33380294e-01 3.03749919e-01 5.88417709e-01 4.23414737e-01 8.36749434e-01 -7.24722087e-01 -2.59119123e-01 3.96285862e-01 -4.48002428e-01 -8.48299921e-01 -1.11172728e-01 -1.44448176e-01 -5.13282955e-01 -5.21647111e-02 9.01985049e-01 6.26869738e-01 3.59343708e-01 5.76770961e-01 -1.17700851e+00 1.41253904e-01 7.79987156e-01 -7.35361457e-01 -1.14651024e-01 3.59112769e-01 3.87376368e-01 8.50470841e-01 -3.88139427e-01 6.22360289e-01 1.23536909e+00 4.89302874e-01 3.64141256e-01 -1.39148462e+00 -5.70913374e-01 3.18136871e-01 2.05537707e-01 -1.37769258e+00 -3.18178833e-01 5.80905616e-01 3.64077352e-02 9.89719272e-01 1.84590757e-01 7.73173034e-01 7.23089755e-01 5.64316511e-01 1.10640454e+00 1.44543433e+00 -4.80519056e-01 9.18884397e-01 -9.41608697e-02 -4.93280470e-01 7.92004347e-01 5.29648736e-02 5.82193553e-01 -4.84495431e-01 -4.66948271e-01 5.65584242e-01 -4.78173226e-01 -9.62976366e-02 -7.18176484e-01 -8.86838555e-01 1.03294492e+00 3.95464867e-01 4.83278409e-02 -4.48563546e-01 6.82093322e-01 2.71820635e-01 4.26424652e-01 4.03545320e-01 4.96987104e-01 -3.71801794e-01 -5.63620031e-01 -1.06565571e+00 7.22378612e-01 8.97901773e-01 9.27504241e-01 7.92767882e-01 -2.85908300e-02 -4.21113484e-02 4.27526832e-01 3.15698028e-01 3.50968242e-01 2.37330765e-01 -1.41178548e+00 3.66390467e-01 9.49578956e-02 4.79435116e-01 -7.20408559e-01 -3.90568107e-01 -5.21396458e-01 -5.32210469e-01 5.09665012e-01 2.22360760e-01 -4.64602083e-01 -6.52509212e-01 1.90835345e+00 4.83971357e-01 -1.50946667e-02 2.19118968e-01 5.98833978e-01 -3.30575079e-01 6.36461437e-01 -1.43102512e-01 -4.55657482e-01 7.81240702e-01 -1.01118731e+00 -3.83524209e-01 -1.76111460e-01 5.79777181e-01 -5.09622812e-01 8.92755270e-01 6.58971071e-01 -1.11712682e+00 -4.29816209e-02 -1.00390959e+00 6.36597395e-01 -1.71291873e-01 -1.10873189e-02 7.11629272e-01 7.10763276e-01 -1.31379509e+00 1.05192912e+00 -1.26197815e+00 -8.32291394e-02 3.84465337e-01 4.60476011e-01 -9.59362462e-03 -1.81660905e-01 -9.27737772e-01 1.16875207e+00 5.38769662e-01 -9.49975476e-02 -1.32091904e+00 -5.94639897e-01 -8.53400707e-01 -2.25633699e-02 8.39343369e-01 -5.87186754e-01 1.83817554e+00 -6.05018616e-01 -2.16708231e+00 -1.90625623e-01 -1.30782992e-01 -8.03550363e-01 8.57338011e-01 -1.39685884e-01 4.76299018e-01 1.03501521e-01 -7.70687163e-02 7.32236385e-01 5.77111244e-01 -1.29222548e+00 -5.84180593e-01 -6.83111325e-02 3.37828517e-01 4.18781936e-01 2.42787883e-01 -3.68492216e-01 -6.91164508e-02 -1.65643796e-01 -1.67764381e-01 -9.77226138e-01 -7.99649715e-01 -6.18284047e-01 -2.80848682e-01 -2.75558978e-01 4.06292260e-01 -4.15224195e-01 1.25160980e+00 -1.68253899e+00 3.21800649e-01 3.54630232e-01 -5.29766716e-02 -9.79923606e-02 -6.46924600e-02 1.00228310e+00 3.61800164e-01 9.50473025e-02 -5.90591133e-01 -5.36762357e-01 2.71354645e-01 2.84420192e-01 -1.97801064e-03 6.36633098e-01 -1.14199845e-02 6.93674088e-01 -1.20947719e+00 -4.60663825e-01 4.19575959e-01 8.86075571e-02 -9.18931723e-01 -2.46181339e-01 -4.43355620e-01 3.64422500e-01 -5.28159797e-01 4.93094563e-01 4.86865699e-01 6.72073513e-02 5.99356949e-01 3.03761899e-01 -1.03086263e-01 7.40459412e-02 -1.31194043e+00 1.65621018e+00 -6.29457712e-01 3.26189518e-01 1.54252529e-01 -1.12633777e+00 5.11361480e-01 2.73390740e-01 5.69435000e-01 -4.92847443e-01 1.99272260e-01 2.28333041e-01 -1.34643942e-01 1.73250109e-01 5.51375151e-01 -1.23889908e-01 -2.37173498e-01 4.81446236e-01 -1.69423651e-02 -6.19230926e-01 5.15110373e-01 1.10386439e-01 1.23846459e+00 4.44102198e-01 6.14105344e-01 -4.26864296e-01 1.97136641e-01 1.03772059e-01 5.27485192e-01 1.00996375e+00 -3.97699326e-01 9.84423980e-02 6.96898103e-01 -4.15018126e-02 -9.60378706e-01 -9.73154426e-01 -5.54560311e-02 6.81684911e-01 1.20249078e-01 -6.49381816e-01 -9.31868911e-01 -8.04823756e-01 1.14051364e-02 1.15522599e+00 -5.07897258e-01 5.53708971e-02 -4.86686796e-01 -7.32595682e-01 3.34911644e-01 1.95216671e-01 7.49836564e-01 -9.88949120e-01 -1.04940379e+00 4.25640613e-01 -4.90266047e-02 -5.41822314e-01 -3.03613186e-01 2.20205665e-01 -8.72006774e-01 -1.00636005e+00 -5.65809608e-01 -1.67586401e-01 5.40715337e-01 -2.11131647e-01 1.06947219e+00 -1.91513360e-01 9.40060839e-02 6.04720891e-01 -3.45959693e-01 -3.57851326e-01 -5.73543847e-01 2.56857395e-01 -3.12579498e-02 -5.02397716e-01 -3.11312169e-01 -6.54831350e-01 -5.54057419e-01 4.73831780e-02 -7.44255483e-01 2.65415106e-02 6.52084768e-01 1.00713658e+00 3.81598353e-01 2.92283535e-01 6.30780995e-01 -6.90736175e-01 9.08941090e-01 -3.20775092e-01 -1.10263598e+00 1.54634621e-02 -8.68606567e-01 3.60171705e-01 6.72857761e-01 -9.56627503e-02 -9.12678361e-01 1.17431851e-02 -2.02491090e-01 -3.03597480e-01 -2.70519312e-02 7.40302086e-01 2.69090563e-01 -2.01778933e-01 3.71597201e-01 3.88766795e-01 2.35066906e-01 -1.90617532e-01 3.50454330e-01 3.71395350e-01 2.36872658e-02 -9.79827166e-01 5.86733818e-01 6.92785904e-02 1.02546759e-01 -5.73770702e-01 -4.61029679e-01 -1.51713505e-01 -1.68191940e-01 -2.45454580e-01 3.76605153e-01 -5.58589101e-01 -1.02356148e+00 3.48210633e-01 -7.20336735e-01 -8.75067830e-01 -4.65148002e-01 4.00376946e-01 -1.22427952e+00 4.08310741e-01 -7.35883892e-01 -1.15467405e+00 -9.12200660e-02 -1.37214541e+00 9.24876750e-01 2.59581327e-01 6.39225096e-02 -8.72943342e-01 4.85305846e-01 -5.17943408e-03 4.28245127e-01 4.41904426e-01 7.34242201e-01 -4.44422960e-01 -5.74582875e-01 -1.76520757e-02 4.26323920e-01 2.49023110e-01 -1.22064613e-01 -3.02377194e-02 -5.70465028e-01 -8.30313087e-01 -8.27180520e-02 -4.88151968e-01 6.23813331e-01 6.16885483e-01 9.54480767e-01 -4.08189684e-01 -2.38056153e-01 1.44069001e-01 1.66572213e+00 4.17515934e-01 4.76620197e-01 7.68617511e-01 -1.15778938e-01 2.57611305e-01 1.05292273e+00 8.59067142e-01 5.05791843e-01 5.26061535e-01 6.15009487e-01 1.48094907e-01 5.44796765e-01 -2.49350518e-01 6.36939108e-01 6.27181888e-01 -1.85817435e-01 -1.73521921e-01 -8.69270921e-01 5.08077264e-01 -2.08718133e+00 -9.11224544e-01 6.94387674e-01 2.29699659e+00 1.11208415e+00 6.91577494e-02 1.46655902e-01 -4.78067175e-02 3.74948025e-01 1.13532454e-01 -6.75981760e-01 -6.26341462e-01 4.38512146e-01 5.36536753e-01 7.17005014e-01 8.24745178e-01 -9.89514649e-01 1.01119399e+00 6.32451820e+00 1.24510884e+00 -6.82502985e-01 1.59984380e-01 6.53649449e-01 -2.78863221e-01 -1.20223820e-01 2.11847097e-01 -4.98247713e-01 4.27170038e-01 1.15662348e+00 -4.84970391e-01 8.74632955e-01 1.11449182e+00 3.15749288e-01 -7.23059833e-01 -1.03749418e+00 5.29013216e-01 -3.92774105e-01 -1.32494628e+00 -3.46259117e-01 2.45379165e-01 8.58642161e-01 8.44329819e-02 -1.31424859e-01 6.74717128e-01 9.16282117e-01 -8.47561955e-01 8.63350093e-01 2.36806631e-01 1.92063764e-01 -1.36030114e+00 8.36746037e-01 7.03489602e-01 -1.17093468e+00 -1.82233855e-01 -3.54478300e-01 -1.05572611e-01 1.09867446e-01 2.82610834e-01 -8.73456299e-01 1.09463084e+00 4.57956731e-01 2.59341002e-01 -1.58777192e-01 1.10527611e+00 -2.68764615e-01 5.85985303e-01 -5.40409505e-01 -2.37616822e-01 7.11710513e-01 -2.70044088e-01 5.96261024e-01 8.89490724e-01 2.09079117e-01 -1.66457668e-01 4.91761893e-01 9.37577844e-01 3.05456996e-01 -9.82294753e-02 -4.39373791e-01 1.35959730e-01 6.34572387e-01 9.16782677e-01 -6.81064844e-01 -2.58640021e-01 8.46762657e-02 6.80909574e-01 4.58147675e-01 3.33755404e-01 -9.75737691e-01 -2.35163063e-01 4.82544929e-01 -4.22973841e-01 3.62113893e-01 -4.40875798e-01 1.12040870e-01 -8.28927040e-01 -1.33062065e-01 -1.07132840e+00 2.59601116e-01 -2.88461834e-01 -9.58956420e-01 3.03601176e-01 4.89616960e-01 -1.08587801e+00 -7.21701980e-01 -1.94663227e-01 -4.14785117e-01 6.42332077e-01 -1.58213353e+00 -5.89605868e-01 2.80719280e-01 3.42925727e-01 7.27185428e-01 1.48059130e-01 7.49737501e-01 -9.72512960e-02 -6.18855894e-01 5.33034086e-01 4.88870472e-01 -4.07447726e-01 4.22060102e-01 -1.44138134e+00 1.23758920e-01 7.27195323e-01 -2.24010363e-01 3.88078362e-01 8.96957576e-01 -5.81267893e-01 -1.47602367e+00 -8.42625380e-01 2.60791689e-01 -1.22570604e-01 5.52514791e-01 -2.98642274e-02 -2.05421627e-01 7.11889625e-01 5.14127135e-01 -5.14452398e-01 2.19576120e-01 1.09913401e-01 4.44218516e-01 -2.41651153e-03 -1.43828785e+00 8.20012689e-01 6.66747332e-01 -5.86433932e-02 -3.44428957e-01 3.15595299e-01 5.14615834e-01 -6.00275636e-01 -8.85149181e-01 4.17150199e-01 2.99433053e-01 -1.25341868e+00 6.61266387e-01 -2.58458942e-01 2.82698303e-01 -2.42485717e-01 -1.63457438e-01 -1.80400050e+00 4.58792560e-02 -8.83839190e-01 -1.52257949e-01 8.46412361e-01 3.31352592e-01 -1.00262868e+00 9.03185964e-01 3.81507933e-01 -1.04273938e-01 -1.26270628e+00 -1.39491999e+00 -1.00487399e+00 5.47680736e-01 -4.54612941e-01 4.36091244e-01 4.70509499e-01 1.78161934e-01 -1.49150938e-01 -2.92218357e-01 1.98591456e-01 7.62520015e-01 7.86243677e-02 6.67670548e-01 -4.75544482e-01 -6.91892743e-01 -3.53944957e-01 -1.07270159e-01 -9.48449433e-01 1.31962582e-01 -5.42434990e-01 4.11214918e-01 -1.50018716e+00 1.95878983e-01 -5.56133032e-01 -2.15013176e-01 3.93839180e-01 1.36168510e-01 -4.11564708e-01 5.62439978e-01 -5.27060665e-02 -6.79070830e-01 1.06911194e+00 1.23765981e+00 -1.24185178e-02 -3.33722413e-01 1.79669291e-01 -3.82083684e-01 7.30545640e-01 1.17772126e+00 -5.40578365e-01 -6.03075564e-01 -4.11420763e-02 6.35027513e-02 4.03395742e-01 2.58065671e-01 -1.26839519e+00 2.26414293e-01 -4.63895500e-01 2.51439549e-02 -5.10640383e-01 3.34267795e-01 -6.15230381e-01 3.62449586e-02 9.88783658e-01 -2.45100230e-01 1.87332734e-01 4.35601622e-01 6.55574620e-01 -3.58556882e-02 -5.73888779e-01 8.08953643e-01 -3.83586735e-01 -3.40744555e-01 1.35399222e-01 -7.27357030e-01 -1.03645846e-01 1.19144583e+00 1.20380118e-01 -1.84936449e-01 -6.31670058e-01 -6.61080241e-01 5.61104655e-01 6.15829229e-01 -3.53821874e-01 4.04837251e-01 -9.86811459e-01 -4.48634803e-01 -2.42011502e-01 -5.36352158e-01 4.44566682e-02 8.30723271e-02 1.06707060e+00 -5.04122376e-01 4.18943465e-01 9.84185375e-03 -5.13039589e-01 -8.64472151e-01 3.94560009e-01 6.54467523e-01 -8.95984113e-01 -4.10073042e-01 5.31174958e-01 -2.68938571e-01 -6.94044590e-01 1.74464047e-01 -4.59084392e-01 2.83526450e-01 -2.29795456e-01 1.51606709e-01 5.73251843e-01 -6.26045465e-02 1.55595496e-01 -3.14568609e-01 2.85996020e-01 8.32339376e-02 -6.34626925e-01 1.36006105e+00 -9.58567932e-02 1.35707840e-01 2.23757952e-01 6.32808328e-01 -2.00729985e-02 -1.67271590e+00 1.15312859e-01 -1.01735443e-02 -3.64639521e-01 1.83142170e-01 -8.27469468e-01 -9.16784346e-01 4.75656658e-01 4.71150577e-01 6.88951323e-03 1.13064086e+00 -3.10362428e-01 3.12327296e-01 6.62326992e-01 1.07939517e+00 -1.30241358e+00 -1.37358844e-01 5.18178582e-01 8.90000522e-01 -9.13335204e-01 2.75834560e-01 7.48261362e-02 -6.51936531e-01 9.87484992e-01 4.34171617e-01 -3.77833605e-01 4.30671722e-01 4.55247670e-01 -3.90571326e-01 7.17546940e-02 -1.02504587e+00 -2.02864617e-01 -4.14313912e-01 4.78339523e-01 -5.73775768e-02 1.06082976e-01 -5.64690948e-01 2.82387286e-01 -3.32093716e-01 -8.63315910e-03 6.34834230e-01 1.60639155e+00 -4.85768855e-01 -1.46918881e+00 -1.55632660e-01 2.59716898e-01 -2.23869115e-01 -6.36823401e-02 -7.13001564e-02 1.24449563e+00 -4.63140965e-01 1.05060470e+00 -8.10348243e-02 -2.74110436e-01 1.55918375e-01 -1.48173217e-02 7.41137445e-01 -2.29277328e-01 -5.34676731e-01 1.91516906e-01 2.31504619e-01 -1.02983940e+00 -3.83197784e-01 -7.97041178e-01 -1.20131826e+00 -6.69069409e-01 -2.72771567e-01 4.20384824e-01 5.90612590e-01 9.58527625e-01 3.28573883e-01 5.16854763e-01 7.40962148e-01 -1.15768397e+00 -1.35334527e+00 -8.06659997e-01 -5.59178233e-01 -4.68991607e-01 1.41777188e-01 -8.82189453e-01 -4.24340248e-01 -5.55180967e-01]
[4.11870002746582, 2.10054349899292]
a29dfa9f-884c-4ca8-97b8-85e5681cc5e3
vision-based-food-analysis-for-automatic
2108.02947
null
https://arxiv.org/abs/2108.02947v2
https://arxiv.org/pdf/2108.02947v2.pdf
A review on vision-based analysis for automatic dietary assessment
Background: Maintaining a healthy diet is vital to avoid health-related issues, e.g., undernutrition, obesity and many non-communicable diseases. An indispensable part of the health diet is dietary assessment. Traditional manual recording methods are not only burdensome but time-consuming, and contain substantial biases and errors. Recent advances in Artificial Intelligence (AI), especially computer vision technologies, have made it possible to develop automatic dietary assessment solutions, which are more convenient, less time-consuming and even more accurate to monitor daily food intake. Scope and approach: This review presents Vision-Based Dietary Assessment (VBDA) architectures, including multi-stage architecture and end-to-end one. The multi-stage dietary assessment generally consists of three stages: food image analysis, volume estimation and nutrient derivation. The prosperity of deep learning makes VBDA gradually move to an end-to-end implementation, which applies food images to a single network to directly estimate the nutrition. The recently proposed end-to-end methods are also discussed. We further analyze existing dietary assessment datasets, indicating that one large-scale benchmark is urgently needed, and finally highlight critical challenges and future trends for VBDA. Key findings and conclusions: After thorough exploration, we find that multi-task end-to-end deep learning approaches are one important trend of VBDA. Despite considerable research progress, many challenges remain for VBDA due to the meal complexity. We also provide the latest ideas for future development of VBDA, e.g., fine-grained food analysis and accurate volume estimation. This review aims to encourage researchers to propose more practical solutions for VBDA.
['Shuqiang Jiang', 'Haisheng Li', 'Xiaoxiao Dong', 'TianHao Li', 'Weiqing Min', 'Wei Wang']
2021-08-06
null
null
null
null
['food-recognition']
['computer-vision']
[-1.32621422e-01 -5.11248350e-01 -5.30059755e-01 -4.43637669e-01 -3.43452126e-01 -4.21915710e-01 -1.07154757e-01 8.45955670e-01 -2.62234688e-01 3.01775753e-01 3.57466310e-01 -2.49142852e-03 1.05784588e-01 -1.10446060e+00 -6.42266750e-01 -6.64117038e-01 -3.38701874e-01 3.43889922e-01 -2.22565636e-01 -3.41160148e-02 -2.66807735e-01 -1.00566410e-01 -1.43737650e+00 1.96859509e-01 9.76565957e-01 1.21759963e+00 1.50649562e-01 5.82654893e-01 -3.48658502e-01 5.04429519e-01 -1.15889512e-01 -3.19362909e-01 3.30257684e-01 -6.37427509e-01 -3.10757995e-01 1.24683149e-01 6.35447323e-01 -9.60374415e-01 7.65587762e-02 1.37226832e+00 6.40627623e-01 6.20271601e-02 6.76281035e-01 -9.55543518e-01 -1.20009673e+00 7.47546136e-01 -9.12421763e-01 1.89637408e-01 7.25413114e-02 4.32742506e-01 6.71118557e-01 -6.30370021e-01 -2.28380933e-01 1.12293088e+00 1.19717968e+00 3.90493989e-01 -9.99154091e-01 -6.36089921e-01 4.08737957e-01 2.07894996e-01 -1.02181029e+00 -4.80854273e-01 6.61009490e-01 -6.40223742e-01 8.09942007e-01 1.60998985e-01 1.32895911e+00 7.87746131e-01 4.13533539e-01 8.08124423e-01 9.69842434e-01 -2.72036403e-01 2.78036445e-01 -1.83420688e-01 1.97466537e-01 9.50952232e-01 8.95273149e-01 2.50085503e-01 -5.81790656e-02 5.97855859e-02 8.77429366e-01 1.60571605e-01 2.24266320e-01 -7.10186660e-01 -1.24111235e+00 1.18966281e+00 5.71628213e-01 1.63385585e-01 -7.13753223e-01 1.18888505e-01 7.86523819e-01 1.27098054e-01 7.35376716e-01 1.25546297e-02 -4.28544879e-01 5.16862035e-01 -1.23758709e+00 4.49640036e-01 6.20130181e-01 5.14034510e-01 9.28613544e-02 2.42967650e-01 -2.04146385e-01 8.91473830e-01 6.67407930e-01 1.03748429e+00 5.27860463e-01 -7.76414096e-01 1.69591054e-01 4.69347864e-01 1.22436367e-01 -1.31464028e+00 -9.95330334e-01 -4.97739047e-01 -1.37962270e+00 2.62573332e-01 7.29848027e-01 -1.31352916e-01 -7.28731692e-01 1.51027763e+00 7.06424475e-01 -3.44237536e-01 -3.70685846e-01 1.34232628e+00 1.38154280e+00 4.31836337e-01 6.64100587e-01 -2.19486386e-01 1.73249328e+00 -1.11066008e+00 -8.07439983e-01 -2.66516730e-02 3.75559986e-01 -5.68670571e-01 9.72523093e-01 4.91288275e-01 -1.32453084e+00 -6.64317608e-01 -1.15237021e+00 -2.99946070e-01 -5.05891919e-01 3.42559665e-01 8.44982564e-01 8.78308117e-01 -6.40660226e-01 3.99508446e-01 -9.45598185e-01 -3.38989705e-01 5.07826269e-01 2.08854750e-01 7.56480247e-02 -1.22765034e-01 -1.21938789e+00 7.47581065e-01 1.40397355e-01 3.20253402e-01 -9.31503713e-01 -1.18810451e+00 -1.11870039e+00 -6.62933215e-02 2.14382321e-01 -1.04414177e+00 1.38810849e+00 -9.50005710e-01 -1.41386700e+00 9.72890735e-01 3.82615060e-01 -5.35362661e-01 4.67451960e-01 -1.30227759e-01 -3.42688143e-01 9.09930542e-02 1.27005830e-01 7.02389300e-01 4.08438057e-01 -8.03149045e-01 -5.31052768e-01 -7.13619530e-01 -1.89435482e-01 1.56037763e-01 -2.18908921e-01 -9.88969114e-04 1.69175237e-01 -5.55565834e-01 -2.09341541e-01 -4.16483670e-01 -3.59748632e-01 8.39626551e-01 -1.58596069e-01 -2.60194153e-01 2.36066785e-02 -1.00030768e+00 1.20531809e+00 -1.61149466e+00 -1.03971153e-01 -2.31149718e-01 5.29763818e-01 3.78787726e-01 -4.81023416e-02 -1.71658173e-01 1.59693122e-01 -1.78958997e-01 -3.35809648e-01 9.59893130e-03 2.19751939e-01 -2.72453934e-01 4.71101224e-01 9.17919099e-01 -1.32856071e-01 1.22569382e+00 -1.20049560e+00 -5.10918081e-01 7.90405273e-01 5.01329005e-01 -5.83853126e-01 -6.60801455e-02 -5.83680630e-01 2.63461798e-01 -3.28406751e-01 9.72445786e-01 8.52477670e-01 -2.88653582e-01 1.80975929e-01 -7.13272989e-01 -3.33931029e-01 -8.16345811e-02 -1.09912395e+00 1.85280657e+00 -1.73848629e-01 -6.84719458e-02 4.46192324e-01 -1.29442894e+00 8.05002451e-01 1.20184153e-01 8.47504318e-01 -1.25017130e+00 4.51285392e-01 2.79370517e-01 -1.41228577e-02 -7.96711028e-01 5.18369824e-02 -2.14413047e-01 -1.68101955e-02 2.19441384e-01 -2.11183459e-01 3.37056369e-01 3.49951446e-01 -4.40466583e-01 3.75647515e-01 3.28465849e-01 9.16391313e-01 -6.73504293e-01 4.12421644e-01 5.78312457e-01 4.05861914e-01 1.79408759e-01 -8.81294608e-01 2.15767220e-01 -2.11568445e-01 -9.59570110e-01 -1.23809481e+00 -1.01945305e+00 -1.19129263e-01 1.14093041e+00 -6.42706007e-02 1.89742610e-01 -7.61945486e-01 -4.98274356e-01 4.44457173e-01 2.49737754e-01 -6.66157126e-01 -6.06591627e-02 -6.53079391e-01 -1.26495612e+00 3.06070775e-01 8.19654584e-01 8.47074509e-01 -8.49423707e-01 -8.24474931e-01 6.77032590e-01 -3.04849684e-01 -5.54980814e-01 -6.95318878e-01 1.76056735e-02 -1.00694835e+00 -1.20080423e+00 -1.34339821e+00 -9.40142214e-01 3.15378249e-01 5.05386770e-01 1.59299576e+00 1.36749402e-01 -5.29946327e-01 2.97651678e-01 -1.71248391e-01 -7.43287802e-01 -3.25957537e-01 -1.30695596e-01 6.02873340e-02 -5.43354392e-01 1.13254654e+00 -3.13993067e-01 -1.57233632e+00 5.88272139e-02 -6.00614011e-01 -1.57559499e-01 5.03413677e-01 5.11103868e-01 1.14232063e+00 1.78660765e-01 7.01251924e-01 -5.40014207e-01 4.21812862e-01 -7.22710371e-01 -7.24398553e-01 7.71970898e-02 -7.51629114e-01 -3.42776328e-01 6.48561239e-01 -5.01398444e-01 -7.18856633e-01 2.17163309e-01 -4.01802003e-01 -5.74317127e-02 -2.30171546e-01 5.25476635e-01 2.89657939e-04 1.31139770e-01 8.12735200e-01 8.29342678e-02 4.24711943e-01 -6.73324227e-01 5.13130069e-01 3.64286989e-01 3.79610360e-01 -3.17179441e-01 2.16926277e-01 3.70286107e-01 -3.96978483e-02 -7.42470503e-01 -9.52117264e-01 -3.99552524e-01 -5.42778075e-01 -2.01195702e-01 1.14451480e+00 -1.21267116e+00 -6.96484625e-01 6.28952920e-01 -5.02531230e-01 -6.11860156e-01 -1.58449456e-01 6.08383894e-01 -2.87172079e-01 5.31512618e-01 -7.03063428e-01 -4.59135324e-01 -1.25599134e+00 -1.09143543e+00 7.57429361e-01 3.01334620e-01 -2.80555248e-01 -1.22901952e+00 2.58692682e-01 3.58132750e-01 6.91578150e-01 5.02035499e-01 1.11050200e+00 -1.10922791e-01 6.04631659e-03 2.27956951e-01 -5.51441967e-01 1.52773097e-01 2.36010924e-01 -2.05304086e-01 -6.13080919e-01 -2.35747397e-01 4.23844345e-02 -4.43968356e-01 9.33423221e-01 1.45894861e+00 1.02291131e+00 -1.52335778e-01 -2.43109763e-01 6.62558675e-01 1.65080512e+00 2.62188047e-01 2.08571613e-01 2.93163091e-01 7.04646170e-01 5.95720887e-01 5.23337007e-01 2.18046144e-01 9.01135087e-01 5.51257491e-01 5.12740314e-01 -6.17876291e-01 -7.71873116e-01 -1.02482967e-01 3.24317306e-01 1.02638280e+00 2.70052969e-01 -1.85636356e-02 -3.37276399e-01 7.05717802e-01 -1.41960156e+00 -7.77323365e-01 -5.87440968e-01 1.96009088e+00 8.85727048e-01 -5.18259943e-01 9.08824801e-01 4.81965132e-02 4.16883141e-01 -1.00492507e-01 -1.07203603e+00 -4.17147338e-01 2.88990855e-01 5.12933172e-02 5.70819974e-01 9.75914001e-02 -1.62539697e+00 2.51445770e-01 6.65724993e+00 2.56704062e-01 -9.88695681e-01 2.63321936e-01 7.60457814e-01 -1.70424253e-01 -2.32578442e-02 -1.13407278e+00 -6.97482347e-01 6.05217218e-01 8.49693537e-01 2.33969212e-01 4.63533223e-01 8.76622438e-01 3.71242523e-01 -2.01749861e-01 -1.00928974e+00 8.49567831e-01 6.42037317e-02 -1.06367338e+00 -2.33672827e-01 -5.78958839e-02 4.82285470e-01 8.63636881e-02 3.61586735e-02 3.79339784e-01 3.14348131e-01 -8.24984491e-01 7.00861692e-01 1.79451019e-01 1.00405931e+00 -6.38476491e-01 5.26672065e-01 8.33510682e-02 -1.76860750e+00 -7.11472556e-02 -9.77659464e-01 -1.23319805e-01 -5.24152890e-02 1.09075308e+00 4.01338823e-02 2.89329261e-01 8.84195328e-01 6.73324883e-01 -2.04784751e-01 1.34290457e+00 3.23235482e-01 2.64873356e-01 -2.82798767e-01 -1.36488438e-01 1.94238007e-01 -2.91900873e-01 -4.18011695e-02 1.28054929e+00 4.10638064e-01 1.85436353e-01 6.33705616e-01 1.00352776e+00 1.07391290e-01 5.28619885e-01 -3.64926457e-01 -6.28921809e-03 -1.13163941e-01 1.31383157e+00 -9.80053067e-01 -2.57014066e-01 -8.33841085e-01 7.82998145e-01 3.94526832e-02 -1.19518414e-01 -8.76149118e-01 2.47172341e-02 7.52369642e-01 7.33998120e-02 1.66074067e-01 -3.31715494e-02 -3.12796056e-01 -1.06892359e+00 -2.26773500e-01 -1.07046771e+00 6.32537603e-01 -2.78679073e-01 -1.61116457e+00 -4.50514257e-02 -2.55822223e-02 -8.80407512e-01 2.77423322e-01 -6.80040300e-01 -2.01226696e-01 6.83324993e-01 -1.96490371e+00 -9.92747903e-01 -5.55348337e-01 3.56203258e-01 8.53013515e-01 1.73704773e-01 1.03346753e+00 9.10467267e-01 -6.91230536e-01 7.52946079e-01 2.18097910e-01 2.56394744e-01 4.87588644e-01 -1.30127347e+00 2.91227758e-01 3.44724745e-01 -2.48279020e-01 1.64270833e-01 5.22361755e-01 -9.62499082e-01 -1.61938190e+00 -1.48035944e+00 6.45139396e-01 -1.33440852e-01 3.85093719e-01 -1.34515494e-01 -6.91640854e-01 2.08300650e-01 9.79496315e-02 -4.29609157e-02 8.94733012e-01 -2.06693739e-01 -1.43808678e-01 -4.12475854e-01 -1.56228542e+00 2.34617859e-01 9.93923843e-01 1.34874493e-01 -3.37797672e-01 3.82802784e-01 5.56572318e-01 -4.89019185e-01 -1.05484188e+00 2.16759250e-01 1.10763633e+00 -8.24742734e-01 1.47002888e+00 -2.49844968e-01 6.16946995e-01 -2.01113418e-01 3.45182698e-03 -1.37778509e+00 -8.28873754e-01 2.91907698e-01 -4.76821631e-01 8.41404676e-01 -3.44965458e-02 -3.04571569e-01 8.09576809e-01 4.23677623e-01 -1.09738797e-01 -6.95512950e-01 -3.93997878e-01 -5.87010443e-01 5.97598195e-01 1.51069120e-01 9.36694860e-01 1.05315113e+00 -1.17035232e-01 1.72127485e-01 -4.35012668e-01 -3.02751772e-02 1.19205654e+00 3.05949509e-01 2.40348250e-01 -1.33498979e+00 -1.35174438e-01 -6.86460257e-01 1.57024056e-01 -9.22537446e-01 -4.56445038e-01 -9.80662048e-01 7.05567822e-02 -1.89153993e+00 5.27756155e-01 -2.41238400e-01 -3.38375181e-01 3.91081899e-01 -1.74199134e-01 5.13014734e-01 -1.51837409e-01 -5.43115474e-02 -3.10250998e-01 2.51371235e-01 1.70910358e+00 -6.82932198e-01 -3.67681265e-01 -2.93283284e-01 -9.92154360e-01 6.37584925e-01 9.94828761e-01 -3.54880214e-01 -4.89756286e-01 -7.22817242e-01 3.16023856e-01 -1.77493379e-01 4.00714934e-01 -8.01533818e-01 -2.09024474e-01 -2.52182454e-01 1.08761418e+00 -7.55752504e-01 -2.84410655e-01 -9.39853609e-01 1.63324267e-01 1.01515675e+00 -1.87526211e-01 -2.50661597e-02 2.81587929e-01 3.80075037e-01 2.60393202e-01 -5.96487410e-02 9.52628493e-01 -7.02002525e-01 -5.87105274e-01 4.56037551e-01 -3.57616603e-01 -2.07023874e-01 9.90243852e-01 -6.38071448e-02 -2.68773317e-01 6.54350221e-02 -6.62243009e-01 4.97774392e-01 7.31363073e-02 1.05433419e-01 3.24105203e-01 -1.62002969e+00 -9.48149562e-01 1.47608235e-01 -5.75774908e-03 -1.47872508e-01 6.36836290e-01 1.00448823e+00 -6.84502184e-01 6.31520510e-01 -4.02932435e-01 -5.68396270e-01 -9.86321449e-01 1.01699150e+00 5.27455986e-01 -3.75588626e-01 -1.00378728e+00 5.73887706e-01 3.02833825e-01 -3.35507572e-01 4.29855108e-01 -1.05862105e+00 -5.26956201e-01 1.98332071e-01 1.06278884e+00 7.18594313e-01 7.82540143e-02 -3.60007137e-01 -2.35789075e-01 7.70733535e-01 2.35717893e-01 9.49885190e-01 1.27462280e+00 -5.55367529e-01 1.15129396e-01 1.50376990e-01 8.15925419e-01 -5.53252637e-01 -1.03999150e+00 -7.40124062e-02 -4.81759518e-01 -8.24318752e-02 6.28895819e-01 -1.29938412e+00 -1.63795841e+00 1.08501363e+00 1.31339335e+00 2.88541824e-01 1.30751753e+00 -4.17485803e-01 1.45816326e+00 -1.33493111e-01 3.15039098e-01 -9.22615051e-01 -9.37468037e-02 -1.65135145e-01 8.12263668e-01 -1.58496165e+00 3.98948044e-01 -2.25754052e-01 -2.96866834e-01 1.10011053e+00 5.31424344e-01 -1.65598541e-01 9.03246343e-01 1.57963976e-01 2.97112763e-01 -4.38637316e-01 3.27097923e-02 -1.19555026e-01 3.78524244e-01 9.45913196e-01 6.46147966e-01 5.43696046e-01 -5.21970451e-01 8.14997733e-01 1.12971447e-01 4.64096129e-01 -1.70020640e-01 3.24980468e-01 -7.06836462e-01 -7.77081549e-01 -5.36482871e-01 8.33998144e-01 -5.69285929e-01 -1.47724807e-01 1.53048590e-01 5.90116620e-01 3.39035094e-01 9.43906844e-01 2.81200465e-02 1.48734763e-01 3.36402565e-01 1.95132662e-02 7.99210012e-01 -3.40123981e-01 -8.62047493e-01 4.13060874e-01 2.86747999e-02 -5.22502959e-01 -7.26555288e-01 -6.55649841e-01 -1.13448632e+00 -7.67288268e-01 -4.33333628e-02 -4.53506410e-01 6.27309799e-01 6.34336174e-01 9.27921236e-02 5.88847399e-01 1.38131618e-01 -8.32414925e-01 -7.89582849e-01 -7.85415769e-01 -7.16738880e-01 2.71354616e-01 3.00638556e-01 -8.44646394e-01 9.95123535e-02 1.27303869e-01]
[11.563934326171875, 4.40177583694458]
ddb9ef95-ca21-4e07-a5a2-39272fc0cb70
polarized-color-image-denoising
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Polarized_Color_Image_Denoising_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Polarized_Color_Image_Denoising_CVPR_2023_paper.pdf
Polarized Color Image Denoising
Single-chip polarized color photography provides both visual textures and object surface information in one snapshot. However, the use of an additional directional polarizing filter array tends to lower photon count and SNR, when compared to conventional color imaging. As a result, such a bilayer structure usually leads to unpleasant noisy images and undermines performance of polarization analysis, especially in low-light conditions. It is a challenge for traditional image processing pipelines owing to the fact that the physical constraints exerted implicitly in the channels are excessively complicated. In this paper, we propose to tackle this issue through a noise modeling method for realistic data synthesis and a powerful network structure inspired by vision Transformer. A real-world polarized color image dataset of paired raw short-exposed noisy images and long-exposed reference images is captured for experimental evaluation, which has demonstrated the effectiveness of our approaches for data synthesis and polarized color image denoising.
['Yinqiang Zheng', 'Mingdeng Cao', 'Haiyang Jiang', 'Zhuoxiao Li']
2023-01-01
null
null
null
cvpr-2023-1
['color-image-denoising']
['computer-vision']
[ 6.95209861e-01 -2.19460621e-01 5.88526964e-01 -1.35729432e-01 -6.00794077e-01 -5.65076530e-01 2.95065612e-01 -3.13803613e-01 -3.07118088e-01 7.60485470e-01 -1.42354339e-01 -7.13755414e-02 1.40070260e-01 -6.41630173e-01 -6.46317005e-01 -1.50978386e+00 3.63264561e-01 -1.79586858e-01 3.46925445e-02 1.40028194e-01 3.02817583e-01 4.17174369e-01 -1.75664783e+00 5.70897385e-02 8.15393746e-01 1.07737827e+00 2.30815887e-01 3.03114384e-01 -1.77139025e-02 4.02995020e-01 -4.41871405e-01 -6.58758283e-01 3.33349168e-01 -3.94674450e-01 -6.33257851e-02 9.99219567e-02 4.31630492e-01 -3.82938832e-01 -2.60351568e-01 1.57097900e+00 6.08131409e-01 -1.68605864e-01 3.07530373e-01 -9.33004081e-01 -6.07471526e-01 -1.10559361e-02 -5.83657801e-01 -5.87651804e-02 -7.39995241e-02 6.77947104e-01 3.83173347e-01 -6.31935477e-01 6.41606331e-01 8.46840262e-01 4.80746835e-01 5.24319053e-01 -1.56274211e+00 -5.66564679e-01 -2.59740978e-01 1.11025147e-01 -7.71225393e-01 -7.19734192e-01 1.26616907e+00 -1.07302308e-01 4.22014683e-01 1.92962345e-02 8.42412591e-01 1.20122576e+00 2.70629704e-01 4.61251987e-03 1.92626858e+00 -2.27609023e-01 2.38784179e-01 8.25631395e-02 2.04875711e-02 4.37412679e-01 6.38378024e-01 2.88038641e-01 -7.70345151e-01 1.99792418e-03 4.40663099e-01 -1.77546740e-01 -6.24338865e-01 -2.99037695e-01 -9.08862948e-01 -1.09311379e-01 2.39537507e-01 -8.99539515e-03 -3.75643045e-01 -1.04362462e-02 -2.62199389e-03 3.63888852e-02 5.94329059e-01 7.18461335e-01 1.82196811e-01 6.42713159e-02 -5.59316635e-01 -1.59934103e-01 4.51021373e-01 2.59978920e-01 8.10495555e-01 1.63456481e-02 1.83437031e-03 8.50779176e-01 1.50833309e-01 9.67986643e-01 -3.23147953e-01 -1.42547512e+00 1.08645663e-01 3.82770121e-01 3.57750177e-01 -1.14084113e+00 -2.56494820e-01 -4.42419797e-01 -9.64231074e-01 6.10824049e-01 6.99911118e-01 -9.83850285e-02 -1.03815031e+00 1.61874104e+00 2.85865426e-01 1.14012763e-01 1.93788841e-01 1.29555953e+00 5.30923843e-01 7.02783287e-01 -2.36060992e-01 -7.63255596e-01 1.21284306e+00 -4.94486094e-01 -9.73602653e-01 -1.42447501e-01 -1.90767810e-01 -8.50840807e-01 1.16600132e+00 9.95198667e-01 -1.24639368e+00 -2.07066357e-01 -1.15158784e+00 -8.01328421e-02 1.91339757e-02 1.39205575e-01 5.46203554e-01 8.54281425e-01 -7.81240582e-01 4.99037981e-01 -7.02546716e-01 -1.85142718e-02 5.41822970e-01 4.88110892e-02 -3.07117313e-01 -6.51393652e-01 -6.18203044e-01 4.27482367e-01 -1.71631873e-01 4.99182135e-01 -6.29248023e-01 -7.79431343e-01 -3.14002573e-01 -1.34194180e-01 2.99552381e-01 -5.73361814e-01 5.27734339e-01 -6.07923567e-01 -1.96275318e+00 7.43737757e-01 -1.44911081e-01 -9.41996500e-02 2.15709180e-01 -1.40216826e-02 -4.04204071e-01 4.77220148e-01 -2.61864156e-01 9.13512036e-02 1.04656565e+00 -1.57196248e+00 -7.68572316e-02 -5.25518715e-01 -8.17142725e-02 8.19194168e-02 -2.18586370e-01 -4.43590693e-02 -5.99340856e-01 -8.95613357e-02 4.39869374e-01 -7.03889847e-01 -2.80298889e-01 4.57910039e-02 -5.99644780e-01 6.26077294e-01 9.27471638e-01 -4.91129518e-01 6.34077370e-01 -2.30368710e+00 -1.82015046e-01 3.04142796e-02 2.31829911e-01 4.16424245e-01 -2.54375786e-01 2.78966844e-01 -2.68201400e-02 -1.20402977e-01 -2.88831860e-01 -2.44390473e-01 -6.09857440e-01 5.07632196e-02 6.38635457e-03 7.67955303e-01 1.78988397e-01 4.35325027e-01 -6.79853857e-01 -2.12687626e-01 1.18260540e-01 6.96386397e-01 -4.90372866e-01 1.91712931e-01 -8.76394957e-02 8.21970701e-01 -8.10658410e-02 8.38341534e-01 1.31622815e+00 -1.62384406e-01 3.75094473e-01 -8.84508252e-01 -5.47896862e-01 7.95918331e-02 -9.69954550e-01 1.54516447e+00 -3.42375964e-01 4.98093754e-01 5.27548373e-01 -9.56120610e-01 9.46129441e-01 8.74340236e-02 3.73254299e-01 -1.32202387e+00 9.14742574e-02 3.46683085e-01 -2.23100930e-01 -8.10257614e-01 2.50026196e-01 -4.63920265e-01 2.68296897e-01 2.72731870e-01 -2.30412826e-01 -2.39847586e-01 1.17923401e-01 -4.20602299e-02 8.23255420e-01 2.85894368e-02 -4.00999457e-01 -1.94529623e-01 3.77659172e-01 -1.84945434e-01 7.24252284e-01 5.85139096e-01 -9.85603686e-03 8.29957247e-01 7.55766988e-01 -3.86188835e-01 -1.22718179e+00 -8.51818979e-01 -3.25250961e-02 2.20845774e-01 6.18056595e-01 -1.42520726e-01 -5.97594976e-01 -1.12393409e-01 -3.07096243e-01 3.74942034e-01 -1.15776502e-01 2.26021130e-02 -4.69684303e-01 -1.23813629e+00 3.54182243e-01 -1.29165292e-01 7.14789331e-01 -6.31658733e-01 -4.72343981e-01 -4.11086231e-02 -2.45612517e-01 -1.22193706e+00 3.06074798e-01 1.16230890e-01 -5.09018838e-01 -1.35327792e+00 -4.52128351e-01 -2.96253324e-01 8.41984451e-01 6.15581453e-01 9.74129498e-01 -1.42133847e-01 -3.19779545e-01 3.20435762e-01 -5.71606383e-02 -2.74908513e-01 -1.47606522e-01 -6.97507322e-01 6.10630028e-02 5.02541125e-01 4.88761216e-02 -8.29270422e-01 -1.00580502e+00 1.88218027e-01 -9.07775700e-01 3.66534561e-01 6.30507350e-01 1.12755239e+00 6.48173928e-01 1.30456507e-01 3.26811373e-01 -7.50673890e-01 3.59142751e-01 5.42300344e-02 -1.18627608e+00 1.15334146e-01 -6.65329158e-01 -1.22951403e-01 6.88627124e-01 -1.79082513e-01 -1.73978806e+00 -1.88239645e-02 1.34737864e-01 3.49368304e-02 -2.16877326e-01 3.73752296e-01 -3.37434500e-01 -4.06762540e-01 5.28651476e-01 1.84772655e-01 4.11400944e-01 -4.53547180e-01 3.31443474e-02 5.31080186e-01 6.64407849e-01 -4.33323860e-01 7.73816705e-01 1.04210305e+00 4.25964266e-01 -1.17131650e+00 -5.35717010e-01 -1.96243450e-01 6.32034913e-02 -5.02250254e-01 6.79158866e-01 -9.73549902e-01 -1.06569970e+00 1.01785541e+00 -1.06577408e+00 -1.99128479e-01 7.80436397e-02 4.68087077e-01 -3.46775316e-02 5.38632631e-01 -6.27293169e-01 -9.26233172e-01 -1.47331640e-01 -1.20924449e+00 8.49456191e-01 4.72829372e-01 7.08595932e-01 -4.36235398e-01 -7.23283365e-02 6.36501431e-01 5.96244752e-01 2.18092903e-01 9.23296154e-01 5.79403579e-01 -9.64619219e-01 -9.94924307e-02 -4.90816236e-01 8.63160014e-01 -1.20861344e-01 1.52522191e-01 -1.29711759e+00 -7.05137700e-02 3.03411037e-01 -4.81123239e-01 8.36919248e-01 4.99067456e-01 1.30413616e+00 1.12002626e-01 -1.32958572e-02 9.06567752e-01 1.73060405e+00 1.59536272e-01 1.05723655e+00 1.21905252e-01 6.70623720e-01 7.98676729e-01 4.98152256e-01 1.80798665e-01 -9.66430604e-02 7.13482857e-01 5.80331743e-01 -3.98286968e-01 -3.23785663e-01 2.72333592e-01 1.66516393e-01 6.13030732e-01 -2.74738997e-01 -5.27343333e-01 -8.07611227e-01 3.07549536e-01 -1.26760054e+00 -7.23284125e-01 -6.45516455e-01 2.12953353e+00 6.70188665e-01 -2.95021355e-01 -5.32683849e-01 1.18902829e-02 6.47237778e-01 1.01735629e-01 -3.62771720e-01 1.75053373e-01 -6.88124835e-01 1.83383927e-01 5.92355251e-01 3.18765163e-01 -7.31506705e-01 3.61407429e-01 5.54100037e+00 5.39579630e-01 -1.67691219e+00 2.03623017e-03 7.68660307e-01 -2.53124148e-01 -5.48077583e-01 5.11041726e-04 -3.09959859e-01 8.06237698e-01 4.49418545e-01 3.75159204e-01 6.37272537e-01 2.07042843e-01 3.76451343e-01 -8.10939312e-01 -5.57316065e-01 1.45481122e+00 -5.40669914e-03 -1.18469059e+00 -1.76239803e-01 -4.42215726e-02 5.67191005e-01 1.39101846e-02 4.72512394e-01 -5.93063176e-01 -2.78830141e-01 -6.93707585e-01 3.52970630e-01 6.81721330e-01 8.44318748e-01 -5.79643309e-01 6.11558437e-01 1.37429565e-01 -3.46983850e-01 -7.21056312e-02 -4.76537019e-01 -2.48537250e-02 4.30050433e-01 1.25599432e+00 -2.82422870e-01 4.55728203e-01 9.78198051e-01 5.91118753e-01 -2.74693459e-01 9.16916072e-01 -1.94489911e-01 7.64473975e-01 -3.82593095e-01 2.76898444e-01 -1.52891845e-01 -7.62543380e-01 7.11873293e-01 6.77036047e-01 4.34702873e-01 2.11461365e-01 -4.04100567e-01 9.76575136e-01 2.22130455e-02 -3.94123703e-01 -7.31121182e-01 -1.45648286e-01 6.47520199e-02 1.75996220e+00 -7.34063089e-01 5.14980778e-02 -5.20072460e-01 6.87362432e-01 -1.86744690e-01 9.28909779e-01 -4.75781530e-01 -2.87871093e-01 7.33159840e-01 2.78601557e-01 -8.61245915e-02 -4.12506819e-01 -5.54048657e-01 -1.35440242e+00 2.65616089e-01 -7.06761122e-01 -2.65298843e-01 -9.23378348e-01 -1.29279006e+00 9.12471265e-02 -5.43702662e-01 -1.13426673e+00 7.89200664e-01 -7.61692762e-01 -3.90892059e-01 1.05034459e+00 -1.89173031e+00 -9.09757197e-01 -6.14062607e-01 4.44711030e-01 -2.72893101e-01 3.24816167e-01 5.61279774e-01 5.47381639e-01 -8.17576587e-01 -1.06192335e-01 3.52505952e-01 -2.84597725e-01 9.86169338e-01 -7.30001152e-01 -1.74496204e-01 1.24826169e+00 -4.60659325e-01 6.69712305e-01 7.39388347e-01 -4.81162786e-01 -1.93860400e+00 -7.23420560e-01 1.61845475e-01 -1.90022022e-01 2.61593461e-01 -3.56475234e-01 -8.80445957e-01 -1.46580622e-01 4.12129074e-01 3.66615623e-01 5.83497465e-01 -4.27357823e-01 -3.11574131e-01 -5.78211546e-01 -1.11029756e+00 4.85681295e-01 8.81726027e-01 -4.64547366e-01 2.06123397e-01 2.43360206e-01 1.92443505e-01 -2.00320736e-01 -4.34105605e-01 4.11503166e-01 6.88180864e-01 -1.37783146e+00 9.50669408e-01 -7.96488225e-02 5.34582794e-01 -7.08030343e-01 5.67957871e-02 -1.25823557e+00 -5.03201447e-02 -7.69302189e-01 4.50257182e-01 1.18861115e+00 2.88839370e-01 -8.73497605e-01 7.26363957e-01 5.67540526e-01 -3.42752814e-01 -3.56499404e-01 -7.26021051e-01 -5.18402696e-01 -5.03350735e-01 -2.18336448e-01 1.20630831e-01 8.71972024e-01 -5.22787333e-01 1.72793999e-01 -4.17597830e-01 5.54660738e-01 1.26818669e+00 1.85773328e-01 6.42499924e-01 -1.19933712e+00 -6.62468150e-02 -2.08615176e-02 -2.34105304e-01 -6.80023015e-01 -9.30979922e-02 -3.93694609e-01 2.71628678e-01 -1.14337099e+00 2.83434451e-01 -5.63133478e-01 -8.26354325e-02 -3.27581428e-02 -7.82543644e-02 9.19269621e-01 -3.87674533e-02 1.04625367e-01 -3.31056267e-01 4.13836926e-01 1.35810220e+00 -1.51017666e-01 -4.27550226e-02 -3.10494304e-01 -6.93556309e-01 4.44753706e-01 6.09243095e-01 -3.49987745e-01 -3.11648220e-01 -5.80219746e-01 8.98127198e-01 -1.59409400e-02 9.08230901e-01 -1.03614104e+00 2.88734764e-01 -1.79346669e-02 3.51088792e-01 -1.87168777e-01 5.76944530e-01 -8.81768703e-01 3.86629701e-01 2.11502448e-01 1.51565745e-01 -6.06843829e-01 -1.51106521e-01 8.00052404e-01 -3.84075195e-01 1.02971442e-01 1.11322355e+00 -1.95412397e-01 -4.38721627e-01 -8.34942833e-02 -1.85929745e-01 -2.97169119e-01 7.75076151e-01 -2.88185000e-01 -1.04881811e+00 -1.95494995e-01 -1.25397488e-01 -1.56893194e-01 9.35504615e-01 -3.62614840e-01 4.94897366e-01 -9.14462090e-01 -2.08292574e-01 4.60037827e-01 4.47371136e-03 -5.71607761e-02 8.21707070e-01 1.41477191e+00 -7.46168971e-01 2.52826791e-02 -3.55465740e-01 -7.48523831e-01 -1.17113841e+00 3.11927557e-01 4.14091080e-01 2.97369868e-01 -6.73914254e-01 7.70475090e-01 1.72901064e-01 2.39551701e-02 -1.58521444e-01 -2.26680562e-01 -1.75386034e-02 6.24946356e-02 4.35352296e-01 5.11297941e-01 2.90581971e-01 -2.34763980e-01 -3.54817472e-02 6.97147608e-01 2.15204403e-01 -4.32768017e-02 1.49718583e+00 -3.99838179e-01 -5.62955201e-01 4.38314043e-02 9.68285263e-01 2.76999563e-01 -1.51111376e+00 -9.04278550e-03 -6.35440767e-01 -7.40876675e-01 3.33793461e-01 -7.10356891e-01 -1.21008289e+00 1.03296709e+00 5.90468347e-01 1.57445326e-01 1.40379739e+00 -2.30656341e-01 6.34202659e-01 6.55627668e-01 3.43016654e-01 -1.28106284e+00 1.55168489e-01 5.50200380e-02 3.83954883e-01 -1.05214608e+00 -1.20642662e-01 -6.96995676e-01 -3.24576795e-01 1.11212087e+00 4.33361441e-01 1.27755687e-01 3.49537849e-01 3.28947514e-01 5.47909439e-01 -3.81373137e-01 -3.89162183e-01 1.19971201e-01 -3.27777475e-01 7.98628569e-01 1.03943393e-01 -3.48718435e-01 -3.29799265e-01 3.10372621e-01 3.74528199e-01 -7.00306445e-02 8.33833992e-01 7.35025108e-01 -6.11476600e-02 -1.07317960e+00 -3.65328103e-01 2.99089044e-01 -7.55017459e-01 4.61587384e-02 3.49618844e-03 2.75264621e-01 1.74496770e-01 7.58162379e-01 -1.46984220e-01 2.58929357e-02 2.59866953e-01 -2.02097688e-02 4.50697094e-01 -3.29437822e-01 -8.02504793e-02 2.78653920e-01 9.06946659e-02 -8.95729423e-01 -1.00635600e+00 -4.69797999e-01 -6.94437146e-01 -1.08874403e-02 -3.21914434e-01 -1.57779321e-01 9.57540512e-01 3.80713403e-01 6.28367543e-01 2.60839432e-01 5.91627121e-01 -8.37120295e-01 -1.59962073e-01 -5.72518170e-01 -9.53275084e-01 5.03720522e-01 3.99317563e-01 -7.16907144e-01 -4.96543348e-01 1.02005349e-02]
[10.459046363830566, -2.630034923553467]
0f7c6b5a-d95c-48dc-b3cc-eb5a8913aff7
open-challenges-in-synthetic-speech-detection
2209.0718
null
https://arxiv.org/abs/2209.07180v3
https://arxiv.org/pdf/2209.07180v3.pdf
Open Challenges in Synthetic Speech Detection
In this paper the current status and open challenges of synthetic speech detection are addressed. The work comprises an initial analysis of available open datasets and of existing detection methods, a description of the requirements for new research datasets compliant with regulations and better representing real-case scenarios, and a discussion of the desired characteristics of future trustworthy detection methods in terms of both functional and non-functional requirements. Compared to other works, based on specific detection solutions or presenting single dataset of synthetic speeches, our paper is meant to orient future state-of-the-art research in the domain, to quickly lessen the current gap between synthesis and detection approaches.
['Dimitrios Tzovaras', 'Konstantinos Votis', 'Patrick Aichroth', 'Artem Yaroshchuk', 'Anastasios Vafeiadis', 'Christoforos Papastergiopoulos', 'Luca Cuccovillo']
2022-09-15
null
null
null
null
['synthetic-speech-detection']
['audio']
[ 5.44034719e-01 6.46056533e-01 -6.12415560e-02 -2.57130891e-01 -6.98445082e-01 -7.84979522e-01 7.95751214e-01 -6.29800558e-02 8.40513632e-02 5.26856899e-01 5.25777817e-01 -3.29066902e-01 -2.95689911e-01 -4.57980305e-01 -2.11097509e-01 -3.61208841e-02 1.77328333e-01 4.03395951e-01 4.40639973e-01 -6.12620115e-01 1.39625937e-01 3.86706799e-01 -2.37630939e+00 6.48984253e-01 4.51817274e-01 7.88488686e-01 5.26214987e-02 7.52525568e-01 -1.95788890e-01 7.81604886e-01 -1.50958776e+00 -2.99099416e-01 4.55756225e-02 -4.00701851e-01 -8.48638952e-01 2.99811512e-01 4.61377412e-01 1.30332962e-01 -1.23295508e-01 1.07265496e+00 8.15102041e-01 -4.17406037e-02 1.68093383e-01 -1.34280717e+00 -8.33038330e-01 1.12311292e+00 2.41536215e-01 1.42624244e-01 1.13453555e+00 9.19749364e-02 8.73941720e-01 -7.45550513e-01 8.60794544e-01 1.18303919e+00 3.46943319e-01 9.83320117e-01 -7.66027570e-01 -7.06190288e-01 1.09426584e-02 -5.09332158e-02 -1.28655529e+00 -1.07602751e+00 5.49646378e-01 -5.20889342e-01 1.35303020e+00 1.03175414e+00 4.97859389e-01 1.83972526e+00 -6.64253652e-01 6.88782513e-01 8.82558525e-01 -7.40661144e-01 1.87563643e-01 4.49872285e-01 2.82582603e-02 2.45400041e-01 2.64786541e-01 2.58022577e-01 -6.98719561e-01 -3.44496310e-01 1.41805992e-01 -1.11951840e+00 -2.64573455e-01 -1.94648251e-01 -1.42182839e+00 3.81972939e-01 -6.98058188e-01 7.94091105e-01 -8.33452046e-02 -2.71686226e-01 7.12267458e-01 5.46786547e-01 5.25574565e-01 2.63301462e-01 -3.58102560e-01 -6.06123924e-01 -1.24626589e+00 4.23744947e-01 1.28041649e+00 1.51917565e+00 2.02229410e-01 7.01927781e-01 -2.99968094e-01 9.74830687e-01 5.57871640e-01 5.60176849e-01 5.68463564e-01 -9.26790118e-01 5.02669454e-01 5.71919262e-01 8.56442824e-02 -7.26049840e-01 -2.08206177e-01 -2.52158284e-01 9.80557781e-03 -1.01155981e-01 1.45813107e-01 -2.06611544e-01 -6.00035548e-01 1.43281543e+00 1.02637662e-02 -1.53963715e-01 4.04908955e-01 5.83289623e-01 1.19868267e+00 6.13315105e-01 -4.38776225e-01 -6.31846368e-01 1.47013962e+00 -4.81126696e-01 -1.23550832e+00 -3.23744774e-01 4.77703810e-01 -1.29174590e+00 1.04186177e+00 3.28100801e-01 -1.03732324e+00 -3.67161781e-01 -1.30000114e+00 3.28460366e-01 -5.75288415e-01 1.79894909e-01 4.97927554e-02 1.66939223e+00 -1.01401138e+00 -5.68820015e-02 -3.23475629e-01 -8.84525299e-01 -5.66274114e-02 1.23966765e-02 -1.45136043e-01 5.47168672e-01 -1.45422292e+00 9.41775322e-01 3.76209229e-01 -4.15295452e-01 -8.94227982e-01 -6.68072283e-01 -8.39808524e-01 -3.21478307e-01 7.48306394e-01 -1.12038337e-01 1.55854285e+00 -6.36556327e-01 -1.66334188e+00 9.50967848e-01 2.19077498e-01 -3.37953717e-01 6.31722569e-01 1.32132589e-03 -1.36317253e+00 -1.32491797e-01 2.37696931e-01 9.98674855e-02 8.41210186e-01 -1.22794771e+00 -5.82617760e-01 -6.22114278e-02 -6.21192828e-02 -2.14293897e-01 -4.03335154e-01 1.15414584e+00 -1.14570051e-01 -8.20269585e-01 -3.05228502e-01 -7.31156111e-01 2.19883651e-01 -3.69540066e-01 -5.14876544e-01 -2.56910652e-01 1.27292323e+00 -3.85219187e-01 1.71531010e+00 -2.08927226e+00 -1.99442342e-01 -1.63553908e-01 -5.97038344e-02 7.35525966e-01 -1.81686908e-01 1.05128515e+00 -2.03460246e-01 3.87982756e-01 -1.46634862e-01 -2.65783638e-01 3.48971218e-01 2.05141738e-01 -5.14053404e-01 5.63958287e-01 1.45257553e-02 3.79380405e-01 -9.83388484e-01 -3.62848341e-01 4.81278986e-01 2.00974494e-01 3.27376306e-01 1.74739823e-01 -3.86917233e-01 2.41518803e-02 -1.44843310e-01 8.48409951e-01 5.02208233e-01 5.30468225e-01 1.32035673e-01 8.20268616e-02 -7.95889795e-01 5.81525624e-01 -1.67835224e+00 1.28240442e+00 -3.37560922e-01 8.12099457e-01 3.52451086e-01 -9.95012999e-01 1.26446927e+00 9.70041573e-01 3.16780597e-01 -5.40126145e-01 1.36380300e-01 5.30228853e-01 -1.08910449e-01 -7.69754827e-01 5.04072547e-01 4.91469681e-01 -3.87197405e-01 5.86067617e-01 2.11694941e-01 -3.41654420e-01 7.02683508e-01 -6.08322024e-02 9.83508468e-01 -2.57353336e-02 5.92669666e-01 -2.92799890e-01 1.01940608e+00 -2.56228954e-01 3.69301617e-01 7.47774661e-01 -5.90283453e-01 6.73465133e-01 4.99180585e-01 -2.12059796e-01 -1.17636192e+00 -5.94195962e-01 -2.92464852e-01 1.09865713e+00 -3.86262476e-01 -9.67867315e-01 -1.19764853e+00 -6.26023471e-01 -2.87947655e-01 9.97481346e-01 -4.60754037e-01 1.08185023e-01 -5.89496136e-01 -3.24904829e-01 1.47321808e+00 2.42825553e-01 1.84329242e-01 -1.17290497e+00 -8.03966641e-01 1.40665233e-01 -4.63049948e-01 -1.60234153e+00 -1.19451985e-01 -2.60087490e-01 -2.05837682e-01 -9.32492197e-01 -3.82885486e-01 -6.00402176e-01 -5.34589775e-02 1.34965822e-01 1.08040929e+00 4.73511778e-02 -3.25491548e-01 4.11529124e-01 -8.69046867e-01 -7.20465004e-01 -1.46991372e+00 -1.03964999e-01 1.67591631e-01 -6.78403527e-02 4.83119875e-01 -2.54195720e-01 2.19896615e-01 7.53404796e-01 -8.00727189e-01 -4.49507415e-01 1.97731197e-01 2.89091617e-01 -7.41703585e-02 -3.10790867e-01 7.53971338e-01 -5.26483834e-01 1.20179093e+00 -3.84351790e-01 -6.83324873e-01 4.92291421e-01 -8.17094803e-01 -3.00330222e-01 3.16234946e-01 -3.04455429e-01 -1.09138846e+00 -5.13161644e-02 -2.98861682e-01 -5.57885505e-02 -5.55849552e-01 1.76618323e-01 -2.61792570e-01 5.91072068e-02 1.12242246e+00 5.71276963e-01 -9.36799720e-02 -4.83341515e-01 2.63869315e-01 1.41253781e+00 3.47577780e-01 -5.21554291e-01 5.98582685e-01 3.23638558e-01 -6.19982600e-01 -1.37172306e+00 -3.56265873e-01 -4.65567917e-01 -4.57793981e-01 -5.09122014e-01 3.91237855e-01 -7.11118877e-01 -2.08785564e-01 3.68310869e-01 -1.23348737e+00 4.98435460e-02 -6.18339360e-01 1.54573396e-01 -8.17419887e-01 7.34984636e-01 -1.01438165e-01 -1.26846635e+00 -3.80043566e-01 -1.37862027e+00 1.13352382e+00 -4.82786149e-01 -8.72857392e-01 -5.11630893e-01 2.42620766e-01 7.24670529e-01 4.32315499e-01 6.87184185e-02 3.55478436e-01 -8.65832090e-01 1.09070770e-01 -3.67496282e-01 1.23632155e-01 6.00339651e-01 2.15946227e-01 6.05148435e-01 -1.19938922e+00 1.57923736e-02 3.88670377e-02 -1.95356369e-01 2.14128047e-01 -1.68247819e-01 4.58985984e-01 -4.42811012e-01 -1.02527827e-01 -1.67263344e-01 7.65695632e-01 2.22373441e-01 5.87200642e-01 4.13096428e-01 8.51629823e-02 9.40801919e-01 8.39181602e-01 8.51330280e-01 1.23375375e-02 8.63904238e-01 4.96699065e-01 4.68740195e-01 -7.24113941e-01 -1.18558973e-01 5.76615930e-01 9.06640947e-01 1.47080004e-01 -7.56122112e-01 -1.00615907e+00 8.67491484e-01 -1.87885189e+00 -1.07161164e+00 -3.25342745e-01 1.94325244e+00 5.29145002e-01 -4.52601649e-02 5.29742301e-01 7.85187006e-01 9.94205356e-01 4.79149759e-01 1.34134218e-01 -8.64192069e-01 -4.34058040e-01 -4.41798419e-02 2.88838208e-01 1.69624239e-01 -1.03506088e+00 8.14655662e-01 8.20928574e+00 9.05484796e-01 -8.05618584e-01 3.17365766e-01 -3.44749689e-01 7.39053870e-03 -3.08591902e-01 -4.41208705e-02 -7.06190586e-01 4.48547810e-01 1.46256006e+00 -3.53985757e-01 2.80510843e-01 7.14434743e-01 5.15725315e-01 3.99471343e-01 -8.84955466e-01 6.57359183e-01 4.45412338e-01 -1.34271932e+00 -2.15717908e-02 -1.55557886e-01 4.35120225e-01 2.27134287e-01 -9.22612026e-02 6.62120730e-02 1.84982926e-01 -6.11946166e-01 1.47855198e+00 1.11488976e-01 7.59086609e-01 -4.72646773e-01 5.97786546e-01 4.74492460e-01 -9.90187883e-01 -2.31330901e-01 -1.17781356e-01 -1.51166007e-01 1.05662622e-01 3.40656072e-01 -7.19483495e-01 7.30229676e-01 6.92963600e-01 3.61510277e-01 -2.82149553e-01 6.65155828e-01 -3.00696224e-01 7.01518476e-01 -1.90311119e-01 -4.37795728e-01 1.79653941e-03 2.12293357e-01 1.06495893e+00 1.69728136e+00 3.61410528e-01 -4.24995482e-01 7.18259811e-02 9.81560946e-01 2.54409224e-01 4.26380545e-01 -7.71107256e-01 -4.22418118e-01 8.34329188e-01 9.41816688e-01 -3.55556279e-01 -2.83931851e-01 -6.54502630e-01 4.00825530e-01 -2.93441981e-01 -9.75548774e-02 -6.60251975e-01 -3.69864166e-01 8.23334038e-01 1.79120064e-01 1.65141612e-01 -8.06333274e-02 -1.56746268e-01 -9.09874678e-01 2.77471215e-01 -1.45351970e+00 3.69720519e-01 -5.69956183e-01 -6.37980878e-01 6.81129634e-01 2.37995058e-01 -1.52207482e+00 -6.27949357e-01 -2.69191831e-01 -3.51166636e-01 5.76283216e-01 -8.29598963e-01 -1.26656771e+00 -1.25900179e-01 2.10149214e-01 9.40547287e-01 -5.04143655e-01 1.28362882e+00 5.07996321e-01 -5.95545053e-01 5.66791713e-01 -1.51750579e-01 -1.70266092e-01 7.36987710e-01 -9.42860484e-01 9.79745209e-01 1.07492387e+00 1.56090707e-01 5.05098999e-01 1.21402752e+00 -6.98389769e-01 -1.35753202e+00 -8.21901441e-01 1.06933820e+00 -5.68271637e-01 1.01266551e+00 -8.47366333e-01 -3.90298992e-01 4.63263899e-01 4.19846684e-01 -8.45475644e-02 6.36404037e-01 -2.00873181e-01 -3.49309891e-01 3.12873162e-02 -1.25203824e+00 2.87955195e-01 1.24181461e+00 -6.53649330e-01 -6.15086854e-01 5.86352944e-01 8.69722664e-01 -4.79330629e-01 -8.63341570e-01 5.62303782e-01 6.07452691e-01 -9.42892611e-01 6.49741828e-01 -4.09857213e-01 -2.12790236e-01 -4.55644011e-01 -3.96334767e-01 -1.02833438e+00 2.07765967e-01 -1.31260455e+00 -2.00964376e-01 1.51721048e+00 7.68757999e-01 -3.15272123e-01 5.55047333e-01 5.37004992e-02 -7.14350879e-01 -1.59753308e-01 -1.26624703e+00 -1.17918265e+00 -3.42302471e-01 -1.05379570e+00 7.83356309e-01 9.58438277e-01 4.61376101e-01 1.04882024e-01 -6.02385640e-01 1.76187173e-01 3.43047142e-01 -2.18922049e-01 9.93763149e-01 -1.11385751e+00 1.02368854e-01 -4.76826757e-01 -7.99853861e-01 -5.32616794e-01 1.23398550e-01 -4.17434096e-01 -9.96309370e-02 -1.30314481e+00 -4.16737914e-01 2.65718907e-01 3.26213002e-01 3.44457626e-01 6.20746732e-01 1.86890930e-01 -2.72880122e-02 -1.07372068e-01 -4.03620452e-01 1.91970915e-02 8.29710960e-01 -2.99880743e-01 -4.74125473e-03 2.89278477e-01 -7.37844169e-01 5.09032369e-01 6.91772938e-01 -4.51046199e-01 -6.77280962e-01 -1.16518274e-01 2.99579889e-01 -6.68532029e-02 3.13860457e-03 -1.26446915e+00 3.37142617e-01 -1.79918855e-01 -6.12770498e-01 -7.47749865e-01 3.45736414e-01 -6.12612486e-01 3.00700992e-01 2.59182006e-01 -3.03495824e-01 -1.64260954e-01 8.76923129e-02 2.21369103e-01 -1.77147806e-01 -6.13456905e-01 5.50784349e-01 -1.32079154e-01 -8.52237105e-01 -3.86467636e-01 -9.93399680e-01 2.23082289e-01 1.38009882e+00 -5.84134340e-01 -5.41503668e-01 -4.52356428e-01 -7.41580009e-01 -2.60946810e-01 3.01931590e-01 1.16523671e+00 6.08293653e-01 -7.70773411e-01 -1.03251314e+00 3.49766910e-01 8.14836740e-01 -9.21570659e-01 -1.91525340e-01 3.82843912e-01 -3.62068981e-01 5.75922668e-01 -3.13592590e-02 -3.39746892e-01 -1.48978329e+00 5.82041025e-01 2.65499353e-01 3.36405665e-01 -4.38719511e-01 5.07713139e-01 -5.53858876e-01 -6.23307288e-01 4.99751598e-01 -1.17748953e-01 -3.94534171e-01 1.65248141e-01 6.73676014e-01 7.40880430e-01 6.91226006e-01 -1.19042027e+00 -5.94866097e-01 4.67875935e-02 2.17594668e-01 -3.88088226e-01 8.21176052e-01 -2.24632457e-01 -1.70849383e-01 5.96982956e-01 7.55826235e-01 4.46616918e-01 -4.26480800e-01 -1.50536895e-02 2.67476827e-01 -4.26908135e-01 -1.08338833e-01 -1.02274203e+00 -6.91763759e-01 2.30028331e-01 7.10577250e-01 8.05375993e-01 7.17870951e-01 2.52147883e-01 4.75436926e-01 2.85052031e-01 3.26388448e-01 -1.49363458e+00 -8.13072696e-02 5.15214801e-01 1.29283798e+00 -7.61394858e-01 -2.92773582e-02 -9.53540564e-01 -3.37365746e-01 1.21478951e+00 3.58250946e-01 4.09644186e-01 3.69797945e-01 8.17842484e-01 4.59761322e-01 -1.79692179e-01 -8.18331122e-01 -4.42515701e-01 -5.03879786e-02 1.18839276e+00 6.59572423e-01 2.65467554e-01 -6.89892888e-01 3.80723804e-01 -5.51437497e-01 -2.73880184e-01 9.15450335e-01 1.05430210e+00 -5.26152074e-01 -1.62619984e+00 -5.30678868e-01 2.07431510e-01 -5.32386124e-01 1.45413101e-01 -1.27167952e+00 8.50699425e-01 3.17023933e-01 1.72793794e+00 -2.94154376e-01 -6.00500762e-01 9.87056494e-01 1.19133048e-01 2.06471965e-01 -8.80226374e-01 -7.76861370e-01 -2.09647506e-01 1.08203804e+00 -4.88573372e-01 -5.33983469e-01 -9.53709900e-01 -7.08618641e-01 -2.64477879e-01 -6.74224854e-01 5.30360006e-02 1.02017248e+00 7.42055357e-01 2.76971936e-01 5.29544592e-01 4.39676285e-01 -1.80533841e-01 -7.36151397e-01 -1.14114976e+00 -3.01357925e-01 6.36896119e-02 2.34967873e-01 -4.88783598e-01 -4.37678605e-01 -3.40110436e-02]
[14.38897705078125, 6.90024471282959]
44a459cd-ac4d-47d5-9869-0271a83fb2aa
unsupervised-pre-traing-for-sequence-to
1910.12418
null
https://arxiv.org/abs/1910.12418v2
https://arxiv.org/pdf/1910.12418v2.pdf
Unsupervised pre-training for sequence to sequence speech recognition
This paper proposes a novel approach to pre-train encoder-decoder sequence-to-sequence (seq2seq) model with unpaired speech and transcripts respectively. Our pre-training method is divided into two stages, named acoustic pre-trianing and linguistic pre-training. In the acoustic pre-training stage, we use a large amount of speech to pre-train the encoder by predicting masked speech feature chunks with its context. In the linguistic pre-training stage, we generate synthesized speech from a large number of transcripts using a single-speaker text to speech (TTS) system, and use the synthesized paired data to pre-train decoder. This two-stage pre-training method integrates rich acoustic and linguistic knowledge into seq2seq model, which will benefit downstream automatic speech recognition (ASR) tasks. The unsupervised pre-training is finished on AISHELL-2 dataset and we apply the pre-trained model to multiple paired data ratios of AISHELL-1 and HKUST. We obtain relative character error rate reduction (CERR) from 38.24% to 7.88% on AISHELL-1 and from 12.00% to 1.20% on HKUST. Besides, we apply our pretrained model to a cross-lingual case with CALLHOME dataset. For all six languages in CALLHOME dataset, our pre-training method makes model outperform baseline consistently.
['Zhiyun Fan', 'Bo Xu', 'Shiyu Zhou']
2019-10-28
null
null
null
null
['sequence-to-sequence-speech-recognition']
['speech']
[ 6.97141171e-01 3.08591902e-01 3.04710388e-01 -6.64375901e-01 -1.54028177e+00 -4.74929422e-01 2.11603865e-01 -2.90406376e-01 -6.10177517e-01 6.92329109e-01 5.25210261e-01 -5.82150042e-01 7.46039987e-01 -4.34221417e-01 -8.67970049e-01 -4.86330301e-01 1.91806197e-01 4.16401058e-01 1.40949279e-01 -2.00145304e-01 -1.76188469e-01 -9.59338844e-02 -1.12441099e+00 7.68101633e-01 9.31401908e-01 7.01898575e-01 7.23041773e-01 1.08721316e+00 -2.02237919e-01 6.11015618e-01 -6.98561311e-01 -4.17050332e-01 1.15435980e-01 -8.54293525e-01 -6.40863419e-01 -6.32027090e-02 1.84573364e-02 -2.94074088e-01 -2.17173383e-01 9.29414570e-01 7.52480507e-01 6.50749877e-02 3.63661975e-01 -5.50364554e-01 -5.75670004e-01 1.20012069e+00 -3.77171397e-01 5.37695624e-02 2.24345982e-01 2.86523461e-01 9.03632164e-01 -1.10342729e+00 2.46141791e-01 1.30963004e+00 4.94783252e-01 8.95452559e-01 -9.78644252e-01 -7.94651687e-01 -5.67468926e-02 -7.68057108e-02 -1.41684401e+00 -1.17805600e+00 3.43807697e-01 -3.26463968e-01 1.33139753e+00 2.58599252e-01 1.40404582e-01 1.14946377e+00 -1.64528713e-01 7.47231066e-01 8.95542800e-01 -5.89235783e-01 2.06456915e-01 -3.78428102e-02 -2.25882590e-01 3.71037126e-01 -6.09930873e-01 2.91733176e-01 -7.05901802e-01 2.90651977e-01 4.42289621e-01 -5.34179151e-01 -2.17138827e-01 8.54348958e-01 -1.20135057e+00 6.11714721e-01 -1.21758832e-03 1.35982618e-01 -2.54809558e-01 1.72622632e-02 7.04221964e-01 4.49934602e-01 4.22500730e-01 4.49055620e-02 -6.57969475e-01 -4.31776941e-01 -1.09504521e+00 -3.86085153e-01 6.16013467e-01 1.30042815e+00 6.95555985e-01 5.62467635e-01 -2.63014823e-01 1.39600348e+00 3.75828654e-01 8.16276968e-01 8.93415034e-01 -4.00673121e-01 1.03412974e+00 -2.01815084e-01 -5.00069082e-01 -8.44050273e-02 1.88302457e-01 -3.53070170e-01 -7.33543992e-01 -4.53266203e-01 -3.01322397e-02 -6.38840914e-01 -1.12029505e+00 1.68299758e+00 3.55532914e-02 2.82132983e-01 6.50171936e-01 7.53843486e-01 8.86432350e-01 1.30602717e+00 7.05697313e-02 -4.36779618e-01 1.19765019e+00 -1.35613561e+00 -6.48414612e-01 -4.51039732e-01 8.99405897e-01 -1.02171814e+00 1.29719853e+00 1.89983338e-01 -1.12666166e+00 -8.51063550e-01 -8.72595191e-01 -9.82341617e-02 9.09472480e-02 5.68009257e-01 -3.75854701e-01 4.78307664e-01 -1.05420363e+00 2.19388559e-01 -6.32718444e-01 -1.79447621e-01 7.72548392e-02 2.91000158e-01 -3.16696167e-01 -8.60775560e-02 -1.37810361e+00 5.91662765e-01 6.34312570e-01 1.70526996e-01 -1.21145058e+00 -6.07134342e-01 -9.32065785e-01 3.09152119e-02 1.83585927e-01 -1.55435175e-01 1.58288515e+00 -1.13305795e+00 -2.19444847e+00 7.01287448e-01 -5.02828896e-01 -4.94910866e-01 2.55770266e-01 -3.52673680e-01 -8.07785451e-01 -1.58826113e-01 -3.78066138e-03 5.95409691e-01 7.15393305e-01 -9.71478403e-01 -5.70308983e-01 9.52638239e-02 -7.41307497e-01 3.81638706e-01 8.98301229e-03 4.49640602e-01 -5.82342029e-01 -8.14706206e-01 -2.23364174e-01 -8.71574640e-01 -8.45446512e-02 -8.92575264e-01 -6.96706593e-01 -3.53509635e-02 6.11144483e-01 -1.12075043e+00 1.25953197e+00 -2.45401669e+00 -7.09856600e-02 -5.54086529e-02 -5.13884306e-01 7.01015770e-01 -5.19466102e-01 7.69318640e-01 -5.11390567e-02 5.97935133e-02 -5.99589169e-01 -7.57305980e-01 -2.03655168e-01 2.74889350e-01 -4.88615900e-01 4.61402647e-02 5.94001293e-01 8.43573749e-01 -8.14963818e-01 -4.22421932e-01 3.19952890e-02 3.84713292e-01 -5.04067242e-01 8.01216245e-01 -2.52733082e-01 6.06982827e-01 -1.95549410e-02 3.74391347e-01 5.29091656e-01 4.53007847e-01 2.61832535e-01 1.64331362e-01 -3.63182336e-01 1.06163383e+00 -7.39227295e-01 1.88549650e+00 -7.39715815e-01 5.14830649e-01 1.72390218e-03 -8.17231953e-01 1.07542944e+00 6.05666459e-01 -9.79051143e-02 -7.54112601e-01 8.73803906e-03 4.80149090e-01 2.21438810e-01 -6.32615089e-01 4.45953757e-01 -5.39828658e-01 -1.61344856e-01 2.76407629e-01 2.61813641e-01 -3.11660469e-02 -1.05615817e-01 -4.99222353e-02 1.13716829e+00 1.61786735e-01 8.65473822e-02 2.49165557e-02 7.10202038e-01 -2.35282332e-01 6.69121265e-01 3.73765439e-01 9.21236277e-02 9.51240838e-01 1.14697814e-01 2.75973290e-01 -1.20211852e+00 -9.70598936e-01 7.86185563e-02 1.33692682e+00 -5.24869561e-01 -4.18239444e-01 -8.87543023e-01 -6.91965580e-01 -2.64089376e-01 9.20568168e-01 -1.45892143e-01 -6.39941841e-02 -8.25943112e-01 -3.80366623e-01 1.16417754e+00 4.39157963e-01 5.15902042e-01 -1.31548393e+00 2.68333495e-01 5.12023628e-01 -1.01779066e-01 -1.34772563e+00 -9.07173634e-01 4.00113195e-01 -4.71627921e-01 -3.27560574e-01 -7.10128963e-01 -1.14615309e+00 4.14729923e-01 -1.17118180e-01 8.56297493e-01 -9.70998183e-02 3.18912297e-01 -4.11538929e-01 -6.46548927e-01 -1.36553124e-01 -1.12461782e+00 3.47951829e-01 1.71792090e-01 2.20946118e-01 1.97283745e-01 -4.02918130e-01 -1.88834921e-01 2.35943571e-01 -5.77327013e-01 2.97156215e-01 9.52131629e-01 8.02851021e-01 7.18142688e-01 -3.49655628e-01 9.73285556e-01 -9.05042470e-01 2.33184099e-01 -5.18723309e-01 -5.24613976e-01 2.87983567e-01 -1.28759146e-01 2.97215730e-02 1.15875614e+00 -4.09907579e-01 -1.33278453e+00 3.89496535e-01 -9.80009794e-01 -3.20110261e-01 -2.63273209e-01 5.90130746e-01 -6.44402087e-01 6.49443626e-01 3.87601286e-01 6.82412922e-01 -1.62908956e-01 -7.99369276e-01 3.82591575e-01 1.48348868e+00 9.25583303e-01 -3.37004036e-01 6.53682172e-01 -3.69588703e-01 -8.02811980e-01 -1.08340526e+00 -6.62309825e-01 -3.82692367e-01 -6.41966164e-01 1.53203219e-01 9.18754160e-01 -1.17869484e+00 -3.06374669e-01 5.01512170e-01 -1.29178560e+00 -7.83681035e-01 -1.37635380e-01 6.20556831e-01 -4.19985086e-01 3.11773956e-01 -8.22688520e-01 -9.04901803e-01 -6.39255524e-01 -1.22878349e+00 1.13584459e+00 -2.20553041e-01 -4.70010564e-02 -5.53143680e-01 2.27601036e-01 4.24281925e-01 3.56295675e-01 -4.95241284e-01 5.75732768e-01 -1.02498877e+00 -2.24197820e-01 2.28367567e-01 -4.73557338e-02 9.54735458e-01 1.51981577e-01 -6.94479942e-02 -1.25287867e+00 -2.82184005e-01 -1.52079344e-01 -4.25921917e-01 7.23887265e-01 1.20104477e-02 9.82005179e-01 -4.82693642e-01 3.52218300e-02 6.80322886e-01 1.12066197e+00 5.34488440e-01 7.33701408e-01 -2.66657144e-01 7.74636745e-01 6.51141584e-01 5.15814006e-01 1.16950035e-01 4.15656149e-01 5.25600970e-01 -1.22825429e-01 8.42928886e-02 -4.36523050e-01 -6.74826264e-01 1.14137852e+00 2.05110288e+00 5.23751378e-01 -4.51930761e-01 -9.43938315e-01 5.38075387e-01 -1.53537583e+00 -7.18657553e-01 -1.52779952e-01 2.07491684e+00 1.39078581e+00 1.49030518e-02 8.35112482e-02 -4.63785492e-02 9.15211141e-01 -9.57079232e-02 -3.22386503e-01 -5.60578465e-01 -1.08664870e-01 2.33972609e-01 3.20844591e-01 8.59127104e-01 -7.90808737e-01 1.43400586e+00 5.67303419e+00 1.04668617e+00 -1.36517966e+00 1.93594500e-01 6.60510123e-01 -1.43852271e-02 -4.49552655e-01 1.25489801e-01 -1.12179482e+00 9.31108773e-01 1.91075313e+00 2.79718987e-03 4.75076467e-01 5.70174932e-01 3.87794256e-01 1.91866741e-01 -1.13657558e+00 8.86103511e-01 9.69719514e-02 -1.04483318e+00 2.88615306e-03 -2.35761210e-01 5.52264333e-01 5.02397716e-01 -1.86250612e-01 6.67507350e-01 5.00361860e-01 -1.16751015e+00 8.54172587e-01 1.35330141e-01 1.45190132e+00 -7.74995029e-01 8.12980235e-01 5.34410179e-01 -1.19016206e+00 2.01444134e-01 -3.90312314e-01 1.65486395e-01 6.29490197e-01 6.75465524e-01 -1.29581213e+00 6.40259743e-01 3.62435818e-01 5.53672493e-01 -3.96893881e-02 7.42228925e-01 -4.08271551e-01 1.38390982e+00 -3.11964303e-01 4.84389029e-02 2.00143889e-01 -9.41115469e-02 2.60744542e-01 1.77533019e+00 4.90091264e-01 1.61107287e-01 -4.48000692e-02 4.29137915e-01 -2.86394000e-01 3.03001463e-01 -2.51370043e-01 -2.43014365e-01 7.64664233e-01 8.34943414e-01 -9.72478166e-02 -6.08052373e-01 -5.52071750e-01 1.30331588e+00 5.15817583e-01 3.28618526e-01 -8.17238092e-01 -7.43696451e-01 5.72941601e-01 -1.68842748e-01 3.65526915e-01 -2.54458606e-01 -2.06269156e-02 -1.07991481e+00 -9.67089757e-02 -1.18227065e+00 -6.60834908e-02 -7.76255369e-01 -1.16892314e+00 1.16626072e+00 -4.15700436e-01 -9.80823219e-01 -5.77380836e-01 -3.39534372e-01 -7.08388388e-01 1.30666173e+00 -1.57575142e+00 -1.19886351e+00 1.00965649e-01 3.05496663e-01 1.15646470e+00 -4.17794913e-01 8.22756469e-01 5.99566996e-01 -9.18531597e-01 8.55360329e-01 2.04795182e-01 3.84051800e-01 7.66934276e-01 -1.04444325e+00 9.48875964e-01 1.08104455e+00 1.12195134e-01 4.58071887e-01 2.70956039e-01 -7.37625301e-01 -1.23700714e+00 -1.47399640e+00 1.14518309e+00 -1.48642734e-01 5.27929246e-01 -7.70759583e-01 -1.01067352e+00 7.73341894e-01 4.24542725e-01 -1.94844276e-01 8.96666050e-01 -2.01761186e-01 -1.35738999e-01 -9.80623662e-02 -6.50918663e-01 4.76021200e-01 9.81059074e-01 -9.94854212e-01 -6.18111014e-01 -8.35556015e-02 1.33007348e+00 -4.62331116e-01 -7.60836661e-01 2.58565098e-01 2.98431426e-01 -4.79861289e-01 4.00469631e-01 -4.23543334e-01 3.89087409e-01 -3.59951138e-01 -5.41111112e-01 -1.71408260e+00 -2.96387188e-02 -1.16717243e+00 3.32480043e-01 1.82131040e+00 9.80370641e-01 -3.83642673e-01 3.79611880e-01 -1.46757007e-01 -9.99604166e-01 -5.52214265e-01 -8.88213992e-01 -8.74035060e-01 2.27786943e-01 -7.61376679e-01 7.29171634e-01 6.50491834e-01 7.51804700e-03 6.43090844e-01 -4.79656994e-01 2.89236873e-01 1.28614724e-01 -3.72126281e-01 7.50937462e-01 -4.01929140e-01 -5.91549575e-01 1.09805882e-01 1.59314036e-01 -1.47994232e+00 1.97797865e-01 -1.16050243e+00 7.85274565e-01 -1.17486823e+00 -1.29225105e-01 -5.01563251e-01 -2.28252888e-01 6.04893863e-01 -3.13121080e-01 5.76232634e-02 4.19041775e-02 8.33110884e-02 -2.82950163e-01 9.22874451e-01 8.72343123e-01 2.06821248e-01 -2.79107541e-01 -1.12587042e-01 -4.23177838e-01 2.11197957e-01 8.91726553e-01 -6.45950615e-01 -3.79421473e-01 -8.20492566e-01 -1.77260071e-01 2.60486484e-01 -2.55417168e-01 -9.60130394e-01 6.77181482e-02 -1.82083901e-02 -4.87471074e-02 -5.97442448e-01 2.55533576e-01 -4.49546218e-01 -5.05494177e-02 2.41808504e-01 -4.89380419e-01 1.27840322e-02 3.57616544e-01 2.85351992e-01 -3.88795465e-01 -3.08466494e-01 6.64752960e-01 5.58329448e-02 -5.54085374e-01 8.26881006e-02 -6.04498744e-01 4.18259919e-01 6.47814512e-01 -7.23229116e-03 -1.54056281e-01 -2.99279600e-01 -6.20984733e-01 2.10170716e-01 5.59017844e-02 4.38346088e-01 5.59517801e-01 -1.21776927e+00 -1.09436941e+00 7.56326973e-01 8.25114548e-02 1.19164906e-01 2.44383708e-01 5.98682523e-01 -4.32441324e-01 3.73889148e-01 2.99985766e-01 -5.49038291e-01 -1.21979713e+00 3.35121095e-01 1.35875806e-01 1.67138323e-01 -3.33700836e-01 1.21274674e+00 2.32553542e-01 -7.31694758e-01 1.99625567e-01 -4.08153802e-01 2.68823236e-01 -3.24644208e-01 4.73165452e-01 1.26814231e-01 1.40882254e-01 -1.04298794e+00 -4.97277737e-01 2.73195744e-01 -6.19513914e-02 -6.52623534e-01 1.30248320e+00 -2.95557886e-01 1.61469311e-01 5.10230243e-01 1.60569060e+00 4.49965566e-01 -1.21229792e+00 -2.10265025e-01 2.50695851e-02 3.53498049e-02 -1.26736447e-01 -9.04084325e-01 -8.67175937e-01 1.14494157e+00 6.89355135e-02 -1.66676164e-01 9.86511886e-01 -1.25947027e-02 1.31024361e+00 2.27461845e-01 -8.01293924e-02 -1.12919223e+00 -8.41700286e-02 1.09536195e+00 9.08666313e-01 -1.13928664e+00 -7.76501000e-01 -3.63534212e-01 -1.06625783e+00 9.01772320e-01 6.36705339e-01 5.57678342e-02 4.67543691e-01 6.52142644e-01 4.51614022e-01 4.63449597e-01 -1.06387079e+00 -2.78746367e-01 1.94409892e-01 4.63194549e-01 7.26917922e-01 1.23117954e-01 -4.06192355e-02 8.38794768e-01 -8.30686092e-01 -1.44749403e-01 3.31624925e-01 5.64376950e-01 -5.33512115e-01 -1.22006798e+00 -1.33773208e-01 1.64874628e-01 -5.21642447e-01 -7.84314036e-01 -2.08261490e-01 1.92587838e-01 3.21795046e-02 1.18757629e+00 2.25814149e-01 -7.40632236e-01 3.42360914e-01 3.60770524e-01 -1.30385533e-01 -1.10662854e+00 -7.52186775e-01 5.89238644e-01 6.08651578e-01 -2.64954150e-01 2.00789854e-01 -5.00834882e-01 -1.53841031e+00 -3.00930701e-02 -4.33853149e-01 4.03661609e-01 6.38175786e-01 9.41819549e-01 4.52266127e-01 6.99122012e-01 8.82317543e-01 -4.40067202e-01 -5.75945377e-01 -1.28845561e+00 -2.84906060e-01 3.58509570e-02 5.35432518e-01 1.51117399e-01 -3.30347329e-01 3.82397026e-01]
[14.481453895568848, 6.948907852172852]
6a6bf3fa-8819-4094-ba94-774277e3e926
multi-view-semantic-labeling-of-3d-point
1805.03994
null
http://arxiv.org/abs/1805.03994v2
http://arxiv.org/pdf/1805.03994v2.pdf
Multi-View Semantic Labeling of 3D Point Clouds for Automated Plant Phenotyping
Semantic labeling of 3D point clouds is important for the derivation of 3D models from real world scenarios in several economic fields such as building industry, facility management, town planning or heritage conservation. In contrast to these most common applications, we describe in this study the semantic labeling of 3D point clouds derived from plant organs by high-precision scanning. Our approach is optimized for the task of plant phenotyping with its very specific challenges and is employing a deep learning framework. Thereby, we report important experiences concerning detailed parameter initialization and optimization techniques. By evaluating our approach with challenging datasets we achieve state-of-the-art results without difficult and time consuming feature engineering as being necessary in traditional approaches to semantic labeling.
['Volker Steinhage', 'Reinhard Töpfer', 'Katja Herzog', 'Jennifer Mack', 'Bernhard Japes', 'Florian Rist']
2018-05-10
null
null
null
null
['plant-phenotyping']
['computer-vision']
[ 2.02218562e-01 7.32492730e-02 -4.90357913e-02 -3.72016937e-01 -1.92927465e-01 -7.56664634e-01 4.96954739e-01 5.45285225e-01 3.14351246e-02 3.32629979e-01 -6.61011279e-01 -6.34313047e-01 -4.27460402e-01 -1.13425875e+00 -5.82151294e-01 -3.00466120e-01 -1.43871516e-01 1.11071825e+00 3.25119972e-01 -1.31600544e-01 2.68180221e-01 1.58700061e+00 -1.70932865e+00 1.39023438e-02 3.59360725e-01 1.16315866e+00 4.96808350e-01 3.79375428e-01 -7.69088209e-01 3.28418333e-03 -3.89807969e-01 2.79754791e-02 3.24862748e-01 1.83788806e-01 -9.89516914e-01 6.17477834e-01 2.89967120e-01 8.00507963e-02 4.98071730e-01 8.99006009e-01 2.96692520e-01 -1.28179997e-01 5.64792037e-01 -1.35985792e+00 -4.96284425e-01 2.20470518e-01 -5.02741277e-01 -5.02051532e-01 6.52337745e-02 -1.18729211e-01 6.22432947e-01 -7.44571924e-01 5.74066758e-01 1.05080605e+00 8.96026313e-01 1.67983443e-01 -1.42215288e+00 -1.19684897e-01 3.06780249e-01 -1.01858974e-01 -1.52527726e+00 -1.32370340e-02 7.16386497e-01 -5.33796608e-01 8.71010900e-01 2.62772769e-01 8.04973841e-01 4.46504593e-01 -3.11673880e-01 5.78813016e-01 8.19511533e-01 -5.30645728e-01 4.32048798e-01 -7.57116601e-02 -5.91218658e-02 6.13563597e-01 1.35270983e-01 -6.64907992e-02 -2.32711714e-02 1.38054313e-02 1.04159141e+00 -1.73391357e-01 4.00535017e-02 -1.13049567e+00 -1.06482720e+00 6.39998734e-01 6.86638772e-01 4.56076950e-01 -3.54995459e-01 3.23986351e-01 1.15985110e-01 -6.86976463e-02 4.80198473e-01 3.93533140e-01 -1.07901645e+00 2.38313049e-01 -9.31247711e-01 2.57829607e-01 7.05293000e-01 1.36295068e+00 7.60166705e-01 2.65373420e-02 2.88245410e-01 5.83349109e-01 4.09770668e-01 3.68686348e-01 -2.83905923e-01 -1.11672282e+00 -2.72942722e-01 9.37011600e-01 2.92444736e-01 -6.93063855e-01 -7.55576134e-01 -4.45932567e-01 -6.15786672e-01 5.64799011e-01 2.97975719e-01 3.93095315e-01 -1.12443089e+00 1.12885416e+00 5.31894147e-01 1.49830341e-01 -2.63719469e-01 4.69933569e-01 1.03165030e+00 2.43077308e-01 3.85375500e-01 2.01784074e-01 1.35205102e+00 -3.48409146e-01 -2.22216606e-01 -2.36570537e-02 8.88147235e-01 -8.50862920e-01 9.52183008e-01 2.65197963e-01 -8.65093887e-01 -4.39985484e-01 -6.55162513e-01 -2.63430960e-02 -8.69597018e-01 2.12043583e-01 1.10073245e+00 8.58831942e-01 -1.02516103e+00 9.12438869e-01 -7.28793383e-01 -8.53498995e-01 8.93414021e-01 4.37184870e-01 -5.03990114e-01 2.16492012e-01 -4.22751248e-01 1.07777739e+00 6.56671882e-01 5.56714423e-02 -8.10305715e-01 -9.71699774e-01 -7.28251636e-01 1.54459924e-01 3.98475856e-01 -7.13101089e-01 1.26720357e+00 -1.89876243e-01 -1.71160722e+00 1.48563957e+00 8.63442421e-02 -2.05300286e-01 3.35678816e-01 -1.45509601e-01 -9.20886025e-02 -2.91937143e-01 -7.48810917e-02 8.79430890e-01 4.37801003e-01 -1.57197273e+00 -4.13518161e-01 -5.35310209e-01 7.27312714e-02 -7.92049840e-02 1.43153325e-01 -6.03367314e-02 -2.04316616e-01 -8.36753920e-02 7.18286753e-01 -1.01889908e+00 -4.34219658e-01 6.16415977e-01 -4.40231472e-01 1.12129010e-01 1.25687861e+00 -1.53523296e-01 1.52147755e-01 -1.93189418e+00 3.51486616e-02 2.31782347e-01 2.68686041e-02 3.99752706e-01 -1.46426307e-02 1.58128932e-01 -3.08547914e-01 3.77713025e-01 -4.77330893e-01 -2.42954358e-01 1.60879955e-01 3.53026956e-01 -1.79824889e-01 4.42617297e-01 4.00023788e-01 8.63916039e-01 -7.29634404e-01 -2.55830050e-01 1.03999662e+00 5.00503838e-01 -2.18784645e-01 -1.03422385e-02 -6.06965184e-01 6.02718890e-01 -3.96148354e-01 1.04119718e+00 9.59485888e-01 -4.15206641e-01 -6.53544664e-02 -2.07276270e-01 -3.27616990e-01 -2.35034036e-03 -1.17256594e+00 2.04201365e+00 -6.06596529e-01 2.60994494e-01 2.37463951e-01 -1.19239533e+00 1.17234147e+00 1.27395183e-01 9.28786278e-01 -9.08190981e-02 3.00388217e-01 3.08815122e-01 -5.38113713e-01 9.47850347e-02 3.30801010e-01 -1.24305740e-01 3.35190631e-02 4.22843620e-02 2.33338103e-01 -1.06782949e+00 -2.51026541e-01 -2.13623449e-01 6.09588861e-01 7.86539137e-01 4.93263096e-01 -5.80281794e-01 6.10057175e-01 4.54391599e-01 1.24857098e-01 2.64958262e-01 -3.90531309e-02 6.46657825e-01 4.11006302e-01 -6.22263253e-01 -1.11464489e+00 -1.07649004e+00 -4.73628908e-01 5.93917310e-01 1.05474420e-01 -2.74241269e-01 -5.32878160e-01 -6.10854805e-01 3.27295661e-01 9.60017264e-01 -4.05801564e-01 1.58746913e-01 -3.15235674e-01 -8.07503104e-01 1.80637583e-01 4.11256522e-01 3.62175316e-01 -9.55121636e-01 -8.34127903e-01 1.74298480e-01 4.24544901e-01 -1.43057704e+00 5.42504787e-01 5.92028677e-01 -1.03324866e+00 -1.03157210e+00 -4.76870120e-01 -7.62522161e-01 5.31080544e-01 3.42690647e-01 1.50924253e+00 1.57070532e-01 -5.39934695e-01 1.94082558e-01 -4.41212624e-01 -8.70091915e-01 -3.38752866e-01 4.38395321e-01 -4.15297031e-01 -6.14880741e-01 4.73089725e-01 -7.50035942e-01 -1.07394569e-01 2.78121650e-01 -7.35396087e-01 3.95035185e-02 2.25616664e-01 3.61258894e-01 1.08900845e+00 1.21989027e-01 5.80408126e-02 -9.09914911e-01 3.24580185e-02 -1.60304591e-01 -1.31536686e+00 1.56229034e-01 -2.70663023e-01 -2.07818344e-01 3.52472246e-01 1.29365072e-01 -5.92413545e-01 6.77172363e-01 -4.16203320e-01 -1.62013009e-01 -1.07718706e+00 1.65418625e-01 -5.19078732e-01 -5.78933179e-01 5.79466522e-01 -1.57489389e-01 -2.24667534e-01 -7.62893438e-01 4.96576905e-01 3.32389951e-01 4.12173778e-01 -6.11734509e-01 1.01426768e+00 7.73749828e-01 7.77309835e-01 -1.22236383e+00 -8.52738917e-01 -3.76711398e-01 -1.29170489e+00 -1.30128413e-01 8.23744357e-01 -5.14903247e-01 -9.39056814e-01 4.73632216e-01 -1.41835308e+00 -1.79016814e-01 -7.24336445e-01 1.57814324e-01 -9.32402730e-01 1.33261546e-01 -3.60715166e-02 -5.83089232e-01 7.77253062e-02 -1.10864580e+00 1.68712819e+00 3.94644961e-02 -9.52865779e-02 -1.17302811e+00 -2.18741938e-01 -5.97639233e-02 3.31020862e-01 7.02513933e-01 1.35330772e+00 -2.79662341e-01 -9.17914867e-01 -2.58227885e-01 -3.67879778e-01 1.28482178e-01 2.90532887e-01 2.69125521e-01 -1.25282931e+00 2.58886606e-01 -2.09423557e-01 -3.08143608e-02 2.11790532e-01 5.53997755e-01 1.62876475e+00 7.50999451e-01 -5.00647664e-01 8.72733355e-01 1.67661071e+00 -1.00598847e-02 4.56432700e-01 4.10720557e-01 8.32953632e-01 8.19902122e-01 6.55030966e-01 3.79315376e-01 -4.57700640e-02 7.32528985e-01 1.17801154e+00 -2.47977898e-01 -9.02398378e-02 -9.62222740e-02 -5.50526857e-01 1.78691149e-01 -8.19759220e-02 -1.83832735e-01 -1.24506426e+00 5.61707795e-01 -1.62307763e+00 -6.13878310e-01 -5.05521178e-01 2.08946705e+00 1.47013426e-01 -1.94809530e-02 -3.03882688e-01 3.53804231e-01 5.58812380e-01 -5.56686409e-02 -5.77751756e-01 -3.71716499e-01 -1.95183516e-01 4.99991477e-01 7.78429151e-01 3.60344946e-01 -1.28378749e+00 1.30751586e+00 6.80712318e+00 4.60778922e-01 -1.02631187e+00 -1.10965017e-02 2.90813863e-01 2.22913891e-01 -1.91004723e-01 2.80402869e-01 -8.83011937e-01 1.54015850e-02 5.58317184e-01 1.85147479e-01 2.02497289e-01 1.05952704e+00 1.74641445e-01 -4.97072861e-02 -1.06137955e+00 9.53998923e-01 -5.04528582e-01 -1.56219721e+00 -1.45456687e-01 7.64080659e-02 4.40702289e-01 3.83736938e-02 -2.52283335e-01 -1.53909937e-01 7.03311920e-01 -1.13147485e+00 6.73463464e-01 1.87715471e-01 7.59875119e-01 -6.66103780e-01 5.44064403e-01 4.27683115e-01 -1.18180192e+00 2.38812059e-01 -4.70007837e-01 1.15734592e-01 4.24274802e-01 1.15149856e+00 -9.99971211e-01 7.98074365e-01 7.35456586e-01 5.45355380e-01 -5.05464435e-01 1.13170600e+00 -2.19379619e-01 1.64358139e-01 -5.92716753e-01 3.16922069e-01 1.72884822e-01 -2.73637861e-01 3.85448903e-01 8.20615888e-01 6.50571644e-01 -2.23155200e-01 2.39754304e-01 1.07672608e+00 2.81668198e-03 4.99295555e-02 -8.41639102e-01 -7.37388479e-03 3.56693149e-01 1.29312479e+00 -1.31760502e+00 -1.73621885e-02 -3.21121812e-02 7.44956553e-01 7.70114139e-02 -1.95653856e-01 -6.80647075e-01 -1.30823523e-01 6.47462547e-01 3.60493273e-01 4.66993839e-01 -6.53083563e-01 -6.95009649e-01 -6.94383621e-01 -2.97232449e-01 -2.84624528e-02 -1.62179381e-01 -1.05242229e+00 -1.07240677e+00 3.56373876e-01 2.75759816e-01 -1.06147039e+00 -3.67541090e-02 -1.13217044e+00 -2.59920746e-01 1.01174891e+00 -1.76268888e+00 -1.64638722e+00 -6.67313874e-01 2.30470985e-01 3.08234155e-01 -5.39853498e-02 1.41901767e+00 1.15905523e-01 -1.05199225e-01 -2.83680767e-01 1.81700513e-02 -4.58692759e-01 1.28462404e-01 -1.25324333e+00 8.47771704e-01 5.95112264e-01 2.51864433e-01 -1.81679726e-02 4.67790663e-01 -4.53909695e-01 -9.86185789e-01 -1.22414041e+00 8.55875850e-01 -5.04024625e-01 4.44228142e-01 -3.39655995e-01 -7.65062153e-01 5.94549954e-01 -2.97786474e-01 2.46352956e-01 6.19455159e-01 9.79584903e-02 6.75620660e-02 1.93009079e-01 -1.44629872e+00 3.46633017e-01 1.57614791e+00 -3.12095165e-01 -7.94123039e-02 8.17748785e-01 6.65187955e-01 -5.04405439e-01 -1.04064894e+00 7.27029085e-01 3.01673800e-01 -8.54734540e-01 1.32049620e+00 -6.11387908e-01 3.89860682e-02 -6.13082111e-01 -3.93730193e-01 -1.30625546e+00 -6.94833755e-01 -2.21393272e-01 9.20551792e-02 1.13235307e+00 2.08136253e-03 -2.23669663e-01 1.20238698e+00 3.34994823e-01 -5.25653303e-01 -2.48312622e-01 -6.81192279e-01 -9.05683637e-01 1.40005305e-01 -5.56769431e-01 1.14143455e+00 1.04215884e+00 -9.28719819e-01 -4.65379916e-02 2.35778064e-01 5.58987021e-01 6.65007353e-01 5.44877291e-01 8.34256053e-01 -2.04640341e+00 1.81049541e-01 -5.50187945e-01 -7.54929662e-01 -6.37193799e-01 4.47090089e-01 -9.10440624e-01 -9.17258933e-02 -1.75568414e+00 -4.55966651e-01 -9.21576440e-01 1.09740540e-01 4.86868531e-01 5.00720382e-01 1.37787998e-01 2.39283264e-01 -5.02222516e-02 -1.06477970e-02 4.25584018e-01 1.11543334e+00 -1.17015705e-01 -8.39333981e-02 1.44109040e-01 -5.41292429e-01 1.11325860e+00 1.06001139e+00 -4.92418617e-01 -4.06669676e-01 -7.65922189e-01 9.35422480e-02 -4.57573950e-01 6.41074955e-01 -9.95279908e-01 -2.80608714e-01 -3.63072634e-01 3.96132737e-01 -1.05395377e+00 5.87919652e-01 -1.47395015e+00 4.56822753e-01 3.10143143e-01 2.34109461e-01 -1.98946148e-01 5.36027431e-01 -2.34139841e-02 1.10283554e-01 -6.50076270e-01 9.12618577e-01 -6.16113842e-01 -1.10809660e+00 3.85846317e-01 -5.02161533e-02 -4.93810207e-01 1.30901635e+00 -4.95582193e-01 -5.58025353e-02 1.21070117e-01 -1.05458117e+00 4.00378890e-02 9.46472406e-01 2.25690350e-01 4.05900806e-01 -1.26992238e+00 -5.04045963e-01 2.84985244e-01 2.48532057e-01 6.83358490e-01 5.77633008e-02 1.97769031e-01 -1.17663634e+00 6.41286492e-01 -6.42388105e-01 -1.08660555e+00 -1.10791206e+00 5.25802076e-01 2.98250288e-01 -7.92939886e-02 -5.01614392e-01 6.59812093e-01 1.05517693e-02 -1.09037960e+00 3.45683587e-03 -5.05651653e-01 -2.38489404e-01 -1.14200763e-01 -1.61443695e-01 1.96281940e-01 6.88146532e-01 -6.30060792e-01 -4.80817258e-01 8.63413453e-01 4.58090842e-01 3.08218271e-01 1.62313592e+00 2.17543557e-01 -1.46016195e-01 4.96358812e-01 8.06073248e-01 -4.74762440e-01 -9.55974102e-01 1.11628264e-01 3.13468814e-01 -7.25805163e-01 3.05887640e-01 -6.31760836e-01 -9.99901175e-01 1.00328302e+00 8.29219699e-01 4.37005252e-01 9.14152145e-01 1.67913646e-01 3.57406914e-01 5.25542259e-01 8.00832450e-01 -7.64164984e-01 -6.11246526e-01 5.61254799e-01 7.50221968e-01 -1.22643828e+00 2.37506792e-01 -1.13287139e+00 4.38123457e-02 9.75370586e-01 2.41160676e-01 -1.71105832e-01 1.07975912e+00 5.11427104e-01 3.23170535e-02 -3.93064767e-01 -2.41891071e-01 -5.23495972e-01 -1.01114638e-01 1.29785478e+00 4.86118406e-01 1.87531322e-01 1.32331073e-01 -1.25508979e-01 -1.40830979e-01 1.12699494e-01 3.18718582e-01 1.24753451e+00 -4.58247155e-01 -1.49433339e+00 -4.57628906e-01 3.22229624e-01 9.88241378e-03 1.68347955e-01 -4.89750206e-01 9.87605035e-01 2.76497573e-01 2.69190162e-01 2.21697137e-01 2.13417765e-02 6.25965953e-01 2.19720274e-01 7.52336860e-01 -9.46020246e-01 -4.17570084e-01 -1.56976447e-01 4.80580442e-02 -4.75014210e-01 -6.33187234e-01 -7.35762775e-01 -1.14122617e+00 -2.83714086e-01 -4.32147741e-01 -3.21760058e-01 1.42566872e+00 7.27931142e-01 4.95430112e-01 6.08795404e-01 3.58231276e-01 -1.28949213e+00 -1.22171871e-01 -4.88947093e-01 -6.93896472e-01 3.69263917e-01 -1.94887832e-01 -9.72707033e-01 1.79633245e-01 2.06744775e-01]
[8.75876522064209, -2.0179758071899414]
9dd25689-c240-497f-8e5b-583c7b656ccd
neurall-towards-a-unified-model-for-visual
1902.03589
null
https://arxiv.org/abs/1902.03589v2
https://arxiv.org/pdf/1902.03589v2.pdf
NeurAll: Towards a Unified Model for Visual Perception in Automated Driving
Convolutional Neural Networks (CNNs) are successfully used for the important automotive visual perception tasks including object recognition, motion and depth estimation, visual SLAM, etc. However, these tasks are typically independently explored and modeled. In this paper, we propose a joint multi-task network design for learning several tasks simultaneously. Our main motivation is the computational efficiency achieved by sharing the expensive initial convolutional layers between all tasks. Indeed, the main bottleneck in automated driving systems is the limited processing power available on deployment hardware. There is also some evidence for other benefits in improving accuracy for some tasks and easing development effort. It also offers scalability to add more tasks leveraging existing features and achieving better generalization. We survey various CNN based solutions for visual perception tasks in automated driving. Then we propose a unified CNN model for the important tasks and discuss several advanced optimization and architecture design techniques to improve the baseline model. The paper is partly review and partly positional with demonstration of several preliminary results promising for future research. We first demonstrate results of multi-stream learning and auxiliary learning which are important ingredients to scale to a large multi-task model. Finally, we implement a two-stream three-task network which performs better in many cases compared to their corresponding single-task models, while maintaining network size.
['Samir Rawashdeh', 'Ciaran Hughes', 'Senthil Yogamani', 'Sumanth Chennupati', 'Stefan Milz', 'Ganesh Sistu', 'Isabelle Leang']
2019-02-10
null
null
null
null
['auxiliary-learning']
['methodology']
[ 1.72831059e-01 -2.22331043e-02 -4.17971462e-01 -5.54502010e-01 -7.20307887e-01 -2.22635344e-01 5.16442180e-01 -2.10765690e-01 -7.04789639e-01 4.10809755e-01 -2.26731986e-01 -3.06251526e-01 9.48203281e-02 -3.34870398e-01 -8.38695168e-01 -6.60552979e-01 9.18466300e-02 3.81915092e-01 5.78126907e-01 -2.88377523e-01 1.89455017e-01 5.57551146e-01 -2.13097095e+00 4.36930507e-01 3.81703496e-01 1.27316821e+00 6.62531257e-01 1.02493048e+00 9.65223759e-02 1.02980387e+00 -7.13652015e-01 -2.07304776e-01 3.52745533e-01 3.19827020e-01 -6.79716170e-01 1.12671986e-01 8.24541688e-01 -1.82964742e-01 -2.79188395e-01 7.20152318e-01 5.30396581e-01 1.59266919e-01 2.63978660e-01 -1.88410103e+00 -5.21422252e-02 -1.41460244e-02 -3.14369500e-01 1.72622085e-01 -4.69280154e-01 9.20855179e-02 8.74552190e-01 -9.92579579e-01 3.58377397e-01 1.20556831e+00 6.28902197e-01 6.19833767e-01 -6.19447231e-01 -6.90829277e-01 3.49625587e-01 7.38071799e-01 -1.15483308e+00 -6.69109821e-01 7.11098790e-01 -2.77557492e-01 1.46450686e+00 1.95515268e-02 4.51417327e-01 1.11807740e+00 6.14030600e-01 1.20482457e+00 7.05972791e-01 8.58370736e-02 4.01264243e-02 2.24368498e-01 1.70371234e-01 7.84029424e-01 2.55902410e-01 2.12362185e-01 -9.00853753e-01 2.61166334e-01 4.34361666e-01 -1.26849994e-01 2.63502359e-01 -5.79156995e-01 -1.08474505e+00 8.73973429e-01 3.42833936e-01 -1.26563907e-01 -9.53518003e-02 6.17486119e-01 8.06353211e-01 3.24656010e-01 4.96275365e-01 1.78638086e-01 -7.59489298e-01 -2.72746623e-01 -9.45350349e-01 4.79795277e-01 6.14825726e-01 1.30229914e+00 1.08809328e+00 4.94407475e-01 -1.32064000e-02 8.09844494e-01 2.12766409e-01 4.94616717e-01 3.10083330e-01 -1.12758338e+00 4.88868743e-01 1.35350287e-01 -1.75958753e-01 -6.08939767e-01 -6.77694440e-01 -5.03503501e-01 -7.49625564e-01 7.23064005e-01 5.21244407e-02 -1.64457560e-01 -1.03434420e+00 1.36227930e+00 -3.34303081e-02 1.66400835e-01 1.78751081e-01 8.11602414e-01 9.57291484e-01 5.35232782e-01 8.31659511e-02 2.28758663e-01 1.49946570e+00 -1.64041626e+00 -6.67715013e-01 -9.83877897e-01 7.88933873e-01 -7.16871858e-01 6.22606933e-01 5.23165703e-01 -8.48979115e-01 -1.18612087e+00 -1.46642041e+00 -4.49048519e-01 -5.51101327e-01 5.23942590e-01 9.51973915e-01 6.57443583e-01 -1.17469692e+00 3.74261260e-01 -8.82150888e-01 -1.92674011e-01 6.66602135e-01 6.13214016e-01 -2.31338456e-01 -2.13931203e-01 -8.22817326e-01 1.03272605e+00 2.70943522e-01 1.85126901e-01 -1.40069532e+00 -4.00862098e-01 -1.20030749e+00 -1.40387177e-01 4.12674040e-01 -6.64654136e-01 1.55994022e+00 -8.50766778e-01 -1.38708794e+00 7.47052133e-01 -4.59929705e-01 -7.49953508e-01 2.30231345e-01 -3.92253995e-01 -2.36405104e-01 -2.77305305e-01 2.78022755e-02 1.40066957e+00 1.15758514e+00 -1.03601205e+00 -1.02516258e+00 -1.53120160e-01 1.30436987e-01 2.95969784e-01 -2.59668052e-01 -2.94843372e-02 -6.10773325e-01 -3.05142879e-01 -2.16926873e-01 -1.05216646e+00 -5.12274742e-01 1.46527916e-01 -1.78852186e-01 -4.55332458e-01 1.32493758e+00 -1.66697055e-01 6.54751956e-01 -2.21741462e+00 4.31962386e-02 -2.51437187e-01 5.73067188e-01 2.98290372e-01 -3.89333993e-01 3.69053483e-02 -2.48730257e-02 -3.45995575e-01 2.81790365e-03 -1.03933895e+00 -1.83535188e-01 5.97146630e-01 -8.05290416e-02 4.35976982e-01 5.35537601e-01 1.27521396e+00 -5.09473383e-01 -3.76135498e-01 3.95222336e-01 3.22328776e-01 -3.48656416e-01 8.58082622e-02 -1.36906356e-01 2.43995428e-01 -1.94034338e-01 7.99159169e-01 6.99790359e-01 -5.96389035e-03 -2.29891598e-01 -3.79868239e-01 -2.78892994e-01 3.57655406e-01 -1.06205618e+00 1.76265681e+00 -6.54536843e-01 1.36711621e+00 4.06882703e-01 -1.44958568e+00 9.44015741e-01 1.15491800e-01 4.33246851e-01 -7.53899813e-01 1.81822479e-01 1.45337760e-01 1.46813005e-01 -4.92841482e-01 9.74958181e-01 1.19030245e-01 -1.89261764e-01 1.28494781e-02 2.37850875e-01 -4.26054299e-01 7.65438899e-02 8.71060882e-04 7.71340072e-01 1.58588648e-01 9.86733213e-02 -3.94362688e-01 4.12153274e-01 2.12725639e-01 5.74010253e-01 6.15370274e-01 -5.89175463e-01 5.09881258e-01 4.14819896e-01 -7.37700522e-01 -1.20521009e+00 -5.32627106e-01 1.60323270e-02 1.35378766e+00 1.92781061e-01 -5.54709733e-01 -1.92115918e-01 -5.41175008e-01 1.74778402e-01 3.22143018e-01 -6.33402586e-01 -1.05540529e-01 -8.18710566e-01 -7.47356474e-01 7.04967260e-01 1.12867165e+00 6.03776753e-01 -8.76294494e-01 -9.35389519e-01 1.39984161e-01 -3.65259945e-02 -1.72841311e+00 -8.03058967e-02 8.03948820e-01 -8.03210974e-01 -8.61995101e-01 -5.11116743e-01 -1.03039622e+00 1.25419185e-01 7.44638205e-01 1.02233696e+00 -8.04796908e-03 -3.41008127e-01 1.17000774e-01 4.38392498e-02 -1.09168315e+00 -1.01754405e-01 3.34727466e-01 8.62772018e-02 -2.10447744e-01 2.91456342e-01 -2.40358025e-01 -3.43161613e-01 3.23792100e-01 -6.98084712e-01 1.06450319e-01 8.46243024e-01 7.68035054e-01 3.91006082e-01 -2.36821577e-01 4.87294793e-01 -4.21257138e-01 3.46865296e-01 -2.26683170e-01 -6.57542586e-01 -3.04076225e-01 -4.95709211e-01 3.72712091e-02 3.33078027e-01 -3.54079694e-01 -8.25038850e-01 2.85524130e-01 -3.74551058e-01 -7.06565678e-01 -2.90170908e-01 9.95061845e-02 1.01985782e-01 -4.71766561e-01 4.76158202e-01 1.06203193e-02 2.63164669e-01 -1.81985080e-01 2.79145002e-01 5.04603148e-01 3.58679593e-01 -3.14637244e-01 6.32711589e-01 6.98939204e-01 3.20570409e-01 -1.16222501e+00 -8.57047260e-01 -7.08764076e-01 -6.63520098e-01 -2.63535142e-01 9.83245075e-01 -1.28315282e+00 -9.20764923e-01 5.98794222e-01 -1.42335486e+00 -6.06742680e-01 -1.49446338e-01 4.99075770e-01 -6.81577563e-01 2.71984577e-01 -3.68094444e-01 -5.64477921e-01 -8.48292746e-03 -1.59153640e+00 1.47551250e+00 -7.51439575e-03 1.41572133e-01 -9.88030434e-01 -2.75731862e-01 5.08185625e-01 6.94507122e-01 -2.24674493e-01 5.30184984e-01 -4.78267252e-01 -9.16419625e-01 -3.30995582e-02 -4.21210438e-01 5.60162485e-01 -1.96153015e-01 -2.56817460e-01 -1.40781856e+00 -4.64543641e-01 -1.51015878e-01 -7.42416143e-01 1.42672658e+00 6.46927118e-01 1.30579162e+00 3.94074202e-01 -5.85925817e-01 8.49187791e-01 1.17710519e+00 9.62965041e-02 6.49825931e-01 4.35471892e-01 9.61123765e-01 7.62750626e-01 7.51954019e-01 7.64374584e-02 4.92116481e-01 7.59872198e-01 8.12707126e-01 -3.62145662e-01 -3.68619978e-01 3.98490578e-01 5.54489255e-01 7.70345867e-01 -6.02051504e-02 -2.65239239e-01 -8.16780210e-01 7.55966425e-01 -1.94347811e+00 -5.82691193e-01 -3.40549946e-01 1.83218312e+00 4.10601050e-02 5.69323540e-01 7.88777098e-02 5.74252605e-02 1.28104523e-01 4.08179820e-01 -6.46644294e-01 -6.50885463e-01 -2.56155014e-01 2.98805296e-01 9.78288829e-01 4.48144972e-01 -1.39474964e+00 1.05771339e+00 7.00569439e+00 7.80364811e-01 -1.14722741e+00 2.66012073e-01 4.36957479e-01 -3.97749901e-01 2.42262781e-01 -1.67251289e-01 -1.37394810e+00 -2.12951705e-01 1.01255298e+00 3.03827543e-02 -6.41075522e-02 1.25368416e+00 4.29816432e-02 -1.85685486e-01 -1.14368010e+00 1.23654652e+00 3.57127666e-01 -1.41931021e+00 -1.76857680e-01 -7.02603906e-03 6.96563780e-01 6.36884332e-01 1.73662782e-01 5.52214146e-01 7.09140077e-02 -1.12365162e+00 8.19183230e-01 -4.25078012e-02 8.51271987e-01 -8.83782685e-01 9.79492068e-01 2.54961073e-01 -1.59288573e+00 -3.13126087e-01 -5.93205273e-01 -2.50572145e-01 1.32332832e-01 4.47583616e-01 -6.99432135e-01 5.30196846e-01 8.50303411e-01 7.88849235e-01 -6.83818161e-01 9.08661366e-01 1.75897740e-02 2.06466600e-01 -1.78576350e-01 -1.55770585e-01 6.54976010e-01 3.78048837e-01 4.46914554e-01 1.45703745e+00 1.44832358e-01 -6.35884762e-01 3.95487249e-01 4.15071309e-01 1.10469803e-01 -2.76937187e-01 -8.72030020e-01 4.04053599e-01 1.57474786e-01 1.47564530e+00 -4.68460500e-01 -4.30375218e-01 -8.89635324e-01 8.16917181e-01 3.03102434e-01 3.43329459e-01 -9.52676654e-01 -5.95425189e-01 1.31708479e+00 -2.02265620e-01 4.96943206e-01 -7.82727599e-01 -4.80098575e-01 -7.83003390e-01 9.61997211e-02 -6.62228465e-01 2.62632295e-02 -6.49136603e-01 -6.82911873e-01 6.79031968e-01 6.28127977e-02 -1.15474081e+00 -2.80357063e-01 -1.27033019e+00 -3.77569735e-01 6.92094088e-01 -2.26521873e+00 -1.34309220e+00 -6.51199639e-01 5.71203768e-01 1.32645702e+00 -4.26434904e-01 4.81807441e-01 4.76934463e-01 -6.22883379e-01 6.07154071e-01 -3.03040504e-01 -1.09202787e-01 8.55702639e-01 -1.06523979e+00 7.82440841e-01 7.10282445e-01 -8.14237148e-02 5.30760586e-02 4.36875850e-01 -3.83916795e-01 -1.48043323e+00 -1.38795698e+00 7.05762148e-01 -4.56990272e-01 5.18157125e-01 -7.44443893e-01 -5.77036083e-01 9.51258421e-01 4.75363165e-01 -1.35019235e-02 2.76342422e-01 7.70101696e-02 -1.42698228e-01 -2.93455780e-01 -6.08458161e-01 3.80470008e-01 9.24025178e-01 -5.03571033e-01 -7.97923952e-02 4.05947328e-01 7.61273026e-01 -6.07840657e-01 -4.34794128e-01 5.71235538e-01 5.14568090e-01 -7.95485139e-01 9.94173944e-01 -6.16482019e-01 2.88003266e-01 -3.22670996e-01 -2.68749863e-01 -1.08054638e+00 -3.26398313e-01 -5.10045290e-01 -1.66416228e-01 4.75695521e-01 5.41379571e-01 -5.90327621e-01 1.13150001e+00 1.69261783e-01 -8.05425942e-01 -7.52946854e-01 -1.09888399e+00 -9.05913174e-01 -2.17241347e-02 -1.07381725e+00 1.33347869e-01 4.78947699e-01 -5.33772349e-01 5.93536139e-01 -6.71000361e-01 2.28804067e-01 6.69706047e-01 -1.92226082e-01 1.02022958e+00 -1.19901109e+00 -1.03863858e-01 -2.81291127e-01 -6.44420981e-01 -1.70215583e+00 2.10190251e-01 -7.88437128e-01 3.58717024e-01 -1.35534179e+00 -1.36154806e-02 -2.84929723e-01 -1.05063677e-01 5.96268535e-01 7.28671178e-02 4.02059138e-01 2.65324354e-01 9.67272818e-02 -7.70362258e-01 5.69459200e-01 1.23583627e+00 -2.85812527e-01 4.26756218e-02 1.82738796e-01 -5.31942785e-01 6.40279889e-01 9.28584814e-01 -2.98901379e-01 -4.66693759e-01 -7.94504285e-01 9.61358249e-02 -3.90808403e-01 6.71539545e-01 -1.27633631e+00 5.73135197e-01 4.31570932e-02 2.97297299e-01 -9.91548836e-01 9.54804838e-01 -6.56232595e-01 -4.87325966e-01 4.35371399e-01 -2.17572302e-02 3.90120506e-01 7.35103488e-01 4.07023013e-01 -4.37516332e-01 -1.07173845e-01 8.49209428e-01 -5.84648252e-02 -1.49926472e+00 3.78705591e-01 -7.37432718e-01 -3.10059339e-01 1.06584561e+00 -3.33645165e-01 -3.28577518e-01 -3.51924419e-01 -6.63710654e-01 4.80813622e-01 -9.28247496e-02 9.33024943e-01 8.68736625e-01 -1.29302812e+00 -8.10502052e-01 3.53964895e-01 3.68201494e-01 2.23059431e-01 1.60896152e-01 7.91214406e-01 -3.35065484e-01 9.00747895e-01 -4.15454507e-01 -9.84817266e-01 -1.48274589e+00 5.12226701e-01 3.84875506e-01 -1.44195020e-01 -3.01822960e-01 9.94168997e-01 3.07487309e-01 -3.25266272e-01 4.01626348e-01 -5.48882961e-01 -1.67852730e-01 -2.80458182e-02 3.42298090e-01 3.84465933e-01 5.19654930e-01 -4.83347088e-01 -4.07713473e-01 5.64538181e-01 -1.92303658e-01 1.78480685e-01 1.16531169e+00 -3.60507891e-03 2.35565394e-01 3.32901001e-01 1.34235072e+00 -4.83017445e-01 -1.41577792e+00 -1.04827881e-01 -9.58555564e-02 -3.37751620e-02 3.20852369e-01 -1.52095169e-01 -1.21324825e+00 1.30706310e+00 6.65653288e-01 7.17189070e-03 1.08140945e+00 5.07723428e-02 6.83897316e-01 8.52065921e-01 2.53689587e-01 -1.17712259e+00 3.91149580e-01 9.14710164e-01 7.30368137e-01 -1.53973973e+00 -1.50113255e-02 -4.91041720e-01 -7.61722922e-01 1.24759507e+00 1.05029774e+00 -7.15154409e-02 6.33676350e-01 7.50983000e-01 1.13361068e-01 -4.04233426e-01 -1.12447536e+00 -5.69725037e-01 3.08367908e-01 8.53559494e-01 3.03422183e-01 -2.46702835e-01 2.96227813e-01 9.24894884e-02 -1.17020503e-01 -2.54624426e-01 3.48675489e-01 1.04469049e+00 -6.91105545e-01 -1.11219645e+00 -7.25568533e-02 3.02601278e-01 -2.48081326e-01 -4.39098105e-02 -1.12474918e-01 9.41908360e-01 4.23502326e-01 1.01984036e+00 3.22166681e-01 -5.85083485e-01 3.47095042e-01 -1.30168498e-02 3.49417776e-01 -5.73358357e-01 -3.52668881e-01 -9.93041694e-02 4.45765018e-01 -8.96606803e-01 -5.89793026e-01 -5.88068366e-01 -8.00202727e-01 -1.62951604e-01 -2.97662467e-01 -3.30702424e-01 1.10059786e+00 9.57730055e-01 4.02207911e-01 1.02275205e+00 4.47940767e-01 -1.30531335e+00 -2.24671319e-01 -8.91852736e-01 -2.60067791e-01 -1.66737646e-01 6.46941781e-01 -8.67713630e-01 1.30690299e-02 8.11119750e-02]
[8.129928588867188, -1.3603373765945435]
c8299c9a-822b-4541-80f9-8071c0510789
evidential-deep-learning-for-class
2212.02863
null
https://arxiv.org/abs/2212.02863v1
https://arxiv.org/pdf/2212.02863v1.pdf
Evidential Deep Learning for Class-Incremental Semantic Segmentation
Class-Incremental Learning is a challenging problem in machine learning that aims to extend previously trained neural networks with new classes. This is especially useful if the system is able to classify new objects despite the original training data being unavailable. While the semantic segmentation problem has received less attention than classification, it poses distinct problems and challenges since previous and future target classes can be unlabeled in the images of a single increment. In this case, the background, past and future classes are correlated and there exist a background-shift. In this paper, we address the problem of how to model unlabeled classes while avoiding spurious feature clustering of future uncorrelated classes. We propose to use Evidential Deep Learning to model the evidence of the classes as a Dirichlet distribution. Our method factorizes the problem into a separate foreground class probability, calculated by the expected value of the Dirichlet distribution, and an unknown class (background) probability corresponding to the uncertainty of the estimate. In our novel formulation, the background probability is implicitly modeled, avoiding the feature space clustering that comes from forcing the model to output a high background score for pixels that are not labeled as objects. Experiments on the incremental Pascal VOC, and ADE20k benchmarks show that our method is superior to state-of-the-art, especially when repeatedly learning new classes with increasing number of increments.
['Michael Felsberg', 'Lena Klasén', 'Karl Holmquist']
2022-12-06
null
null
null
null
['class-incremental-semantic-segmentation']
['computer-vision']
[ 7.60105789e-01 1.96841821e-01 -2.08107859e-01 -5.63893557e-01 -5.53990543e-01 -4.81393516e-01 6.42924428e-01 2.43476070e-02 -4.75053102e-01 1.10941780e+00 -5.13511181e-01 -2.01257870e-01 1.21078737e-01 -8.27762842e-01 -9.96020675e-01 -1.04553378e+00 2.32779846e-01 6.79229856e-01 7.41036415e-01 5.42634010e-01 6.83837384e-02 3.38182092e-01 -1.73011065e+00 5.11991799e-01 8.77379656e-01 1.00778079e+00 2.24983737e-01 6.40565455e-01 -4.83079910e-01 7.93539643e-01 -9.10844684e-01 -2.87167877e-01 2.12768063e-01 -5.29717326e-01 -7.81525791e-01 7.15786934e-01 6.48169398e-01 -2.88290530e-01 1.43066064e-01 1.34751904e+00 -6.61638677e-02 1.53513312e-01 7.02465892e-01 -1.35727119e+00 -3.26306313e-01 3.74884814e-01 -6.69431627e-01 3.10609311e-01 -3.68871927e-01 1.28630847e-02 4.30791378e-01 -7.61877120e-01 5.70752919e-01 1.04977322e+00 6.33780718e-01 6.11725628e-01 -1.26113641e+00 -5.54244399e-01 7.20540702e-01 5.01624644e-01 -1.34618473e+00 -8.81655440e-02 7.59734571e-01 -6.38318598e-01 4.66224581e-01 -4.49349871e-03 7.50782251e-01 9.49996173e-01 4.88236919e-02 9.37108159e-01 1.12351084e+00 -5.46214461e-01 6.50239348e-01 5.10804713e-01 4.46238220e-01 3.42005283e-01 3.88215899e-01 -3.39238346e-02 -2.71026105e-01 6.28843158e-02 3.31877857e-01 8.04244503e-02 -4.00196612e-02 -4.94697392e-01 -7.84757137e-01 6.56450093e-01 1.71495080e-01 2.50338405e-01 -3.32571387e-01 2.45758340e-01 2.07832898e-03 -1.91790774e-01 6.91429019e-01 -2.33396068e-01 -7.13458180e-01 2.65868664e-01 -1.15957391e+00 -1.08593181e-02 6.74394429e-01 7.75793970e-01 1.06933022e+00 5.82169965e-02 -5.56582958e-03 6.86746895e-01 2.21651778e-01 4.73098010e-01 4.33752298e-01 -9.11815047e-01 -6.05081907e-03 4.25382406e-01 1.85443297e-01 -5.58094025e-01 3.07362005e-02 -7.43622661e-01 -6.11599982e-01 3.85651499e-01 6.28080189e-01 -2.36028433e-01 -1.44000816e+00 1.62943721e+00 6.86280608e-01 7.96663821e-01 1.42305791e-01 2.85040081e-01 3.14809501e-01 8.79090369e-01 -1.21858940e-02 -3.19732249e-01 8.59237909e-01 -1.05065966e+00 -6.98421538e-01 -6.19671524e-01 1.70999989e-01 -5.36128163e-01 4.73028362e-01 6.13885939e-01 -6.48973405e-01 -8.88158262e-01 -1.02505612e+00 4.73659486e-01 -5.28617263e-01 8.42754394e-02 5.01417696e-01 7.12659299e-01 -8.68218958e-01 6.59616709e-01 -1.00459826e+00 -2.76882619e-01 7.53313184e-01 2.52598554e-01 1.23483442e-01 -1.64942339e-01 -8.78578901e-01 6.40584409e-01 8.38374078e-01 1.40341222e-01 -9.81043041e-01 -4.61783797e-01 -3.62645119e-01 -7.46200532e-02 5.43103397e-01 -3.02863210e-01 1.22285366e+00 -1.52523160e+00 -1.08778358e+00 7.93265939e-01 -3.80662143e-01 -6.23513818e-01 5.91735363e-01 -6.50620535e-02 -3.10152322e-01 3.19405273e-02 2.83459157e-01 8.35568368e-01 1.10717475e+00 -1.59937823e+00 -1.32684696e+00 -4.49967712e-01 -2.58532673e-01 3.28810811e-02 -7.29439035e-02 -6.23797655e-01 -4.55658704e-01 -3.66223991e-01 3.84148687e-01 -1.05523229e+00 -2.82262504e-01 1.17836319e-01 -9.20440778e-02 -3.00280690e-01 1.11536312e+00 -4.12297100e-01 7.58692086e-01 -2.26389670e+00 -1.95963800e-01 -5.66920266e-02 -1.52701316e-02 2.49527007e-01 2.72412837e-01 -4.51320529e-01 5.16198613e-02 -1.37470797e-01 -5.31165123e-01 -2.52952874e-02 -2.63551563e-01 7.42806017e-01 -3.29207569e-01 4.22395587e-01 5.28692305e-01 5.65297544e-01 -1.00667989e+00 -4.81755227e-01 2.17880622e-01 2.17269316e-01 -2.53180981e-01 1.57489907e-02 -6.72254264e-01 3.44859123e-01 -1.15498744e-01 5.85898876e-01 1.06058311e+00 -2.54193634e-01 1.07453845e-01 7.74709061e-02 6.53389376e-03 -3.30594301e-01 -1.49240184e+00 1.26960516e+00 -6.47544861e-02 6.62274003e-01 -1.56215519e-01 -1.27080894e+00 7.42379248e-01 -6.63815960e-02 2.38577127e-01 -1.60007268e-01 -5.96344173e-02 1.86358079e-01 2.54524383e-03 -3.60165507e-01 3.72127920e-01 -3.96963805e-01 3.11336040e-01 1.34011790e-01 2.83918321e-01 -1.90686196e-01 2.78810799e-01 -1.68364425e-03 7.56742179e-01 3.60304356e-01 -3.58294807e-02 -5.05233482e-02 4.23622996e-01 2.14137018e-01 8.79202187e-01 1.04136026e+00 -3.67714345e-01 4.48507130e-01 4.08938557e-01 -2.82974035e-01 -6.73548520e-01 -1.34361684e+00 -3.56809258e-01 8.50368559e-01 3.59203905e-01 4.31285888e-01 -8.04771900e-01 -1.09855974e+00 -1.30438730e-01 9.86174524e-01 -7.75244236e-01 -1.50869444e-01 -5.63207746e-01 -1.22158623e+00 6.09185128e-03 5.75507164e-01 6.07233942e-01 -8.62291873e-01 -5.85816741e-01 4.00949985e-01 -2.44444817e-01 -1.14791548e+00 -1.49427708e-02 7.69818127e-01 -1.05454671e+00 -1.09428263e+00 -7.21607685e-01 -8.44361305e-01 8.69071901e-01 1.51640028e-01 1.00380898e+00 -2.18109742e-01 -4.20590729e-01 5.89385390e-01 -1.58083424e-01 -6.98745430e-01 -4.09119517e-01 -3.64175975e-01 -2.61856705e-01 3.63786459e-01 6.49648845e-01 -1.24383032e-01 -3.57834548e-01 1.57617405e-01 -1.04214382e+00 -2.27573644e-02 3.75262052e-01 9.78699088e-01 8.66672277e-01 6.51938081e-01 6.48436129e-01 -9.88685846e-01 -2.42086500e-01 -5.95555544e-01 -6.79773271e-01 3.27221900e-01 -5.23368180e-01 -1.47650182e-01 4.25490499e-01 -7.89011657e-01 -1.54599094e+00 4.42836314e-01 9.75639671e-02 -3.05735886e-01 -5.53310394e-01 4.82296646e-02 -2.43677884e-01 3.15782994e-01 5.14798701e-01 2.59109765e-01 -3.81158143e-01 -3.16684157e-01 4.11576003e-01 5.52092552e-01 7.08279669e-01 -5.66997170e-01 5.47990739e-01 9.08195436e-01 -3.58173847e-02 -9.28673148e-01 -1.23327291e+00 -7.52689362e-01 -1.07066929e+00 -5.04408777e-01 9.46653187e-01 -7.93500125e-01 2.33190909e-01 8.78255546e-01 -1.16058552e+00 -3.83246183e-01 -7.32958972e-01 3.56901646e-01 -3.49781603e-01 3.78980458e-01 -1.56977400e-01 -1.15744877e+00 2.68113524e-01 -9.88732338e-01 8.25055480e-01 4.99238968e-01 1.16418771e-01 -1.04516792e+00 -2.90301353e-01 2.88049519e-01 -1.42538399e-02 3.44389915e-01 9.48451221e-01 -7.12089121e-01 -7.65720546e-01 -2.95458555e-01 -2.08351091e-01 8.50531399e-01 3.00766051e-01 1.56559069e-02 -1.22525597e+00 -6.61556199e-02 2.78868079e-01 -1.41039044e-01 1.23235810e+00 4.64383006e-01 1.25887287e+00 -1.53825387e-01 -5.67349911e-01 1.99049845e-01 1.41702950e+00 6.39631748e-01 3.89828742e-01 1.89496547e-01 5.28738081e-01 6.34933352e-01 6.79081917e-01 3.59630078e-01 1.72575265e-02 2.11356089e-01 5.53864896e-01 -2.10061837e-02 -4.29217726e-01 1.04077339e-01 1.54372439e-01 2.43574604e-01 4.27417248e-01 -3.08754861e-01 -8.83242369e-01 8.00723255e-01 -1.89183092e+00 -9.24178123e-01 -2.71363348e-01 2.35760880e+00 8.38750303e-01 5.02603471e-01 -2.69635409e-01 2.44839057e-01 1.09058654e+00 -3.26836616e-01 -1.01223743e+00 1.57520324e-01 -2.10393757e-01 3.33993793e-01 2.99480021e-01 6.20496154e-01 -1.30549359e+00 9.02097583e-01 5.66597033e+00 7.68227220e-01 -1.01523340e+00 1.24704480e-01 1.05017471e+00 7.21474066e-02 1.32016003e-01 1.06589437e-01 -1.32280338e+00 5.49675405e-01 6.92992926e-01 8.46675634e-02 -3.47860530e-02 9.71738875e-01 -3.29030514e-01 -7.14257240e-01 -1.19226933e+00 5.68574727e-01 1.89649403e-01 -1.10154819e+00 5.53930551e-03 -1.32231608e-01 1.10624135e+00 1.89882051e-02 1.66065156e-01 4.97248501e-01 2.65404612e-01 -5.36391377e-01 9.36753929e-01 4.98586953e-01 4.22376603e-01 -5.40476799e-01 8.48471165e-01 5.86755097e-01 -8.14846933e-01 -1.93887055e-01 -3.79409581e-01 -2.76484136e-02 -5.26168644e-02 8.32536519e-01 -1.08394301e+00 1.88063100e-01 7.45122850e-01 5.65918922e-01 -6.39576316e-01 1.05731416e+00 -2.57325500e-01 7.51749396e-01 -4.41130459e-01 3.81436646e-02 3.08269024e-01 -1.32099211e-01 2.50236005e-01 1.10335541e+00 2.68598050e-01 -1.01680279e-01 3.53409559e-01 9.68612492e-01 9.33597907e-02 -3.69693130e-01 -1.74501300e-01 1.86115414e-01 2.01168612e-01 9.60491300e-01 -1.51019490e+00 -7.97171175e-01 -3.86520237e-01 1.27722645e+00 1.58416256e-01 4.95820254e-01 -9.15951908e-01 -5.61559908e-02 1.84336916e-01 8.69574696e-02 6.99067533e-01 3.55148017e-02 -2.95269608e-01 -1.01742387e+00 7.38979205e-02 -3.29072475e-01 4.02472436e-01 -6.95263863e-01 -1.40884829e+00 4.49753284e-01 1.26855925e-01 -8.85599196e-01 -2.35788956e-01 -8.18831921e-01 -4.94933337e-01 6.11275315e-01 -1.66503882e+00 -9.09033060e-01 -3.75112683e-01 3.05949271e-01 7.68172622e-01 1.51539564e-01 5.24919927e-01 1.84511885e-01 -3.65235984e-01 7.82491490e-02 5.19071043e-01 -2.65271515e-02 5.71730077e-01 -1.39095294e+00 -4.67237458e-03 1.06134176e+00 1.55576244e-01 6.21119402e-02 6.68287456e-01 -8.84785533e-01 -5.15531957e-01 -1.44549465e+00 5.25898695e-01 -5.59895098e-01 6.09951973e-01 -2.39526257e-01 -1.04557157e+00 5.72684109e-01 -1.43256187e-01 4.45417643e-01 5.92623651e-01 -3.06984425e-01 1.43577028e-02 -1.39510587e-01 -1.24510634e+00 2.50605851e-01 5.98108888e-01 -1.17934467e-02 -5.27643263e-01 4.10081536e-01 6.00385368e-01 -1.51382625e-01 -1.77009866e-01 5.73936164e-01 3.49636972e-01 -8.88948858e-01 7.54413128e-01 -4.63527679e-01 1.77121133e-01 -5.29418349e-01 -3.55476707e-01 -1.11473751e+00 -1.60887212e-01 1.13171838e-01 -2.72568434e-01 1.37574577e+00 3.97190332e-01 -4.35459077e-01 1.14515269e+00 4.39969689e-01 5.83238304e-02 -5.94123244e-01 -1.10159445e+00 -9.93506670e-01 2.24080950e-01 -4.35376614e-01 2.91410387e-01 1.00589001e+00 -9.16979492e-01 2.24834129e-01 -9.15473923e-02 3.68547559e-01 8.53920579e-01 1.83503464e-01 6.99198544e-01 -1.49986708e+00 -4.10338849e-01 -1.96009576e-01 -3.18948567e-01 -9.98947620e-01 1.74139172e-01 -7.45056152e-01 4.93885607e-01 -1.48483038e+00 4.26796794e-01 -4.48810577e-01 -3.95042390e-01 3.80677849e-01 -3.52733493e-01 8.73686150e-02 3.51318382e-02 -5.78781962e-02 -7.69525945e-01 4.22209412e-01 8.96471679e-01 -4.84212756e-01 -2.33567923e-01 3.51694852e-01 -3.36065769e-01 1.15136230e+00 6.99388742e-01 -7.84362853e-01 -2.84365296e-01 -2.64346331e-01 -1.47918940e-01 -3.87034029e-01 5.13184845e-01 -1.27772164e+00 1.99348524e-01 -1.13026679e-01 9.25478041e-01 -1.09472322e+00 3.59645396e-01 -1.07399893e+00 7.52962306e-02 5.37388444e-01 -4.77789827e-02 -7.40126848e-01 3.58005822e-01 1.07326829e+00 -4.45451699e-02 -7.56431460e-01 1.05662405e+00 -4.28883284e-01 -9.22373712e-01 1.26036838e-01 -6.31392956e-01 5.65615594e-02 1.14831710e+00 -4.56591010e-01 -2.03662440e-01 5.52333668e-02 -1.07021630e+00 5.46427667e-02 1.98489681e-01 2.40083233e-01 4.46184158e-01 -8.86738181e-01 -5.08748055e-01 1.37743667e-01 3.26693151e-03 4.92650568e-01 3.13762009e-01 3.15396100e-01 -2.04212099e-01 1.78552829e-02 1.49549544e-01 -1.08774567e+00 -1.22785926e+00 5.57347238e-01 3.00920784e-01 -1.57800540e-01 -3.08566242e-01 1.09038544e+00 4.48672384e-01 -3.14765930e-01 5.60242712e-01 -2.58349240e-01 -6.50857314e-02 2.15979218e-01 4.19394344e-01 4.04196143e-01 2.77864523e-02 -3.86525810e-01 -1.03129737e-01 2.74755031e-01 -2.74582684e-01 4.17702831e-03 1.24026299e+00 -2.68675208e-01 -1.89252272e-01 9.81268823e-01 8.60041797e-01 -3.58870536e-01 -1.67167270e+00 -4.74453449e-01 2.18260825e-01 -3.32242429e-01 4.12305593e-02 -1.13899326e+00 -7.74518251e-01 1.02333117e+00 1.15026712e+00 4.05729003e-02 1.02339745e+00 1.66722164e-01 4.73756403e-01 2.87541419e-01 2.08364457e-01 -1.32545567e+00 3.07351947e-01 6.10386074e-01 2.32682601e-01 -1.43546438e+00 -5.69121204e-02 -3.66455287e-01 -3.78034502e-01 1.13342726e+00 9.30712700e-01 1.70439169e-01 8.42427194e-01 1.91849008e-01 3.78790312e-02 2.23546147e-01 -4.42296356e-01 -2.99945891e-01 1.80294901e-01 7.08297908e-01 1.67411380e-02 1.72080640e-02 2.29107618e-01 7.02989221e-01 1.92826822e-01 7.53321871e-02 6.26830518e-01 1.07961893e+00 -7.57566035e-01 -8.28787565e-01 -6.06400311e-01 4.56528962e-01 -5.20439863e-01 4.47241124e-03 -2.97180116e-01 5.96500218e-01 9.67253685e-01 8.35991502e-01 4.05205578e-01 1.55853361e-01 -1.18826739e-01 4.52681243e-01 5.97204804e-01 -7.63027549e-01 1.85352102e-01 2.71643609e-01 -2.82485664e-01 4.44326084e-03 -6.19281232e-01 -7.88313627e-01 -1.25784469e+00 5.73006034e-01 -7.71823883e-01 9.89331082e-02 8.57236803e-01 1.06484842e+00 -1.08321607e-01 8.71140540e-01 4.86446440e-01 -8.19064975e-01 -3.37083817e-01 -7.66479492e-01 -6.09931588e-01 2.25641891e-01 4.09702420e-01 -6.73797488e-01 -6.70610130e-01 6.02779210e-01]
[9.383620262145996, 1.560594081878662]
615dfda9-1a91-4896-b0ea-f1d989d9de9e
ultra-high-resolution-detector-simulation
2303.08046
null
https://arxiv.org/abs/2303.08046v1
https://arxiv.org/pdf/2303.08046v1.pdf
Ultra-High-Resolution Detector Simulation with Intra-Event Aware GAN and Self-Supervised Relational Reasoning
Simulating high-resolution detector responses is a storage-costly and computationally intensive process that has long been challenging in particle physics. Despite the ability of deep generative models to make this process more cost-efficient, ultra-high-resolution detector simulation still proves to be difficult as it contains correlated and fine-grained mutual information within an event. To overcome these limitations, we propose Intra-Event Aware GAN (IEA-GAN), a novel fusion of Self-Supervised Learning and Generative Adversarial Networks. IEA-GAN presents a Relational Reasoning Module that approximates the concept of an ''event'' in detector simulation, allowing for the generation of correlated layer-dependent contextualized images for high-resolution detector responses with a proper relational inductive bias. IEA-GAN also introduces a new intra-event aware loss and a Uniformity loss, resulting in significant enhancements to image fidelity and diversity. We demonstrate IEA-GAN's application in generating sensor-dependent images for the high-granularity Pixel Vertex Detector (PXD), with more than 7.5M information channels and a non-trivial geometry, at the Belle II Experiment. Applications of this work include controllable simulation-based inference and event generation, high-granularity detector simulation such as at the HL-LHC (High Luminosity LHC), and fine-grained density estimation and sampling. To the best of our knowledge, IEA-GAN is the first algorithm for faithful ultra-high-resolution detector simulation with event-based reasoning.
['Thomas Kuhr', 'James Kahn', 'Sahand Sharifzadeh', 'Nikolai Hartmann', 'Hosein Hashemi']
2023-03-07
null
null
null
null
['metric-learning', 'conditional-image-generation', 'metric-learning', 'relational-reasoning']
['computer-vision', 'computer-vision', 'methodology', 'natural-language-processing']
[ 7.07094148e-02 7.28936270e-02 3.71836990e-01 -5.39139152e-01 -1.51128447e+00 -2.57033736e-01 9.25563693e-01 1.59924090e-01 -2.54950970e-01 9.91379917e-01 5.52953705e-02 -2.84677386e-01 -7.26555139e-02 -1.38859582e+00 -1.09737003e+00 -8.87122333e-01 9.48803574e-02 1.39002180e+00 2.52108544e-01 -1.44406147e-02 -4.05800432e-01 9.39493239e-01 -1.33623445e+00 7.16682732e-01 1.35606363e-01 9.20144737e-01 1.10639349e-01 8.71134460e-01 2.88468331e-01 1.06916642e+00 -5.84321499e-01 -4.09697026e-01 1.40930757e-01 -9.36700523e-01 -2.87508219e-01 -4.55222249e-01 -1.69093698e-01 -3.07069600e-01 -6.70023799e-01 8.27320278e-01 9.57716405e-01 -1.36938930e-01 6.64178550e-01 -1.21612680e+00 -1.47993654e-01 9.23430681e-01 -2.67059416e-01 8.23898166e-02 1.92990780e-01 6.64013505e-01 4.65280712e-01 -5.62166333e-01 8.20857644e-01 1.16246533e+00 6.92421317e-01 5.36503971e-01 -1.49010873e+00 -6.10484719e-01 -8.92218471e-01 -8.24323446e-02 -1.36464095e+00 -8.87900144e-02 4.71240669e-01 -2.73445934e-01 1.23915184e+00 3.76379251e-01 7.34529972e-01 1.34114754e+00 5.13353050e-01 2.33206615e-01 1.47454596e+00 -4.98674750e-01 7.05744684e-01 1.68798804e-01 -5.11925340e-01 2.57689685e-01 1.68638200e-01 8.85700822e-01 -6.06325567e-01 -2.01388195e-01 1.08441973e+00 -2.82414466e-01 1.98491350e-01 -2.09646672e-01 -1.27466702e+00 8.77894342e-01 6.63218558e-01 1.66254669e-01 -4.48170781e-01 4.92086589e-01 5.01749098e-01 4.77076508e-02 3.05253386e-01 3.40839595e-01 9.32167098e-02 -8.31192806e-02 -1.04051912e+00 6.15569472e-01 7.00679779e-01 1.06556547e+00 4.14886564e-01 2.63216317e-01 -7.26277947e-01 4.80730683e-02 1.44544676e-01 9.15433049e-01 -2.06787109e-01 -5.06145120e-01 1.03031151e-01 -5.43330982e-02 -4.55338322e-02 -2.78264672e-01 -3.97892773e-01 -4.47018504e-01 -1.15961373e+00 6.16361022e-01 2.25657403e-01 -1.64716065e-01 -1.02955949e+00 1.73411751e+00 5.08508742e-01 7.86572844e-02 1.52163461e-01 9.94888246e-01 7.32663512e-01 7.72116303e-01 2.95252025e-01 -1.50831416e-01 1.42198002e+00 -2.50348479e-01 -5.33551216e-01 -6.90176189e-02 1.23660348e-01 -7.08348334e-01 5.31762540e-01 1.09921381e-01 -1.36885953e+00 -4.59137976e-01 -1.12728751e+00 -1.12465158e-01 -6.91191033e-02 -4.63581145e-01 1.00946462e+00 6.16367638e-01 -8.70066226e-01 5.16307414e-01 -8.68167877e-01 4.46299538e-02 3.65721524e-01 1.52000323e-01 4.98030074e-02 1.37000317e-02 -1.14416718e+00 7.87478447e-01 5.74930847e-01 -3.95056307e-01 -1.02061534e+00 -1.11209452e+00 -5.77533066e-01 -8.82009715e-02 8.09499100e-02 -1.16232419e+00 1.42452121e+00 -3.19542170e-01 -1.47140503e+00 7.16275990e-01 2.65430570e-01 -1.02119899e+00 3.89181763e-01 6.12664044e-01 -5.76623380e-01 1.70530379e-02 3.56923155e-02 6.35315835e-01 5.68649590e-01 -1.05886781e+00 -1.58275291e-01 -1.28230199e-01 -2.30986893e-01 -2.75485162e-02 1.02378058e+00 1.53405964e-01 -1.84405357e-01 -5.88962495e-01 1.05847493e-02 -7.81464934e-01 -4.32691067e-01 -6.11912608e-01 -4.10374254e-01 3.20960134e-01 5.75304389e-01 -3.34080219e-01 4.69285876e-01 -1.91434050e+00 -2.38969345e-02 1.28041878e-01 2.54801184e-01 4.45270650e-02 4.90495086e-01 5.21947265e-01 -1.33519799e-01 -3.51057351e-01 -2.84170687e-01 -2.82775372e-01 2.85184890e-01 2.47386739e-01 -2.65928388e-01 1.85426250e-01 1.23792037e-01 1.47509277e+00 -9.49413121e-01 -4.22540456e-01 7.86304057e-01 7.94013083e-01 -4.14018184e-01 1.86828718e-01 -7.21164644e-01 9.09287035e-01 -3.69581252e-01 5.94630897e-01 9.35073018e-01 -1.87300801e-01 2.00599004e-02 -5.65698743e-01 2.00582929e-02 1.91671118e-01 -8.06618452e-01 1.68758190e+00 -5.23578107e-01 3.16794485e-01 1.10468470e-01 -5.68743348e-01 5.94778538e-01 1.45975903e-01 3.50321442e-01 -1.40729690e+00 3.13222438e-01 -4.47183382e-03 -1.13603689e-01 3.06331426e-01 9.11221027e-01 -8.24679911e-01 -6.37679696e-01 4.80471075e-01 1.96024090e-01 -7.05927193e-01 -2.72075653e-01 6.89100862e-01 1.31901157e+00 -4.44751717e-02 5.56058213e-02 -1.39227688e-01 -1.96717560e-01 1.18884347e-01 2.42638364e-01 1.15260768e+00 2.49138325e-01 8.39431286e-01 2.44427174e-01 -1.06148787e-01 -1.88259053e+00 -1.44604456e+00 -4.75151539e-01 3.33351046e-02 -1.47323877e-01 -5.56203842e-01 -7.58272469e-01 -2.29051948e-01 6.30714893e-02 1.34578431e+00 -3.47943082e-02 -1.43051907e-01 -4.32137966e-01 -1.20158720e+00 5.99013627e-01 7.95160413e-01 4.75307196e-01 -1.19553709e+00 -2.30206475e-01 3.25671166e-01 3.07564050e-01 -1.37646306e+00 1.95234194e-01 4.29505259e-01 -2.86670536e-01 -6.11861646e-01 -1.15779586e-01 2.36550812e-02 5.44914246e-01 -5.32952011e-01 1.74852490e+00 -6.86528385e-01 -7.39090204e-01 3.83730024e-01 -2.32113659e-01 -2.90855616e-01 -1.21120918e+00 -7.00999379e-01 -1.13325417e-01 -4.29870516e-01 1.61278307e-01 -6.07782245e-01 -5.90761185e-01 6.56966195e-02 -7.89949000e-01 3.34117472e-01 5.90724170e-01 8.99345994e-01 1.41023743e+00 3.89392585e-01 3.05089921e-01 -1.17919457e+00 1.10071339e-01 -4.56320971e-01 -1.10750139e+00 -2.17274815e-01 -2.66525120e-01 2.24662811e-01 7.70273507e-01 2.86538284e-02 -1.30912614e+00 -1.22674424e-02 -6.46396935e-01 -2.72505552e-01 -1.63771212e-03 -1.77378595e-01 -4.96682793e-01 -2.59978890e-01 9.07379329e-01 -9.98004675e-02 -4.39138830e-01 -7.20269280e-03 7.70804703e-01 2.76393652e-01 9.28040981e-01 -7.21462250e-01 8.03034186e-01 5.06545365e-01 5.54831743e-01 -3.79276782e-01 -6.14200175e-01 -2.09333724e-03 -6.92169741e-02 -1.73080668e-01 1.19309771e+00 -1.13831055e+00 -1.02357781e+00 5.42133927e-01 -8.26066554e-01 -5.09971023e-01 -1.34313977e+00 6.23380184e-01 -9.16627407e-01 -1.41815946e-01 -6.92084312e-01 -8.06482911e-01 -2.68012971e-01 -1.05542111e+00 1.34257209e+00 4.82881851e-02 1.51596427e-01 -6.07638538e-01 -1.37104973e-01 6.10850491e-02 4.98218060e-01 7.37553000e-01 7.25390315e-01 -2.15498671e-01 -1.38282299e+00 -2.96630293e-01 -3.18709195e-01 1.52775332e-01 -7.19656646e-01 -5.85353136e-01 -8.64224970e-01 -3.10232460e-01 4.67742503e-01 -6.04196191e-01 7.49146819e-01 6.21613026e-01 1.34145987e+00 1.62463456e-01 -4.91670400e-01 8.36655736e-01 1.67498147e+00 1.67085156e-01 1.04819071e+00 -4.52886194e-01 6.49645329e-01 -2.99251050e-01 6.02617919e-01 6.37517989e-01 1.12279750e-01 9.11764324e-01 1.87387928e-01 -2.11412892e-01 -6.54470921e-01 -6.36994421e-01 1.06405929e-01 7.18275130e-01 2.52848417e-01 -5.09383678e-01 -4.80016232e-01 7.59672001e-02 -1.50932789e+00 -1.39614534e+00 -3.76594633e-01 2.46238470e+00 6.23272777e-01 2.87339687e-01 -2.55035400e-01 -1.87541395e-01 5.63864052e-01 -4.78302352e-02 -6.09939218e-01 -6.12651646e-01 -4.10152525e-01 1.00120735e+00 9.72718716e-01 4.84792113e-01 -6.25240803e-01 5.34910083e-01 6.23989201e+00 1.28788745e+00 -7.02353597e-01 7.20823109e-01 7.59006679e-01 -4.43364650e-01 -7.66270578e-01 2.27629125e-01 -1.07725513e+00 7.90357351e-01 1.44480896e+00 1.20237572e-02 5.55827975e-01 5.54023743e-01 -2.29802683e-01 -5.52707314e-01 -1.18645847e+00 1.03420055e+00 -1.69349849e-01 -1.73058379e+00 -2.56316006e-01 8.36949274e-02 8.69437039e-01 2.19360068e-01 -1.20711714e-01 5.70924759e-01 8.65441144e-01 -1.12798524e+00 5.73299766e-01 8.74226093e-01 1.21809995e+00 -1.12121522e+00 4.76466417e-01 4.32892889e-01 -9.03830647e-01 4.78784204e-01 -3.96905571e-01 3.46165955e-01 7.89536774e-01 1.53034413e+00 -1.09485638e+00 8.04802716e-01 5.10606825e-01 -1.85570717e-01 -5.82098514e-02 3.95698935e-01 -3.96526605e-02 5.58655560e-01 -7.08694518e-01 2.01406442e-02 -7.49676302e-02 -1.46744311e-01 6.96331084e-01 1.14238465e+00 5.39941370e-01 2.30032995e-01 -4.84218709e-02 1.24938834e+00 -2.23637164e-01 -6.80481970e-01 -8.45172942e-01 2.64697932e-02 4.25807893e-01 1.14647818e+00 -7.11697996e-01 -3.42203856e-01 1.87031424e-03 9.57220197e-01 5.81588224e-02 -9.70171615e-02 -1.32638347e+00 3.09756428e-01 2.48803124e-01 4.42967772e-01 3.16688657e-01 -5.80251515e-02 -2.20327094e-01 -8.26805413e-01 -2.91353613e-02 -6.89833462e-01 2.54454553e-01 -1.12239897e+00 -1.43034530e+00 4.55411613e-01 1.94425181e-01 -8.24910820e-01 -8.42195094e-01 -2.73209661e-01 -2.07881153e-01 1.15921283e+00 -9.52754974e-01 -1.44961143e+00 -1.62258238e-01 3.72597486e-01 -7.75458217e-02 -2.12142244e-01 1.09490645e+00 4.32135701e-01 1.05602935e-01 4.22228783e-01 3.64321262e-01 -2.30178535e-01 3.40981573e-01 -1.35920680e+00 8.20419848e-01 6.55834854e-01 1.74037553e-02 -2.08777443e-01 9.23743546e-01 -1.02643764e+00 -1.80212367e+00 -1.34394288e+00 5.21599770e-01 -5.11177421e-01 4.80918467e-01 -8.84126484e-01 -4.10447866e-01 7.76906312e-01 2.65733957e-01 4.55423474e-01 1.85752824e-01 -1.66896507e-01 -9.46043059e-02 -2.28759915e-01 -1.70525455e+00 1.27902374e-01 1.02439916e+00 -9.97702003e-01 1.13807276e-01 8.99126589e-01 6.63231492e-01 -9.89745378e-01 -1.08660913e+00 7.24290431e-01 -1.78808942e-01 -1.25919127e+00 1.40961039e+00 6.74569011e-02 2.73755848e-01 -4.52110171e-01 -2.37100556e-01 -1.25269723e+00 -3.06434721e-01 -5.89432538e-01 -9.40044001e-02 1.15626895e+00 7.01794028e-02 -5.42281091e-01 6.45683348e-01 3.69258076e-01 -3.48797381e-01 -3.39748472e-01 -1.20792663e+00 -6.61126971e-01 9.84656736e-02 -8.40870798e-01 7.47099221e-01 4.78244394e-01 -6.36476398e-01 1.23958223e-01 -3.50136757e-01 4.43153203e-01 9.97556090e-01 5.75253591e-02 6.24118030e-01 -6.02982581e-01 -1.14204037e+00 2.21050736e-02 -7.01595426e-01 -4.71605271e-01 -3.14702451e-01 -1.25692642e+00 2.74609476e-02 -1.13480651e+00 2.54597694e-01 -7.34641612e-01 1.29817754e-01 -2.08383769e-01 4.66413587e-01 4.09171700e-01 -8.40113219e-03 -2.63565153e-01 -4.57305849e-01 3.96533310e-01 1.14780521e+00 1.02601023e-02 5.58666706e-01 4.14317613e-03 -1.15002833e-01 3.04409355e-01 6.91934884e-01 -5.16635299e-01 -2.84391195e-01 1.11560233e-01 6.18925571e-01 8.21674585e-01 1.16804731e+00 -1.47079933e+00 3.38900566e-01 4.50839311e-01 8.48560989e-01 -1.09306312e+00 7.98230648e-01 -4.00920779e-01 1.26384127e+00 7.23997653e-02 1.67003386e-02 -1.45674020e-01 4.02059525e-01 4.73570228e-01 -1.99005231e-01 -2.40074784e-01 9.74077165e-01 -4.90068287e-01 -4.68198478e-01 2.13408440e-01 5.17319776e-02 2.48675078e-01 1.23270392e+00 4.92952466e-01 -2.28431225e-01 -1.06015950e-01 -7.18887091e-01 -3.15118730e-01 4.99121547e-01 -2.62083083e-01 2.79782057e-01 -1.63887084e+00 -7.92368293e-01 3.39851618e-01 -1.56494185e-01 3.20648551e-01 8.24416935e-01 3.29388559e-01 -6.32661164e-01 3.11560780e-01 -4.92427684e-02 -7.72724092e-01 -6.47785127e-01 6.09876692e-01 3.72187883e-01 -7.15099096e-01 -1.00501418e+00 7.58746624e-01 5.70860766e-02 -2.83273965e-01 -3.50872397e-01 -1.55294895e-01 8.94862831e-01 -4.57082301e-01 5.07026672e-01 1.17159961e-02 4.41104233e-01 -4.57076967e-01 -2.18616594e-02 -1.10108464e-03 2.19802424e-01 -2.45347321e-01 9.96829569e-01 3.93892765e-01 5.37470318e-02 3.78909022e-01 6.91708505e-01 5.19371852e-02 -1.20504272e+00 2.14663409e-02 -7.82440722e-01 -2.30289415e-01 2.98476487e-01 -1.36817753e+00 -8.65732729e-01 6.58150375e-01 6.42171860e-01 2.11346924e-01 1.04738784e+00 3.29523385e-01 9.69591200e-01 -3.08633614e-02 9.80375469e-01 -8.03623438e-01 -3.12380940e-01 3.19506794e-01 7.40061164e-01 -9.39501286e-01 1.16521426e-01 -1.24488048e-01 -6.30996764e-01 4.83919740e-01 2.46565163e-01 -8.21454749e-02 5.12788236e-01 1.21761858e+00 -5.38791239e-01 -4.69786793e-01 -6.65944993e-01 -8.49376619e-02 -1.35294229e-01 6.55543387e-01 1.12431534e-02 6.45201206e-01 2.11642794e-02 4.17427480e-01 -3.14539015e-01 -2.04839185e-02 2.03312859e-01 6.63670421e-01 1.16472021e-01 -1.33756459e+00 -4.51114506e-01 4.73511249e-01 -5.29334694e-02 -2.10246101e-01 2.55108863e-01 8.42144787e-01 2.14364380e-01 3.42536122e-01 4.39491898e-01 -1.24293715e-01 2.19182238e-01 4.61801626e-02 1.11307168e+00 -3.88562828e-01 -8.70845854e-01 -1.58766925e-01 1.23000838e-01 -7.75236607e-01 -2.73938984e-01 -7.04284251e-01 -1.52775335e+00 -8.43211174e-01 1.94918662e-02 -2.71675992e-04 1.10609972e+00 6.28056884e-01 4.50302541e-01 1.09683454e+00 4.97857720e-01 -9.17686939e-01 -6.27532840e-01 -6.97193086e-01 -5.96252441e-01 4.45897877e-01 -2.52546281e-01 -4.03231323e-01 -9.22691599e-02 -5.50748289e-01]
[9.258605003356934, -3.5985426902770996]
f65991bb-ade2-4b4a-a3a5-0c05405804fb
stanford-mlab-at-semeval-2021-task-1-tree
null
null
https://aclanthology.org/2021.semeval-1.89
https://aclanthology.org/2021.semeval-1.89.pdf
Stanford MLab at SemEval-2021 Task 1: Tree-Based Modelling of Lexical Complexity using Word Embeddings
This paper presents our system for the single- and multi-word lexical complexity prediction tasks of SemEval Task 1: Lexical Complexity Prediction. Text comprehension depends on the reader{'}s ability to understand the words present in it; evaluating the lexical complexity of such texts can enable readers to find an appropriate text and systems to tailor a text to an audience{'}s needs. We present our model pipeline, which applies a combination of embedding-based and manual features to predict lexical complexity on the CompLex English dataset using various tree-based and linear models. Our method is ranked 27 / 54 on single-word prediction and 14 / 37 on multi-word prediction.
['Ethan A. Chi', 'Jillian Tang', 'Zander Lack', 'Patrick Liu', 'Kevin Liu', 'Kathy J. Lee', 'Enok Choe', 'Gordon Chi', 'Niveditha Iyer', 'Erik Rozi']
2021-08-01
null
null
null
semeval-2021
['lexical-complexity-prediction']
['natural-language-processing']
[-9.07423869e-02 -1.02565589e-03 -2.79245913e-01 -2.48654783e-01 -7.47713029e-01 -6.79906130e-01 2.15371132e-01 1.14913070e+00 -9.46125448e-01 -4.63197306e-02 8.44957590e-01 -9.22281742e-01 -1.44390136e-01 -6.10898197e-01 -5.17624244e-02 4.69819516e-01 2.79889375e-01 5.48729777e-01 1.22794226e-01 -5.75402558e-01 8.13551605e-01 -1.72555335e-02 -1.58306575e+00 5.62784910e-01 8.42649221e-01 7.10593224e-01 7.34504938e-01 1.37194145e+00 -6.71710849e-01 8.34089935e-01 -4.04460162e-01 -3.22202474e-01 -1.79617271e-01 -2.62067348e-01 -1.11252427e+00 -4.44630355e-01 4.01339352e-01 -1.38507653e-02 -9.50789172e-03 5.77184439e-01 4.35234904e-01 1.77537575e-02 9.10571456e-01 -3.76219720e-01 -3.67776513e-01 1.15514004e+00 2.44158730e-01 8.07191074e-01 9.71399248e-01 1.20925725e-01 1.58351302e+00 -1.13891864e+00 5.31179190e-01 1.21154773e+00 7.45461166e-01 1.88988701e-01 -1.14536107e+00 -4.58620727e-01 2.48860404e-01 5.97049236e-01 -1.13900459e+00 -5.20348489e-01 3.31688881e-01 -7.93512702e-01 1.98492515e+00 1.64912760e-01 9.15266156e-01 9.55398917e-01 4.88792151e-01 7.80835330e-01 1.00502264e+00 -1.03115082e+00 -5.09670638e-02 -1.23454407e-01 8.97306025e-01 8.53619456e-01 7.86526427e-02 -3.37565809e-01 -8.30981970e-01 9.91211161e-02 -7.88794532e-02 -5.62811852e-01 2.83463094e-02 5.48287809e-01 -1.07748199e+00 1.00889277e+00 -3.77776980e-01 1.77475616e-01 -1.47793353e-01 -4.39328671e-01 5.92265308e-01 5.70065379e-01 3.95405114e-01 9.28444326e-01 -9.85224366e-01 -6.56284988e-01 -8.62710297e-01 5.18970370e-01 1.34443414e+00 1.02035439e+00 4.90696251e-01 -2.73074955e-01 -1.19605266e-01 9.62611079e-01 4.12759364e-01 3.59226793e-01 8.43288958e-01 -2.65228599e-01 7.01489091e-01 8.45165253e-01 -3.96095425e-01 -6.13774776e-01 -1.02717102e+00 -2.05251306e-01 -1.82887614e-01 -2.58464336e-01 2.34977499e-01 8.95270854e-02 -6.53362215e-01 1.23930240e+00 -2.57963508e-01 -6.93487346e-01 -1.48975492e-01 1.42240405e-01 1.22659540e+00 8.84639680e-01 6.36264324e-01 -3.66427958e-01 1.62491953e+00 -8.66798043e-01 -7.98891425e-01 -4.87498999e-01 1.37347627e+00 -1.24285960e+00 1.73423314e+00 2.87469596e-01 -1.24543309e+00 -5.99351227e-01 -1.13923931e+00 -5.18401742e-01 -6.52586102e-01 -1.96088001e-01 1.02153569e-01 6.64729655e-01 -8.21635902e-01 4.55673397e-01 -4.50039715e-01 -3.54996383e-01 -6.84625953e-02 1.15333892e-01 -3.34633626e-02 -1.12264313e-01 -1.45744002e+00 1.45624268e+00 5.91887593e-01 -6.00411773e-01 -3.01566750e-01 -8.89693797e-01 -1.08812582e+00 2.76400834e-01 1.59283094e-02 -7.09324598e-01 1.36462355e+00 -3.85843188e-01 -1.27700675e+00 7.39319623e-01 -3.48899513e-01 -1.85881946e-02 1.04376604e-03 -1.93488047e-01 -4.30820674e-01 -3.55851322e-01 1.28559217e-01 2.77441949e-01 5.57491302e-01 -2.04277843e-01 -9.20725882e-01 -1.29080996e-01 -2.15648338e-01 7.19554365e-01 -6.61035657e-01 3.03026855e-01 -1.78183883e-01 -7.16079473e-01 -9.37254727e-02 -5.98659992e-01 5.34650646e-02 -5.05121887e-01 -2.70671636e-01 -7.56934524e-01 2.75457680e-01 -1.05345833e+00 2.20655322e+00 -1.65758383e+00 4.51753199e-01 8.25350285e-02 4.92489398e-01 -7.52855046e-03 -4.00725663e-01 8.08373392e-01 4.10987288e-02 6.20711863e-01 4.24234629e-01 -3.13172996e-01 3.23403865e-01 -3.33822489e-01 2.34069660e-01 -1.91084385e-01 1.27445891e-01 9.59235668e-01 -6.22927129e-01 -7.52561629e-01 2.37763181e-01 -8.75677466e-02 -7.97030270e-01 4.53370661e-01 -4.03954953e-01 -3.35117579e-01 -1.68112174e-01 3.99003804e-01 4.32824343e-02 -1.45376191e-01 8.05943087e-02 -1.41587779e-01 -3.48141104e-01 1.02859473e+00 -8.17648768e-01 1.30576861e+00 -1.06471205e+00 6.52412355e-01 -5.07996082e-01 -1.81425050e-01 3.74232948e-01 3.23603628e-04 -9.74691734e-02 -9.47063565e-01 2.22300887e-01 1.17549606e-01 4.45710063e-01 -7.47425675e-01 6.43976629e-01 8.06182846e-02 -4.15267080e-01 8.70290339e-01 -6.91635236e-02 -4.02673274e-01 6.42413318e-01 2.13410050e-01 1.42969036e+00 -2.55226851e-01 7.57626116e-01 -6.69968665e-01 4.24566567e-01 -1.82976499e-01 -4.92211580e-02 5.29507875e-01 6.13527149e-02 -6.64954185e-02 2.79368758e-01 -5.68133056e-01 -1.35886145e+00 -7.19254553e-01 -7.67019242e-02 2.05976772e+00 -5.33706248e-01 -1.31535089e+00 -6.05770051e-01 -2.92144686e-01 -5.04627675e-02 1.00199771e+00 -4.22794253e-01 -1.22915730e-01 -5.13224483e-01 -2.11276382e-01 4.63294178e-01 5.53705275e-01 -1.41476512e-01 -1.16625798e+00 -5.59625328e-01 4.70028222e-01 -5.65429986e-01 -1.01569140e+00 -7.73107052e-01 4.65786517e-01 -5.37518144e-01 -8.44676316e-01 2.44622622e-02 -1.12093890e+00 7.17336014e-02 -9.40926149e-02 1.63317072e+00 2.57022679e-01 -4.73007530e-01 3.56274515e-01 -7.31286526e-01 -6.35389447e-01 -7.58268833e-01 7.45202899e-01 4.60527986e-02 -1.10511875e+00 7.76476026e-01 -1.57720372e-02 -3.71849090e-01 4.35789451e-02 -7.15116620e-01 2.57228047e-01 6.29389405e-01 6.79030299e-01 3.80117089e-01 1.26158521e-01 2.10818991e-01 -6.54241204e-01 1.44212449e+00 -4.03709233e-01 -2.86609441e-01 3.68205488e-01 -1.02946496e+00 2.64025837e-01 6.26710296e-01 -5.85060656e-01 -3.86990994e-01 -3.07720248e-02 -6.62442744e-01 5.01608729e-01 1.77505553e-01 1.06447256e+00 -1.47558913e-01 3.11806172e-01 8.51741314e-01 1.88162804e-01 -3.71338397e-01 -4.21454877e-01 5.57644606e-01 7.77957261e-01 -8.10662583e-02 -4.78659958e-01 4.88938004e-01 -7.80489206e-01 -3.81496221e-01 -1.29274368e+00 -1.06734896e+00 -6.68830335e-01 -8.22274268e-01 -2.07768008e-02 9.97832716e-01 -7.95468450e-01 -6.92682624e-01 2.36612886e-01 -1.16061270e+00 -3.80835086e-01 -8.10307637e-03 2.72249311e-01 -3.15358460e-01 6.29658639e-01 -6.48030102e-01 -4.99448776e-01 -6.67341232e-01 -1.05594516e+00 8.99627626e-01 -1.73981667e-01 -1.30177200e+00 -1.32728326e+00 6.00256724e-03 4.77273971e-01 5.89793086e-01 -5.34969449e-01 1.94545734e+00 -9.88149583e-01 -6.91913739e-02 -2.36891925e-01 6.43658862e-02 1.80459797e-01 -1.29522219e-01 7.99094364e-02 -4.52948213e-01 -1.17230661e-01 -3.20966423e-01 -5.60916781e-01 5.18418610e-01 3.47018510e-01 1.12978721e+00 -4.30185914e-01 1.72324076e-01 2.97740549e-01 1.09510922e+00 -3.88744861e-01 1.94816291e-01 6.27803028e-01 7.44812250e-01 5.73970377e-01 2.66931653e-01 6.31295502e-01 1.00169623e+00 4.81695682e-01 -1.19830720e-01 3.20423394e-01 -1.16458170e-01 -4.33220267e-01 5.51475346e-01 1.73711312e+00 4.78744686e-01 -2.84204394e-01 -1.33708227e+00 2.36747384e-01 -1.17705250e+00 -6.25783086e-01 -3.26440006e-01 1.71686685e+00 1.12650180e+00 7.22110450e-01 2.95294046e-01 4.03130352e-01 1.83443472e-01 6.31106794e-02 -3.03916752e-01 -1.02437985e+00 -1.03037052e-01 5.76903880e-01 3.02571505e-01 9.36939776e-01 -7.00228155e-01 1.28114688e+00 7.16157913e+00 1.11962354e+00 -5.73981524e-01 -5.42510636e-02 3.08177859e-01 1.07841440e-01 -5.59433222e-01 -4.29930501e-02 -1.26490307e+00 3.68775874e-01 1.56686783e+00 -8.02191854e-01 5.53580821e-01 5.79489589e-01 2.39938840e-01 -2.78232485e-01 -1.41321087e+00 8.51853609e-01 1.38870791e-01 -1.17487335e+00 4.09586489e-01 -1.89615935e-01 1.24480836e-01 -1.24820776e-01 -1.11852698e-01 7.75797069e-01 6.49659261e-02 -1.36661482e+00 6.42109632e-01 4.47306067e-01 8.51631761e-01 -8.12494338e-01 5.49328208e-01 7.93137789e-01 -1.39741206e+00 -1.09474614e-01 -2.07057640e-01 -6.12531841e-01 -5.62670715e-02 3.51392061e-01 -8.28977346e-01 -3.57136965e-01 4.65780139e-01 2.81993628e-01 -1.05290234e+00 4.58054513e-01 -1.53488547e-01 7.00892866e-01 -1.98058292e-01 -8.88768077e-01 -5.87597452e-02 3.53893459e-01 3.76443058e-01 1.84009254e+00 -6.57110959e-02 2.34836385e-01 3.21718007e-01 3.05804461e-01 3.42041894e-04 1.08271730e+00 -1.82977363e-01 -5.08044064e-01 7.69929290e-01 1.06147170e+00 -5.34770906e-01 -2.28464022e-01 -5.32629728e-01 6.99039519e-01 5.64030468e-01 -3.99603933e-01 -2.57663578e-01 -8.04371178e-01 4.07680541e-01 3.20438743e-01 -2.05565616e-02 -5.55276573e-01 -7.17926264e-01 -1.12564528e+00 -1.05568366e-02 -1.15369511e+00 5.07447064e-01 -4.43319738e-01 -1.33144546e+00 6.80206180e-01 -1.60773903e-01 -6.32452846e-01 -2.28560895e-01 -1.03787100e+00 -4.85739887e-01 1.21059620e+00 -1.36805046e+00 -1.02874827e+00 -1.36513293e-01 2.61675715e-01 1.33081853e+00 -4.22629923e-01 1.25450015e+00 -1.11703500e-01 -3.75997216e-01 7.25754261e-01 -2.76069909e-01 -1.30549431e-01 4.64357853e-01 -1.52147961e+00 9.11031187e-01 4.78854924e-01 -7.61403218e-02 5.62125742e-01 6.46164060e-01 -7.59519875e-01 -1.35218561e+00 -7.00882077e-01 1.83559978e+00 -1.10753310e+00 1.12785769e+00 -4.50520754e-01 -7.49516189e-01 3.46836507e-01 2.01426089e-01 -9.97355223e-01 1.36441350e+00 7.42964387e-01 -2.63559639e-01 4.54966664e-01 -4.86775309e-01 7.72741139e-01 9.46319938e-01 -7.72070229e-01 -1.10805130e+00 3.97435039e-01 9.53806162e-01 -2.80400187e-01 -1.19144225e+00 8.30379203e-02 7.07424402e-01 -5.11402428e-01 4.89899337e-01 -7.47689486e-01 8.30915570e-01 3.85303080e-01 -2.29282320e-01 -1.51125336e+00 -8.78819466e-01 -4.75418866e-01 -1.40833512e-01 5.56225300e-01 9.28316236e-01 -5.22727743e-02 5.74932575e-01 6.10247970e-01 -8.74303430e-02 -7.90090978e-01 -7.55535066e-01 -3.90190303e-01 6.34556592e-01 -9.07684505e-01 3.43896240e-01 6.93046629e-01 7.03231215e-01 1.13417792e+00 3.05976659e-01 -1.76037341e-01 2.34176949e-01 -4.90144312e-01 3.93915862e-01 -1.38657749e+00 -1.96749270e-01 -1.08186412e+00 -3.35810393e-01 -1.07265043e+00 4.94450688e-01 -1.17446744e+00 -3.09253663e-01 -1.57087564e+00 2.34399334e-01 -7.32522905e-02 9.75142345e-02 2.98651755e-01 -5.33041179e-01 -4.27393466e-01 1.85240194e-01 8.18944424e-02 -6.80923223e-01 3.39457095e-01 8.49804580e-01 -6.46239966e-02 -2.03660920e-01 3.96279357e-02 -8.29113245e-01 7.54314363e-01 6.12926006e-01 -2.39024580e-01 -2.35853314e-01 -5.00582755e-01 1.03551245e+00 -1.70508310e-01 -6.22145474e-01 -9.63156819e-01 3.04467291e-01 -2.11868450e-01 5.52319884e-01 -7.06803679e-01 -1.52283773e-01 -5.59455276e-01 -4.55910802e-01 5.53824246e-01 -7.79188812e-01 9.52237070e-01 5.05608737e-01 7.42796808e-02 2.13285089e-01 -5.72222352e-01 7.10885465e-01 -1.98483542e-01 -5.92267692e-01 2.61194762e-02 -1.18698144e+00 5.28505623e-01 5.49086094e-01 -1.18752651e-01 -2.11464316e-01 -4.87558633e-01 -8.38306963e-01 4.67307925e-01 2.48081043e-01 7.47391164e-01 8.43291700e-01 -9.20681357e-01 -9.52477753e-01 2.94248253e-01 6.45054102e-01 -4.35395956e-01 -3.60872149e-02 2.30228066e-01 -8.24837148e-01 5.56395054e-01 -6.49579754e-03 -1.90875754e-01 -1.51459634e+00 3.89374048e-01 7.41262734e-02 -5.67656338e-01 -3.18483055e-01 9.19549644e-01 -4.41095084e-01 -4.10244852e-01 2.42632404e-01 -6.31230175e-01 -5.17818570e-01 3.65424901e-01 7.58520961e-01 6.45669580e-01 4.03767705e-01 -5.35896480e-01 -9.62624773e-02 6.30114257e-01 -6.26812100e-01 -1.91910207e-01 1.17569149e+00 -3.36109012e-01 -7.06509948e-02 7.59771526e-01 1.45990443e+00 5.12516275e-02 -3.21055323e-01 -3.08337212e-01 6.77462518e-01 1.19737973e-02 2.89876401e-01 -1.07867408e+00 1.20892935e-01 8.13569903e-01 5.56660235e-01 1.52972475e-01 8.46468627e-01 -6.54501840e-02 8.50746632e-01 9.95094717e-01 -1.90915883e-01 -1.80712056e+00 4.22557384e-01 1.43429410e+00 8.15245509e-01 -1.07217741e+00 1.85234010e-01 -1.85241312e-01 -5.98338485e-01 1.53831387e+00 5.40573657e-01 4.08810049e-01 1.05439341e+00 3.24359685e-01 1.02259785e-01 -2.15546697e-01 -1.29447889e+00 -2.75196999e-01 6.47309124e-01 2.30228320e-01 9.51420903e-01 1.87102228e-01 -6.10885859e-01 7.96927154e-01 -9.52589154e-01 -5.20694613e-01 5.95322013e-01 9.33846951e-01 -1.04493749e+00 -1.28098094e+00 1.14416093e-01 1.04717374e+00 -3.84252548e-01 -9.00522172e-01 -5.32829404e-01 3.73678267e-01 -1.12710178e-01 1.00817347e+00 -8.60821083e-02 -5.81035435e-01 7.59265721e-01 5.94140470e-01 2.21838802e-01 -1.24556184e+00 -7.67014086e-01 -4.46443999e-04 4.36596721e-01 -3.48366290e-01 4.00662094e-01 -7.62551963e-01 -1.10205019e+00 -5.60012281e-01 -3.07187170e-01 -1.50718898e-01 9.08851624e-01 9.74481046e-01 7.69547839e-03 3.87183696e-01 4.87218916e-01 -5.18437862e-01 -6.36028171e-01 -1.48360026e+00 -2.04905406e-01 2.37833738e-01 -5.97477034e-02 -1.55717552e-01 -4.56402510e-01 9.63042751e-02]
[10.718132972717285, 10.425771713256836]
51c76c94-30e2-404e-9153-572dd3ec96cf
a-language-model-based-generative-classifier
null
null
https://aclanthology.org/2021.emnlp-main.188
https://aclanthology.org/2021.emnlp-main.188.pdf
A Language Model-based Generative Classifier for Sentence-level Discourse Parsing
Discourse segmentation and sentence-level discourse parsing play important roles for various NLP tasks to consider textual coherence. Despite recent achievements in both tasks, there is still room for improvement due to the scarcity of labeled data. To solve the problem, we propose a language model-based generative classifier (LMGC) for using more information from labels by treating the labels as an input while enhancing label representations by embedding descriptions for each label. Moreover, since this enables LMGC to make ready the representations for labels, unseen in the pre-training step, we can effectively use a pre-trained language model in LMGC. Experimental results on the RST-DT dataset show that our LMGC achieved the state-of-the-art F1 score of 96.72 in discourse segmentation. It further achieved the state-of-the-art relation F1 scores of 84.69 with gold EDU boundaries and 81.18 with automatically segmented boundaries, respectively, in sentence-level discourse parsing.
['Manabu Okumura', 'Hidetaka Kamigaito', 'Ying Zhang']
null
null
null
null
emnlp-2021-11
['discourse-segmentation', 'discourse-parsing']
['natural-language-processing', 'natural-language-processing']
[ 3.23347837e-01 9.80565250e-01 -3.11666757e-01 -4.78698492e-01 -1.24211049e+00 -5.23527384e-01 8.19863856e-01 1.77061617e-01 -3.39796811e-01 9.98833418e-01 5.07861376e-01 -5.11432707e-01 4.30868179e-01 -8.34422708e-01 -5.64792454e-01 -6.46956623e-01 1.30210429e-01 6.59034669e-01 2.93050885e-01 -9.79059413e-02 1.29774496e-01 -1.85268983e-01 -1.27867806e+00 7.65870452e-01 1.18221354e+00 5.53578496e-01 2.43635699e-01 6.98586643e-01 -3.68722200e-01 1.31917965e+00 -9.51351047e-01 -4.40683454e-01 -4.77209061e-01 -8.49705398e-01 -1.33896279e+00 3.96278650e-01 5.41203544e-02 -6.07979950e-03 -1.77737191e-01 9.30763841e-01 2.29323685e-01 8.89205337e-02 8.48637104e-01 -8.61613393e-01 -6.76124036e-01 1.07899058e+00 -4.65295047e-01 1.33048505e-01 4.62134808e-01 -3.00010979e-01 1.36101091e+00 -5.77562153e-01 1.00848877e+00 1.40456879e+00 1.48560882e-01 7.86143005e-01 -1.13434947e+00 -1.51560351e-01 3.30518126e-01 1.36827603e-01 -9.23845530e-01 -2.37249702e-01 6.87864661e-01 -5.93854725e-01 1.08927977e+00 3.30615491e-01 2.50514150e-01 1.12331033e+00 -1.44268796e-01 1.04469776e+00 1.27311409e+00 -8.60180199e-01 9.89875048e-02 1.50966033e-01 5.05075455e-01 5.53113282e-01 -8.15133750e-02 -4.14384663e-01 -2.98723251e-01 2.13189095e-01 5.52368402e-01 -7.16581225e-01 -2.69147694e-01 4.28614229e-01 -9.63668048e-01 1.08669734e+00 3.25444549e-01 7.33768940e-01 -1.22816078e-01 -1.76376015e-01 4.51312244e-01 6.96059540e-02 8.04214716e-01 5.79415560e-01 -1.31028593e-01 -1.44737735e-01 -9.04015541e-01 1.38463274e-01 1.01307821e+00 1.03646612e+00 6.22511566e-01 -1.36500984e-01 -6.68658614e-01 8.46502006e-01 2.17682809e-01 1.99985921e-01 4.74780738e-01 -1.05950236e+00 8.94306540e-01 9.23489809e-01 -1.75472409e-01 -7.74612904e-01 -5.18492639e-01 -2.59992391e-01 -8.55820000e-01 -3.63614172e-01 4.97054130e-01 -2.08212823e-01 -8.02747011e-01 1.85617816e+00 2.97082812e-01 -1.22552484e-01 5.18059075e-01 6.33153379e-01 1.24827647e+00 1.02125347e+00 2.67624259e-01 -4.78591621e-01 1.54237628e+00 -1.42367744e+00 -1.13933134e+00 -4.74474013e-01 1.13706803e+00 -7.31447458e-01 1.06549156e+00 4.55413312e-02 -9.80539322e-01 -6.83236957e-01 -8.90338838e-01 -3.41511875e-01 -6.28822669e-02 3.98337692e-01 4.07801300e-01 5.63875496e-01 -1.05651021e+00 3.55105728e-01 -8.55832636e-01 -1.23514868e-01 2.73622990e-01 1.59862801e-01 -4.30952720e-02 -1.20309936e-02 -1.34332466e+00 8.69874001e-01 6.09205782e-01 -4.64375801e-02 -7.20965922e-01 -2.40048945e-01 -1.03258240e+00 2.61189640e-02 4.93822396e-01 -3.82991254e-01 1.34978628e+00 -5.49371839e-01 -1.61989129e+00 1.23951483e+00 -4.84388411e-01 -4.01287884e-01 3.94459456e-01 -3.77825588e-01 -1.66900650e-01 2.01636419e-01 4.61883694e-01 7.98400283e-01 2.56101519e-01 -1.41053355e+00 -6.13001823e-01 -9.28003713e-02 2.98720151e-01 2.39861786e-01 -1.42565683e-01 7.75869489e-02 -2.96905518e-01 -3.75284046e-01 2.46378422e-01 -9.43025529e-01 -1.71297058e-01 -9.48874593e-01 -8.26732278e-01 -8.89849603e-01 6.33492529e-01 -1.09826243e+00 1.46588349e+00 -1.87688351e+00 2.63257742e-01 -3.97600651e-01 1.44178152e-01 4.24179852e-01 -1.04913013e-02 4.16541219e-01 1.20591372e-01 4.32954729e-01 -3.50372791e-01 -6.12613857e-01 -1.50180072e-01 2.89311260e-01 -1.42033994e-01 4.81200106e-02 7.45287836e-01 1.01971483e+00 -1.04128957e+00 -8.33360314e-01 4.93133813e-02 2.63512135e-01 -3.23747575e-01 6.32403016e-01 -5.57940483e-01 8.54622424e-01 -6.16164327e-01 2.82831550e-01 3.44668835e-01 -5.39428174e-01 6.11066401e-01 2.63164639e-01 -1.16080582e-01 8.18091214e-01 -7.76729405e-01 1.59145319e+00 -3.41798455e-01 6.88247383e-01 -1.36117220e-01 -1.11290205e+00 1.03612745e+00 6.39005005e-01 1.04426675e-01 -4.76702899e-01 3.03716451e-01 2.69386202e-01 1.40721038e-01 -8.37986648e-01 6.14973485e-01 -1.42496690e-01 -4.65165794e-01 5.04629791e-01 1.14141501e-01 -1.14923917e-01 5.45576394e-01 2.93006808e-01 8.50127161e-01 2.86306530e-01 1.71865866e-01 -1.31900176e-01 8.23925912e-01 2.04383731e-01 5.83260059e-01 3.55265468e-01 1.30803257e-01 7.16461957e-01 8.38484645e-01 -4.92359214e-02 -7.83447564e-01 -4.75261569e-01 -6.48369640e-02 1.17796791e+00 -1.31526798e-01 -4.88371730e-01 -1.20470953e+00 -8.41000497e-01 -5.41663945e-01 1.09709954e+00 -5.62974274e-01 2.56250292e-01 -1.19751894e+00 -7.48905897e-01 5.22545457e-01 5.20619094e-01 6.47676706e-01 -1.32761395e+00 -3.57382923e-01 4.10692304e-01 -7.13320315e-01 -1.47387290e+00 8.98203254e-02 3.48493941e-02 -7.58819103e-01 -1.07938313e+00 -6.26011610e-01 -1.05463088e+00 6.45521522e-01 -1.79561079e-01 1.14806747e+00 1.33730233e-01 2.53631890e-01 -2.32044458e-01 -6.66592896e-01 -2.92135291e-02 -1.02752316e+00 5.43496609e-01 -6.42377377e-01 -3.81489366e-01 1.68527648e-01 -8.79669935e-02 -7.68383518e-02 -6.86338395e-02 -7.02829123e-01 3.11886758e-01 2.02449143e-01 1.02252078e+00 4.56984699e-01 -2.40299076e-01 9.35266376e-01 -1.50588524e+00 7.76771069e-01 -3.27536196e-01 -2.76657909e-01 5.17104924e-01 -3.64315987e-01 1.01300068e-02 4.60546046e-01 -2.33922631e-01 -1.50328696e+00 -3.53014082e-01 -4.22168225e-01 3.80072743e-01 -2.28179350e-01 6.21147096e-01 -3.43099236e-01 9.10462379e-01 5.30904233e-01 -2.24115357e-01 -3.76138180e-01 -4.37464088e-01 5.95510364e-01 9.76015091e-01 4.60979849e-01 -6.83947504e-01 1.55240014e-01 -1.44351706e-01 -2.02507526e-01 -8.26147735e-01 -1.48059762e+00 -4.05360907e-01 -8.63461435e-01 -4.98422720e-02 1.41250193e+00 -9.21582699e-01 -2.90769279e-01 3.23551744e-01 -1.50613809e+00 -4.07543868e-01 -1.78547874e-01 2.18218401e-01 -4.76065040e-01 4.08452094e-01 -9.48088586e-01 -9.81396079e-01 -3.08714479e-01 -1.22381234e+00 1.03166783e+00 4.81050462e-01 -5.06280839e-01 -1.26578677e+00 6.47126287e-02 7.91488349e-01 -4.10511084e-02 4.95795131e-01 1.22680223e+00 -9.45103943e-01 -5.34755409e-01 2.46082600e-02 -2.34126866e-01 3.65441769e-01 2.05009535e-01 -1.31350338e-01 -1.36606479e+00 -1.67216331e-01 1.42046109e-01 -6.03060305e-01 8.77351999e-01 3.21751416e-01 7.06239581e-01 -3.03053796e-01 -4.73397970e-01 4.45300080e-02 1.03877854e+00 2.55786449e-01 6.70008421e-01 4.02328581e-01 7.07421958e-01 9.21452224e-01 7.47676790e-01 6.58868849e-02 6.02339923e-01 5.64248621e-01 1.16971716e-01 -1.33786246e-01 -5.00095844e-01 -3.12121809e-01 2.10233435e-01 1.25188160e+00 -8.36126283e-02 -5.97550392e-01 -9.08576965e-01 5.60108662e-01 -1.97733808e+00 -6.67611241e-01 -6.19570971e-01 1.75565839e+00 1.34627891e+00 1.78425685e-01 -7.48654529e-02 1.81004956e-01 9.16388512e-01 3.48958164e-01 -7.57735074e-02 -4.81446922e-01 -2.72783667e-01 1.40233114e-01 -1.13224216e-01 7.89319754e-01 -1.34607863e+00 1.47012269e+00 5.51856804e+00 8.68295610e-01 -8.53868186e-01 2.99384326e-01 1.09557271e+00 5.58592558e-01 -3.79842728e-01 1.14436880e-01 -1.09157014e+00 2.67473340e-01 1.29067409e+00 5.59035353e-02 -6.78628609e-02 7.36787498e-01 4.40793969e-02 -4.12635773e-01 -1.14292908e+00 4.33959723e-01 9.88791287e-02 -1.29530561e+00 -1.55604213e-01 -1.49683086e-02 1.14563882e+00 -3.78193140e-01 -2.69864202e-01 6.41015589e-01 3.63362938e-01 -1.17132676e+00 4.68489289e-01 9.71649438e-02 8.04558635e-01 -4.35422808e-01 1.02952659e+00 7.39593804e-01 -7.84176469e-01 3.76178622e-01 -3.66391629e-01 -3.22208375e-01 4.30802703e-01 5.28574526e-01 -1.31742156e+00 7.96515405e-01 1.31757423e-01 4.04862374e-01 -4.56875354e-01 5.41717172e-01 -1.02218652e+00 1.02126324e+00 -3.25605087e-02 -1.37034073e-01 4.18334603e-01 -1.25373706e-01 1.83862478e-01 1.36643445e+00 1.16528524e-02 1.34945631e-01 4.91809994e-01 8.02943826e-01 -2.80736327e-01 1.69675082e-01 -2.97529966e-01 -2.32455656e-01 5.74722350e-01 1.08290339e+00 -9.20515299e-01 -5.44739068e-01 -1.48823544e-01 7.37509072e-01 5.95948279e-01 2.91222006e-01 -6.71274483e-01 -8.79863575e-02 -6.58892766e-02 -1.69000223e-01 -4.89978008e-02 -1.84071884e-01 -4.39167768e-01 -9.14110303e-01 -9.41925347e-02 -7.77782738e-01 3.02710235e-01 -2.73450673e-01 -1.02251911e+00 9.10041749e-01 1.31757453e-01 -7.37580836e-01 -6.93344593e-01 -4.71825570e-01 -6.71871006e-01 8.44010949e-01 -1.55868673e+00 -1.32732558e+00 -4.03389484e-02 -1.07155055e-01 9.32337582e-01 7.65028000e-02 1.08539224e+00 7.57352635e-03 -6.78839266e-01 4.10847485e-01 -1.15145044e-02 4.98013079e-01 5.98092198e-01 -1.31703186e+00 3.05928379e-01 7.51905859e-01 2.56183654e-01 2.72143543e-01 4.86050069e-01 -7.35510647e-01 -5.62410653e-01 -9.88596380e-01 1.61678195e+00 -2.80183882e-01 5.67290902e-01 -2.92234898e-01 -1.08578956e+00 7.12963402e-01 6.80025518e-01 -6.71310663e-01 7.72453904e-01 4.02745605e-01 3.77186574e-02 6.26764774e-01 -8.71261001e-01 3.49698216e-01 8.16746891e-01 -4.43320841e-01 -1.01615310e+00 6.79845870e-01 1.01724732e+00 -6.95669949e-01 -9.28297341e-01 3.27586949e-01 -1.76605925e-01 -6.67103767e-01 5.05819201e-01 -5.15546799e-01 9.01658058e-01 -7.23451376e-02 -1.19274281e-01 -1.11325109e+00 -1.22856528e-01 -5.05887210e-01 5.40419258e-02 1.85942495e+00 8.27924967e-01 -2.95488596e-01 7.12619185e-01 6.49807692e-01 -4.75168914e-01 -9.13404822e-01 -9.44376767e-01 -5.26109993e-01 3.94517541e-01 1.48538146e-02 3.59975338e-01 9.54769552e-01 1.93058327e-01 1.00662553e+00 -1.64345056e-01 -7.04129860e-02 1.15986928e-01 3.03491414e-01 5.35319686e-01 -1.10704613e+00 -1.65962353e-01 -2.01994866e-01 2.77444482e-01 -1.18903673e+00 6.30243599e-01 -1.01212883e+00 1.76397905e-01 -2.04349923e+00 2.56057709e-01 -4.95795667e-01 2.31720835e-01 3.79777759e-01 -6.49288654e-01 -1.00741684e-01 2.51513034e-01 2.30061576e-01 -6.90609276e-01 6.44303679e-01 1.62065387e+00 -1.43632963e-01 -3.35503906e-01 -1.65098291e-02 -6.55092418e-01 7.06063330e-01 8.53238583e-01 -5.25702953e-01 -3.44054937e-01 -4.80352372e-01 -6.31477609e-02 2.55898476e-01 -2.21912012e-01 -6.85269654e-01 -1.09509444e-02 -8.49101245e-02 -4.51008491e-02 -7.27577031e-01 3.95541877e-01 -2.56637558e-02 -3.16180795e-01 1.83695674e-01 -6.42151415e-01 -2.54713774e-01 -5.63793480e-02 1.59553334e-01 -5.71067691e-01 -7.40724206e-01 4.65499520e-01 -2.48754129e-01 -4.31065351e-01 -5.00352085e-01 -4.64390308e-01 5.74320734e-01 8.62569928e-01 7.75981396e-02 -9.00055110e-01 -2.91978180e-01 -8.90302122e-01 3.35311204e-01 1.99003130e-01 2.36098111e-01 2.91579425e-01 -9.60846007e-01 -9.11913991e-01 -1.79255307e-02 -2.91777343e-01 5.79765141e-01 9.29378420e-02 6.20293438e-01 -3.67979228e-01 6.48420393e-01 2.12598011e-01 -6.56247377e-01 -1.26020062e+00 1.79248825e-01 -1.68025121e-01 -1.06028461e+00 -4.54676777e-01 1.14312565e+00 2.50882387e-01 -3.44867826e-01 2.79400181e-02 -3.63601595e-01 -7.01593101e-01 3.09586853e-01 2.90154457e-01 3.27850968e-01 5.31181321e-03 -8.44138265e-01 -4.51050587e-02 3.75655115e-01 -1.31173179e-01 -1.32848263e-01 1.23485255e+00 -1.76320821e-01 -2.75347561e-01 6.45438313e-01 1.07633090e+00 -4.89707664e-03 -1.08170044e+00 -1.10439092e-01 5.50374329e-01 3.38432975e-02 -1.82153791e-01 -7.47171283e-01 -5.72639406e-01 1.18788087e+00 -1.23790065e-02 4.64134306e-01 7.41244137e-01 4.00932938e-01 7.78732061e-01 2.89949119e-01 9.94070917e-02 -1.11719024e+00 2.89612830e-01 9.51310873e-01 7.84195662e-01 -1.26680434e+00 -3.24843712e-02 -1.07025838e+00 -8.61557305e-01 8.95120800e-01 6.94265366e-01 -9.78087112e-02 1.67131096e-01 3.20636332e-02 2.81227201e-01 -7.75843635e-02 -5.84332883e-01 -2.72050619e-01 3.17825705e-01 4.83536154e-01 1.10295367e+00 3.43864292e-01 -6.35572851e-01 7.87881970e-01 -3.66192222e-01 -4.71852392e-01 5.58138311e-01 8.22699964e-01 -7.22995162e-01 -1.63192332e+00 -2.16514677e-01 1.53551221e-01 -5.28672516e-01 -6.15961254e-02 -4.75439310e-01 9.18862998e-01 -2.72775274e-02 1.33534276e+00 3.38656940e-02 -1.94747388e-01 7.75046945e-02 4.39749599e-01 4.00685459e-01 -1.13206327e+00 -6.71900272e-01 3.98950875e-01 8.64558995e-01 -2.45660152e-02 -7.64800727e-01 -6.44184291e-01 -1.65677714e+00 2.19541103e-01 -5.42019367e-01 4.71396327e-01 2.86188811e-01 1.17366970e+00 1.74989004e-03 9.31972802e-01 4.21695054e-01 -5.76644421e-01 -4.48908538e-01 -1.54513061e+00 -3.20341408e-01 4.74459648e-01 -8.96524936e-02 -5.74745357e-01 -1.72986373e-01 3.16180676e-01]
[10.79276180267334, 9.330761909484863]
88dd2647-e76a-45d7-8c65-4e379884b358
non-rigid-3d-shape-retrieval-based-on-multi
1904.00765
null
http://arxiv.org/abs/1904.00765v1
http://arxiv.org/pdf/1904.00765v1.pdf
Non-rigid 3D shape retrieval based on multi-view metric learning
This study presents a novel multi-view metric learning algorithm, which aims to improve 3D non-rigid shape retrieval. With the development of non-rigid 3D shape analysis, there exist many shape descriptors. The intrinsic descriptors can be explored to construct various intrinsic representations for non-rigid 3D shape retrieval task. The different intrinsic representations (features) focus on different geometric properties to describe the same 3D shape, which makes the representations are related. Therefore, it is possible and necessary to learn multiple metrics for different representations jointly. We propose an effective multi-view metric learning algorithm by extending the Marginal Fisher Analysis (MFA) into the multi-view domain, and exploring Hilbert-Schmidt Independence Criteria (HSCI) as a diversity term to jointly learning the new metrics. The different classes can be separated by MFA in our method. Meanwhile, HSCI is exploited to make the multiple representations to be consensus. The learned metrics can reduce the redundancy between the multiple representations, and improve the accuracy of the retrieval results. Experiments are performed on SHREC'10 benchmarks, and the results show that the proposed method outperforms the state-of-the-art non-rigid 3D shape retrieval methods.
['Zhixun Su', 'Ximin Liu', 'Nannan Li', 'Haohao Li', 'Shengfa Wang']
2019-03-20
null
null
null
null
['3d-shape-retrieval']
['computer-vision']
[-3.78443569e-01 -8.55317116e-01 -1.38894111e-01 -3.66814196e-01 -9.83817101e-01 -6.87113464e-01 5.66873014e-01 -1.81031749e-01 7.48693049e-02 5.26284277e-02 5.16471744e-01 3.20338070e-01 -6.72245860e-01 -7.11326241e-01 -7.27099553e-02 -1.12812614e+00 3.05784345e-01 6.08834028e-01 1.54823214e-01 -1.41079322e-01 3.57370585e-01 7.82091916e-01 -1.33372617e+00 1.86987206e-01 3.82672936e-01 9.57221925e-01 1.15688562e-01 -4.55235206e-02 -8.37657675e-02 1.80661023e-01 -1.75069422e-01 -7.05471113e-02 3.90531749e-01 -5.73053397e-02 -4.41027552e-01 -1.99910905e-02 1.98236108e-01 -3.27791154e-01 -4.76171196e-01 1.06979930e+00 8.39199126e-01 2.24189281e-01 1.08497214e+00 -8.89511883e-01 -1.05215824e+00 -5.43801151e-02 -7.66365409e-01 5.60083352e-02 3.71489078e-01 -4.63948846e-01 1.23935187e+00 -1.50896406e+00 4.39747691e-01 1.83373952e+00 2.68870741e-01 1.83370605e-01 -8.80785644e-01 -5.42758048e-01 -1.43554240e-01 2.65946567e-01 -1.92980707e+00 -3.06251407e-01 1.00973558e+00 -4.33321685e-01 5.18397868e-01 3.15061748e-01 3.24884772e-01 5.82280695e-01 -1.57078598e-02 8.78721654e-01 1.03598940e+00 -3.40527222e-02 -2.16187254e-01 -1.94893479e-01 -1.73973814e-01 7.86794722e-01 3.76423359e-01 7.17128888e-02 -1.74333557e-01 -3.86990756e-01 9.27378953e-01 4.91706401e-01 -2.29620054e-01 -8.66851747e-01 -1.36714840e+00 8.02739680e-01 4.39098209e-01 5.08508086e-01 -1.67023335e-02 -3.40158969e-01 3.56353939e-01 2.75695354e-01 4.67038631e-01 -1.71179593e-01 -1.74252823e-01 9.74692851e-02 -5.37939250e-01 6.57121092e-02 4.82292414e-01 9.17110205e-01 9.10771906e-01 -1.29157558e-01 -1.00826092e-01 1.15768409e+00 8.09570551e-01 1.08812642e+00 5.54841220e-01 -6.21948600e-01 5.23425043e-01 9.24045026e-01 -7.84218498e-03 -1.49485147e+00 -2.18273699e-01 -1.97486594e-01 -9.70847607e-01 -1.46073118e-01 1.68023109e-02 5.79886019e-01 -5.86083293e-01 1.33332884e+00 4.09475714e-01 9.81424823e-02 -4.19433527e-02 1.20319736e+00 1.04013443e+00 5.34597814e-01 -5.40276289e-01 -1.50304288e-01 1.00032318e+00 -6.13936901e-01 -4.38428819e-01 4.70669895e-01 6.31530702e-01 -1.20573282e+00 6.50465786e-01 3.86709087e-02 -9.02618825e-01 -6.74815476e-01 -1.15261185e+00 -7.57240281e-02 -1.33836269e-01 1.29870445e-01 3.46497208e-01 4.33047086e-01 -5.58175862e-01 5.07334232e-01 -7.62968242e-01 -3.60970944e-02 2.27514848e-01 1.80982009e-01 -5.90994954e-01 -5.43130696e-01 -1.03435898e+00 8.33219647e-01 1.88314363e-01 8.25716257e-02 -6.25492036e-01 -2.57340819e-01 -7.41441309e-01 -1.84800640e-01 9.00472850e-02 -4.66360986e-01 4.09079820e-01 -7.69871324e-02 -1.19949043e+00 9.85969782e-01 -2.30033278e-01 6.15746498e-01 1.80546656e-01 9.40004885e-02 -5.57535708e-01 2.07607329e-01 1.09258503e-01 3.83267775e-02 8.10007632e-01 -1.34048104e+00 -1.92764536e-01 -7.49035418e-01 -7.71738663e-02 5.25895715e-01 -3.26047480e-01 -2.59573646e-02 -5.35612822e-01 -6.43491089e-01 8.87054682e-01 -9.73239362e-01 2.49010324e-01 -7.79322758e-02 2.66442820e-02 -5.33056736e-01 1.00722647e+00 -4.29052144e-01 1.08127820e+00 -2.25589037e+00 9.15086091e-01 5.38682818e-01 1.94775447e-01 1.02565028e-01 -4.18518931e-01 3.04882646e-01 -5.23079634e-02 2.47397553e-02 -9.79808122e-02 -1.38867080e-01 -1.36352152e-01 4.02546793e-01 -8.24670568e-02 7.00737953e-01 -6.85829818e-02 7.81431735e-01 -7.68054605e-01 -6.59775198e-01 3.21317434e-01 5.53300083e-01 -2.23668158e-01 2.23552257e-01 5.15214205e-01 4.88347530e-01 -1.01870024e+00 9.56934690e-01 9.72302020e-01 -6.91071302e-02 -2.19617531e-01 -6.14022195e-01 8.16843212e-02 -1.38600439e-01 -1.55860817e+00 2.01863194e+00 -3.34126651e-01 -1.82373866e-01 -1.72573179e-01 -1.18463969e+00 1.41943610e+00 7.97155723e-02 8.45460415e-01 -5.56715012e-01 1.03374049e-01 5.01660347e-01 -2.44737417e-01 -4.61842120e-01 -2.73374151e-02 -1.24376528e-01 2.42066849e-02 5.87776661e-01 -5.41299358e-02 -2.15009272e-01 -1.95630103e-01 -2.41669476e-01 5.16451538e-01 2.02625275e-01 1.43534511e-01 -4.21203852e-01 1.20722568e+00 -7.26572037e-01 6.25713766e-01 2.78357249e-02 3.89983021e-02 8.94557595e-01 -9.34449211e-02 -6.35236919e-01 -8.05056334e-01 -1.31832206e+00 -3.83096933e-01 6.36160374e-01 6.23475790e-01 -2.97515631e-01 -1.26595885e-01 -6.12153769e-01 2.45850235e-01 1.68414056e-01 -3.96285444e-01 -4.77875590e-01 -6.20886028e-01 -5.19705176e-01 2.00877443e-01 3.24466616e-01 4.11083072e-01 -5.27910411e-01 1.32306824e-02 7.05126822e-02 -1.85760871e-01 -7.82915473e-01 -8.72458637e-01 -5.12918830e-01 -1.03356206e+00 -1.13411438e+00 -1.08875668e+00 -7.37996519e-01 6.28374875e-01 1.12153995e+00 6.30743206e-01 1.23240352e-01 -1.82962403e-01 8.29153001e-01 -5.11865914e-01 6.04386888e-02 -9.56785679e-02 -1.50112748e-01 3.72577220e-01 3.06857705e-01 5.95842421e-01 -7.72427797e-01 -5.57650149e-01 7.50145316e-01 -9.78105903e-01 -3.66531551e-01 5.92620671e-01 8.74406993e-01 1.11472189e+00 -2.58781224e-01 3.05370867e-01 -1.75750390e-01 4.47418422e-01 -4.07070816e-01 -4.26344186e-01 6.16344988e-01 -5.65010011e-01 2.54694730e-01 3.55481446e-01 -4.46216881e-01 -6.93005383e-01 -3.26795697e-01 3.32531393e-01 -1.02341425e+00 2.44885057e-01 5.68699419e-01 -5.73446870e-01 -3.99974108e-01 2.36055896e-01 5.27062297e-01 1.88961536e-01 -7.69249380e-01 4.67041910e-01 7.02201903e-01 -1.56059235e-01 -7.38400936e-01 1.11415243e+00 5.32050014e-01 4.96387124e-01 -6.87589407e-01 -8.95369709e-01 -7.14577913e-01 -9.04843748e-01 -7.01563014e-03 5.85217535e-01 -9.17782366e-01 -5.06290615e-01 3.78874898e-01 -1.03747749e+00 8.08083177e-01 4.05198336e-02 7.45170891e-01 -5.97053647e-01 8.77115071e-01 -2.27704257e-01 -5.71146190e-01 -3.79464895e-01 -1.33320999e+00 1.42625070e+00 3.12644571e-01 2.76641101e-01 -9.38539267e-01 3.17224532e-01 2.69835263e-01 2.54972965e-01 6.24817163e-02 9.72266376e-01 -7.63783097e-01 -6.23334289e-01 -1.82074487e-01 -3.78588319e-01 3.85148793e-01 6.61373198e-01 -1.06211022e-01 -6.59535706e-01 -7.15790987e-01 3.65191013e-01 -1.55834198e-01 8.78158689e-01 9.11873132e-02 1.10683739e+00 -1.49059966e-02 -2.73930818e-01 6.52954757e-01 1.30073619e+00 1.59715414e-01 5.36809385e-01 1.12955309e-01 9.72020328e-01 3.92601073e-01 7.35170603e-01 4.48088080e-01 4.17070627e-01 8.07794094e-01 2.39612043e-01 2.87974447e-01 2.98716500e-02 -1.80287927e-01 1.79925531e-01 1.74827445e+00 -5.30144453e-01 3.65323395e-01 -6.29989505e-01 3.06170702e-01 -1.82615089e+00 -1.00843620e+00 7.09149688e-02 2.42190981e+00 3.29898417e-01 -3.62018675e-01 -1.53184697e-01 1.72763709e-02 7.91904628e-01 4.76839542e-01 -5.33323705e-01 5.65448664e-02 -3.36668968e-01 1.12104237e-01 3.61868553e-02 3.05282801e-01 -1.11792219e+00 4.49374080e-01 4.96348000e+00 1.07903171e+00 -8.06274295e-01 4.31508608e-02 5.08305542e-02 1.19047314e-01 -7.86500275e-01 3.99547480e-02 -6.34459794e-01 1.21675722e-01 1.81698218e-01 -4.88942176e-01 5.15708566e-01 7.38838255e-01 -7.92149454e-02 4.86194074e-01 -9.35827076e-01 1.61573374e+00 4.92886066e-01 -8.72837842e-01 6.63889229e-01 1.90015584e-01 7.33330727e-01 7.11445650e-03 6.62587509e-02 1.47444129e-01 -2.40998462e-01 -8.07995319e-01 2.15518087e-01 1.08284640e+00 7.48481452e-01 -9.22405064e-01 6.51713371e-01 1.45894408e-01 -1.81779265e+00 2.11907223e-01 -9.46052730e-01 4.98879552e-01 -8.13316777e-02 5.32732308e-01 -2.20949408e-02 1.21746397e+00 4.58327562e-01 1.32248914e+00 -5.95703781e-01 1.04861152e+00 1.09560609e-01 -2.47982502e-01 -2.69071400e-01 7.21288174e-02 -1.96141377e-02 -8.41073513e-01 8.37039769e-01 6.20665848e-01 6.84931993e-01 3.30084860e-01 4.11459655e-01 6.57890737e-01 5.33456728e-02 5.72571814e-01 -7.40572989e-01 1.44948468e-01 4.29727912e-01 1.27623999e+00 -2.86051482e-01 -5.70121631e-02 -7.73133099e-01 8.65268111e-01 1.49903238e-01 2.51391679e-01 -5.35714030e-01 -2.80459911e-01 7.09798336e-01 -1.71699226e-01 3.31336290e-01 -5.86391091e-01 1.04639225e-01 -1.50099909e+00 1.46442980e-01 -7.15960562e-01 5.44757962e-01 -6.24431252e-01 -1.84350336e+00 1.85010657e-01 5.59594445e-02 -1.87138748e+00 1.30450293e-01 -5.10766506e-01 -3.45833778e-01 9.97819781e-01 -1.71191859e+00 -1.30391371e+00 -1.86450511e-01 8.90157402e-01 2.00670093e-01 -5.22257090e-01 1.08411181e+00 4.32842225e-01 -1.23857372e-01 4.80892539e-01 5.94429016e-01 1.03639498e-01 1.03412914e+00 -8.64570916e-01 -1.92317709e-01 2.34771237e-01 4.01205450e-01 7.15562224e-01 -1.68395028e-01 -4.64074194e-01 -1.91663921e+00 -8.17296803e-01 4.77335155e-01 -3.56231689e-01 4.26334232e-01 1.96191981e-01 -1.12437475e+00 1.60034686e-01 -4.39115226e-01 2.97210366e-01 1.01006114e+00 -3.33124539e-03 -7.99353361e-01 -3.21132183e-01 -1.00993550e+00 2.62262493e-01 1.19012499e+00 -8.60101819e-01 -9.48216915e-01 2.30153769e-01 5.73555827e-01 -2.63473719e-01 -1.28804660e+00 7.23269880e-01 7.22087979e-01 -7.95506060e-01 1.28839111e+00 -5.72857797e-01 1.21167883e-01 -6.28222406e-01 -7.25770831e-01 -1.07712579e+00 -7.07583666e-01 -6.39482364e-02 -2.08273098e-01 1.19744921e+00 -9.54281688e-02 -5.73355556e-01 2.32832402e-01 3.80925149e-01 2.02678367e-02 -7.82379568e-01 -1.33030212e+00 -9.33529198e-01 2.83234775e-01 9.79702994e-02 7.80937254e-01 9.20437336e-01 -4.85598087e-01 2.46909216e-01 -3.15122366e-01 1.55925304e-01 8.12910140e-01 8.20971668e-01 5.89416742e-01 -1.77256525e+00 -1.19639067e-02 -6.21424258e-01 -7.71831632e-01 -1.16175675e+00 1.66805297e-01 -1.38744307e+00 -5.93747973e-01 -1.37212694e+00 5.57299674e-01 -4.19936597e-01 -7.92392313e-01 2.26341620e-01 -8.91241729e-02 -1.30857691e-01 1.72537223e-01 7.60562360e-01 -5.41248560e-01 1.12157035e+00 1.71126890e+00 -3.88150841e-01 3.12265344e-02 6.50720522e-02 -4.85113978e-01 5.91873288e-01 3.46614808e-01 -3.10851693e-01 -4.16163176e-01 -5.23310006e-01 3.61104533e-02 3.65126729e-02 8.20726603e-02 -7.21012712e-01 1.48081854e-01 -2.22081468e-01 5.37721753e-01 -7.93071926e-01 4.43884343e-01 -1.01295924e+00 1.93336979e-02 2.23416030e-01 1.68705463e-01 8.50014761e-03 -1.70459270e-01 7.82205582e-01 -4.25088227e-01 -2.05488548e-01 6.99924290e-01 -1.86189413e-01 -3.59563649e-01 8.71868432e-01 4.87651855e-01 1.16973303e-01 7.32797027e-01 -1.97189227e-01 4.77873348e-02 -1.54872715e-01 -6.37233913e-01 1.87052682e-01 4.32120830e-01 8.25966239e-01 1.05784917e+00 -2.23968053e+00 -8.08279157e-01 2.96217173e-01 3.83114398e-01 -1.42397597e-01 3.46008241e-01 6.14270449e-01 -4.22874577e-02 3.37466300e-01 -1.71773478e-01 -9.22775269e-01 -1.25448215e+00 5.50274551e-01 2.55294383e-01 -2.66113341e-01 -3.59091043e-01 3.94557357e-01 2.63818920e-01 -5.91725409e-01 -2.17493266e-01 1.94134608e-01 -4.92601305e-01 3.63200188e-01 4.52397376e-01 5.11578500e-01 5.10066152e-02 -1.14398491e+00 -5.03445864e-01 1.56375206e+00 -4.17101160e-02 9.02986154e-02 1.53817844e+00 -2.55736142e-01 -4.18322504e-01 5.83419323e-01 1.79393220e+00 7.79169276e-02 -5.71187317e-01 -7.28586197e-01 -1.23195313e-01 -8.30846250e-01 -2.59776157e-03 -6.04329631e-02 -1.14736784e+00 1.00366032e+00 7.17610657e-01 -4.99835461e-02 9.82438624e-01 4.37288499e-03 7.58488178e-01 5.49633801e-01 8.11512232e-01 -8.32558990e-01 2.66688973e-01 6.55103385e-01 1.25983334e+00 -1.10032928e+00 3.82813454e-01 -2.81361192e-01 -3.16839010e-01 1.45725846e+00 3.21386188e-01 -2.85010576e-01 1.02102947e+00 -5.10850549e-01 -2.26317272e-01 -2.38040268e-01 -9.66660529e-02 -1.19861729e-01 8.89621377e-01 4.67851788e-01 2.15635747e-01 9.37942192e-02 -4.53441262e-01 5.90721965e-01 3.05811673e-01 -4.85896468e-01 -7.04886317e-02 6.08929276e-01 -3.61051053e-01 -1.52126396e+00 -3.92304301e-01 3.37294340e-01 2.39980221e-02 2.66332656e-01 -2.66512156e-01 4.81322408e-01 -1.97664469e-01 6.93242729e-01 -4.72282797e-01 -5.92845321e-01 5.13208210e-01 2.69469060e-02 6.70199811e-01 -5.16551077e-01 2.19931200e-01 3.97883803e-01 -5.24796247e-01 -5.59271693e-01 -8.32335293e-01 -7.56527245e-01 -1.06495786e+00 -6.40152693e-02 -4.56063300e-01 9.26661566e-02 3.42802554e-01 9.19622898e-01 4.09188896e-01 -1.22012980e-01 1.30356717e+00 -8.32763672e-01 -8.46536577e-01 -7.07716763e-01 -7.53358483e-01 7.43979931e-01 1.27938483e-02 -1.31794548e+00 -5.19138634e-01 -4.61270303e-01]
[8.17963981628418, -3.917731523513794]
b5de2482-9bc5-43dd-8a7f-1770c06511fd
normalization-and-back-transliteration-for
null
null
https://aclanthology.org/2021.calcs-1.15
https://aclanthology.org/2021.calcs-1.15.pdf
Normalization and Back-Transliteration for Code-Switched Data
Code-switching is an omnipresent phenomenon in multilingual communities all around the world but remains a challenge for NLP systems due to the lack of proper data and processing techniques. Hindi-English code-switched text on social media is often transliterated to the Roman script which prevents from utilizing monolingual resources available in the native Devanagari script. In this paper, we propose a method to normalize and back-transliterate code-switched Hindi-English text. In addition, we present a grapheme-to-phoneme (G2P) conversion technique for romanized Hindi data. We also release a dataset of script-corrected Hindi-English code-switched sentences labeled for the named entity recognition and part-of-speech tagging tasks to facilitate further research.
['Thamar Solorio', 'Dwija Parikh']
null
null
null
null
naacl-calcs-2021-6
['transliteration']
['natural-language-processing']
[-3.55116501e-02 -2.54857004e-01 -4.81954776e-02 -6.54369831e-01 -1.02949774e+00 -1.06474113e+00 2.79981017e-01 1.29433975e-01 -4.49239165e-01 1.05938625e+00 2.28624806e-01 -6.74843490e-01 3.04620087e-01 -4.22405750e-01 -5.23863554e-01 -2.77552366e-01 4.15370017e-01 5.72955251e-01 6.86751725e-03 -3.71169090e-01 2.38719150e-01 3.15109611e-01 -7.87076652e-01 3.34783643e-01 1.23421121e+00 -3.52511048e-01 5.55339932e-01 6.20482743e-01 -3.94490629e-01 1.05926096e+00 -4.20793116e-01 -8.61573398e-01 2.90473938e-01 -9.36634898e-01 -8.67439806e-01 -2.94509411e-01 3.62206817e-01 3.18035215e-01 -2.21063524e-01 1.46818841e+00 2.59106904e-01 -4.14643623e-02 5.58121860e-01 -8.66609335e-01 -7.27001488e-01 1.08713007e+00 -3.52620423e-01 1.61085576e-01 5.35772443e-01 -3.89982879e-01 7.63197482e-01 -8.74173582e-01 1.21152222e+00 7.16166615e-01 5.76406896e-01 5.11560142e-01 -8.40918958e-01 -8.50349963e-01 -4.02783424e-01 7.59415030e-02 -1.81749403e+00 -4.40557390e-01 3.73804927e-01 -9.49612021e-01 1.16422641e+00 1.68795034e-01 3.61814380e-01 7.15797484e-01 3.82547468e-01 3.91104817e-01 1.06400299e+00 -8.49635601e-01 -3.17442507e-01 3.91176462e-01 -3.26419085e-01 7.94580996e-01 1.80189192e-01 -4.92257267e-01 -3.30917358e-01 2.02332288e-01 3.89656484e-01 -4.07987498e-02 7.28900805e-02 9.79293510e-02 -1.16733956e+00 8.92395198e-01 -2.62581766e-01 8.18368316e-01 -1.39808543e-02 -3.95617127e-01 5.49946368e-01 4.93269235e-01 3.26717049e-01 4.36016738e-01 -6.79429829e-01 -7.15619087e-01 -1.08083284e+00 -3.04305762e-01 7.87634969e-01 1.75601459e+00 9.25789416e-01 -1.25301644e-01 2.79719085e-01 1.30615723e+00 4.55301136e-01 7.81487465e-01 6.77431166e-01 -5.47056794e-01 9.18283999e-01 4.95233774e-01 -2.27202863e-01 -6.51329339e-01 -8.69660079e-02 -8.85161161e-02 -3.77352655e-01 -1.89918175e-01 4.15941656e-01 -1.91744685e-01 -5.81190825e-01 1.29671478e+00 1.42322049e-01 -7.75041401e-01 1.26736417e-01 3.67159367e-01 4.91885632e-01 7.61176765e-01 1.50607843e-02 -1.68323413e-01 1.29481101e+00 -1.30579638e+00 -7.92825758e-01 -3.45827758e-01 8.76489758e-01 -1.33470011e+00 9.58601058e-01 -3.04566361e-02 -8.01997304e-01 -2.64508307e-01 -7.47501493e-01 -2.89207637e-01 -5.78432858e-01 9.95299444e-02 2.05782309e-01 9.69224513e-01 -9.96725142e-01 1.44923300e-01 -8.06881070e-01 -7.88642585e-01 -1.23405606e-01 9.03437659e-02 -6.73012853e-01 -2.41227597e-01 -1.05800021e+00 1.03899789e+00 4.93182480e-01 -2.78218210e-01 -5.73582232e-01 -2.16369763e-01 -1.14063096e+00 -3.74049813e-01 -2.29507647e-02 8.16308707e-02 1.30871129e+00 -9.20367241e-01 -1.46626723e+00 1.36654174e+00 -1.62725911e-01 1.26304189e-02 4.22842532e-01 1.00609779e-01 -1.00728047e+00 -2.38969892e-01 5.53655684e-01 1.45372316e-01 4.99367833e-01 -7.67661750e-01 -9.80063856e-01 -1.18174993e-01 -2.75768638e-01 1.57328129e-01 -1.08914942e-01 8.90301764e-01 -6.37805045e-01 -8.18192482e-01 2.68417280e-02 -1.42572057e+00 -1.42736688e-01 -5.70065558e-01 -2.94231713e-01 2.07293808e-01 2.92054117e-01 -1.49437654e+00 1.61480653e+00 -2.17807293e+00 -8.31536427e-02 1.37805864e-01 -1.67461023e-01 2.44634137e-01 -1.38258696e-01 8.98186207e-01 -1.82242095e-01 2.18056142e-01 -6.16645277e-01 -2.64172822e-01 -1.93328023e-01 4.44800317e-01 1.55353591e-01 6.53313041e-01 -1.31938934e-01 6.02637768e-01 -1.12050092e+00 -8.35512757e-01 -4.05895784e-02 1.18613847e-01 -5.52313566e-01 7.00111389e-02 3.25308114e-01 7.23675787e-01 -1.32904291e-01 7.36846805e-01 5.56981087e-01 3.66469413e-01 4.43062097e-01 6.91031635e-01 -7.91973770e-01 5.34470558e-01 -5.58108807e-01 2.02957249e+00 -9.67669249e-01 8.19646716e-01 2.49644835e-02 -5.47885299e-01 8.15630674e-01 3.79981160e-01 2.53426041e-02 -7.12157369e-01 3.64559665e-02 7.29401171e-01 9.78181884e-02 -6.51724994e-01 8.53484929e-01 -1.75788432e-01 -6.09914958e-01 1.62627727e-01 1.19894840e-01 -1.94380417e-01 7.63355017e-01 1.23524338e-01 8.39146852e-01 1.19387627e-01 9.08344924e-01 -4.34351295e-01 7.64763176e-01 3.20147663e-01 7.41940260e-01 4.92991090e-01 -1.52395532e-01 6.88759863e-01 2.41333187e-01 7.75246099e-02 -1.34714270e+00 -8.56901765e-01 -2.43844107e-01 1.36969399e+00 -6.05743229e-01 -3.85116965e-01 -8.22980583e-01 -8.40171456e-01 -5.33103287e-01 9.49522972e-01 -2.12505892e-01 4.63031203e-01 -8.82599711e-01 -4.66728449e-01 8.04085135e-01 1.08169250e-01 7.60342181e-02 -9.73205864e-01 1.79695785e-01 4.56541121e-01 -6.05602384e-01 -1.11887920e+00 -1.02972209e+00 3.46974045e-01 -4.14154410e-01 -7.89923906e-01 -7.12175906e-01 -1.45158362e+00 8.69690895e-01 -9.21873301e-02 8.39549005e-01 -7.05090992e-04 8.78181681e-02 -1.84497386e-01 -8.18930924e-01 -1.60855323e-01 -1.30881870e+00 5.88385463e-01 -2.73084223e-01 -3.25286776e-01 4.68874395e-01 -2.58790076e-01 1.63756222e-01 1.09599218e-01 -6.83230817e-01 -2.42046528e-02 4.67754453e-01 3.31014067e-01 3.43477339e-01 -3.38407695e-01 2.38337982e-02 -1.42515695e+00 3.37703228e-01 -6.32401884e-01 -6.96462512e-01 4.49851900e-01 -4.42974776e-01 3.11089605e-02 1.05661476e+00 -7.71804452e-02 -1.27070642e+00 3.21374208e-01 -6.11724734e-01 4.72661167e-01 -2.23605230e-01 6.47653341e-01 -3.00475448e-01 -1.61083639e-01 5.12958825e-01 3.55303705e-01 -5.48438847e-01 -5.96613348e-01 2.59946316e-01 1.21972597e+00 7.03304529e-01 -4.47386384e-01 8.32352638e-01 -1.51862167e-02 -7.12833881e-01 -1.06841254e+00 -4.82218772e-01 -7.80945837e-01 -1.01546764e+00 -1.26379251e-01 1.21838844e+00 -1.18659532e+00 1.30137101e-01 6.36717081e-01 -1.25993872e+00 -3.58771741e-01 1.93395521e-02 6.81985199e-01 -1.66760191e-01 4.96209830e-01 -9.81637180e-01 -3.43577117e-01 -2.76123360e-03 -9.82526064e-01 5.67411005e-01 1.18880831e-01 -3.29694510e-01 -9.61772263e-01 7.01064050e-01 5.49265504e-01 3.46656859e-01 -9.23325717e-02 1.13311386e+00 -8.68577719e-01 -4.79931496e-02 -1.88436389e-01 -1.49574950e-01 3.53379250e-01 1.84723109e-01 3.29324991e-01 -2.92655081e-01 -2.87469804e-01 -1.52750220e-02 4.74253371e-02 7.84064010e-02 -2.78207839e-01 2.40232706e-01 -4.68143463e-01 1.70546159e-01 4.63101178e-01 1.42839587e+00 7.22503781e-01 5.64200222e-01 5.12471914e-01 9.18884158e-01 3.21604133e-01 4.93825585e-01 4.83494043e-01 8.38313699e-01 4.55109239e-01 -3.25504184e-01 1.65303826e-01 -2.83637077e-01 -4.46832150e-01 6.40774906e-01 2.15722561e+00 1.25730753e-01 -1.46417201e-01 -1.48638725e+00 8.95408273e-01 -1.32600808e+00 -7.72232711e-01 -7.95739353e-01 2.03563643e+00 1.27119005e+00 -3.48137975e-01 -3.41048598e-01 -2.03424081e-01 1.14044333e+00 -2.07903281e-01 9.47243795e-02 -7.90667772e-01 -8.46547410e-02 3.44159991e-01 7.66283035e-01 7.55992591e-01 -8.87733996e-01 1.33451104e+00 5.67197037e+00 8.35643649e-01 -1.06657648e+00 6.94536269e-01 -1.21037215e-01 4.75131720e-01 -1.99786767e-01 4.42403853e-01 -8.93320203e-01 6.13646269e-01 1.28230250e+00 -2.49509439e-01 9.13727641e-01 1.05618203e+00 -9.32305492e-03 -9.67379063e-02 -9.81606841e-01 9.77768958e-01 2.21137628e-01 -7.73657143e-01 -4.29801673e-01 -4.35556471e-02 1.25470507e+00 6.85579360e-01 -4.93239611e-01 5.26290417e-01 6.25466049e-01 -5.16403198e-01 1.13829064e+00 4.83739283e-03 1.19141710e+00 -6.75414920e-01 6.51083827e-01 2.84503311e-01 -1.33999705e+00 2.55649090e-01 -3.83100837e-01 -1.15474174e-02 1.30608499e-01 4.49718118e-01 -9.58880901e-01 4.74227816e-01 5.03946900e-01 5.95687866e-01 -9.20575023e-01 9.57111478e-01 -6.61578953e-01 6.82649910e-01 5.72527461e-02 -1.55746490e-01 1.68982044e-01 -5.30123830e-01 2.63687819e-01 1.82959843e+00 7.32326627e-01 -3.40602428e-01 7.18121901e-02 2.03344896e-01 -1.71524972e-01 7.58900464e-01 -6.16440356e-01 -4.80107576e-01 4.71649915e-01 1.06041729e+00 -1.04618454e+00 -3.47443312e-01 -9.79302943e-01 1.44132793e+00 3.40230048e-01 3.11998516e-01 -6.80900455e-01 -1.02314293e+00 4.83746827e-01 -1.57920912e-01 3.52106422e-01 -5.51091015e-01 -1.95039921e-02 -1.55556428e+00 -1.30808100e-01 -1.07961321e+00 2.92743176e-01 -5.44736147e-01 -1.09930277e+00 7.55193710e-01 -3.14154595e-01 -1.23770881e+00 -1.88123628e-01 -3.27885479e-01 -1.74765348e-01 7.05477536e-01 -1.26806295e+00 -9.45994318e-01 7.74084777e-02 7.13904977e-01 5.36435783e-01 -3.02739829e-01 8.98135483e-01 1.08522904e+00 -5.43165922e-01 6.00071847e-01 8.68631482e-01 7.62534380e-01 1.04599464e+00 -1.00260758e+00 7.14002371e-01 1.39002240e+00 3.23020428e-01 8.56228352e-01 7.24577665e-01 -8.74798179e-01 -1.00510144e+00 -1.35149848e+00 2.01096559e+00 -4.99603450e-01 1.06548440e+00 -6.74177766e-01 -5.18612623e-01 1.06288421e+00 5.83923101e-01 -3.17027539e-01 9.96400595e-01 -5.31423502e-02 -2.38498449e-01 2.61533380e-01 -1.03129554e+00 6.04371130e-01 9.17627275e-01 -1.15951788e+00 -6.65094495e-01 4.29791510e-01 4.25960809e-01 -3.82524252e-01 -9.74924624e-01 -4.68755454e-01 3.43281776e-01 -5.47638297e-01 1.50411308e-01 -1.96303621e-01 5.05267262e-01 -4.45433080e-01 -3.44152212e-01 -1.32711685e+00 -3.15898836e-01 -7.23241925e-01 9.01521802e-01 1.74687898e+00 8.05405200e-01 -5.28964818e-01 2.04772279e-01 5.11007726e-01 -4.60318953e-01 3.74170989e-01 -9.27567184e-01 -9.17499423e-01 3.44310611e-01 -3.35708946e-01 2.92974293e-01 1.43021929e+00 3.93633306e-01 2.11793229e-01 -4.78978217e-01 -1.12489223e-01 1.85305312e-01 -9.31948423e-02 5.36969841e-01 -7.95379639e-01 -1.18152268e-01 -9.98633578e-02 -3.98624510e-01 -6.13470614e-01 3.99197191e-01 -1.55918086e+00 6.54652059e-01 -1.43298066e+00 4.08033192e-01 -3.66202325e-01 2.23629504e-01 5.03107965e-01 3.05644155e-01 3.85735691e-01 2.74356097e-01 2.08820954e-01 -6.49819136e-01 1.56756386e-01 8.45252037e-01 1.17604673e-01 -9.83094350e-02 -2.55796224e-01 -4.26145136e-01 5.93675911e-01 8.21626782e-01 -1.37603605e+00 2.08315626e-02 -6.71304822e-01 6.98835015e-01 3.58698487e-01 -7.64573932e-01 -1.14661074e+00 2.46582955e-01 -3.99734974e-01 -2.11293340e-01 -3.69743496e-01 -5.00055075e-01 -8.29440951e-01 3.30799222e-01 4.04229879e-01 -1.88467905e-01 4.40906286e-01 -1.35173202e-01 -1.09849321e-02 -3.12205195e-01 -8.44802976e-01 1.01697302e+00 -3.23202193e-01 -5.88277221e-01 9.16838646e-02 -1.34346879e+00 7.24012971e-01 9.81906056e-01 -1.33218821e-02 -6.51866123e-02 -4.42721769e-02 -2.43879735e-01 -1.97775304e-01 1.08452344e+00 5.94825029e-01 -9.54969451e-02 -1.05339706e+00 -8.77402306e-01 3.72559458e-01 4.56796587e-01 -7.35184729e-01 1.29223689e-01 6.85747564e-01 -1.44455850e+00 5.49472630e-01 -5.49200833e-01 3.14919441e-03 -1.15747142e+00 4.07740325e-01 -2.37468168e-01 -1.83668911e-01 -2.31655583e-01 6.01981580e-01 -2.04937190e-01 -1.15267670e+00 -1.93726867e-01 -1.55780539e-01 -6.09359331e-02 1.79325975e-02 2.20608756e-01 2.32999384e-01 3.99325848e-01 -1.24320817e+00 -6.38532758e-01 4.01670635e-01 -8.37684199e-02 -2.27633640e-01 1.15268636e+00 -5.17469227e-01 -7.20792055e-01 4.91362512e-01 1.52751136e+00 9.95056868e-01 -3.43297929e-01 3.43816802e-02 3.79384130e-01 -4.55267429e-01 -6.21022105e-01 -5.74107945e-01 -8.24910998e-01 6.13924921e-01 1.74177527e-01 -2.26019189e-01 5.40112376e-01 -1.22666873e-01 8.48186791e-01 5.43993413e-01 5.92856884e-01 -1.56853712e+00 -7.93881834e-01 1.25998628e+00 4.33676839e-01 -1.01902115e+00 -4.49618876e-01 -1.52498096e-01 -7.40035653e-01 1.07213187e+00 2.53917038e-01 2.94670850e-01 7.76820183e-01 4.05823022e-01 5.37434697e-01 2.71766305e-01 -1.64985836e-01 -9.63397026e-02 -1.38201177e-01 5.80775499e-01 1.07055593e+00 1.37919262e-01 -9.15851474e-01 4.19640332e-01 -7.08182931e-01 -2.44910210e-01 1.11361110e+00 1.22663105e+00 -7.80786574e-02 -1.54053974e+00 -3.25853556e-01 2.46090069e-01 -9.39247131e-01 -6.71989799e-01 -1.71791524e-01 6.51765108e-01 2.89423347e-01 1.08658934e+00 -2.60694116e-01 -1.31440535e-01 1.34484276e-01 3.44984114e-01 2.66461909e-01 -1.10233951e+00 -1.01407719e+00 -5.37589751e-02 1.58293739e-01 -6.40571490e-02 -3.15344512e-01 -1.02745664e+00 -1.42642987e+00 -4.30942804e-01 -3.08228564e-02 5.06825507e-01 1.13822126e+00 7.94368446e-01 -1.01822458e-01 1.38602540e-01 4.27822471e-01 -2.60380477e-01 8.61107633e-02 -7.85881400e-01 -7.28150606e-01 3.55101436e-01 -9.75179598e-02 2.37004533e-02 -2.57736266e-01 5.57648897e-01]
[10.224018096923828, 10.088836669921875]
1e89d14c-5116-41c7-97e4-884f1ea5cd0b
gaitmpl-gait-recognition-with-memory
2306.0465
null
https://arxiv.org/abs/2306.04650v1
https://arxiv.org/pdf/2306.04650v1.pdf
GaitMPL: Gait Recognition with Memory-Augmented Progressive Learning
Gait recognition aims at identifying the pedestrians at a long distance by their biometric gait patterns. It is inherently challenging due to the various covariates and the properties of silhouettes (textureless and colorless), which result in two kinds of pair-wise hard samples: the same pedestrian could have distinct silhouettes (intra-class diversity) and different pedestrians could have similar silhouettes (inter-class similarity). In this work, we propose to solve the hard sample issue with a Memory-augmented Progressive Learning network (GaitMPL), including Dynamic Reweighting Progressive Learning module (DRPL) and Global Structure-Aligned Memory bank (GSAM). Specifically, DRPL reduces the learning difficulty of hard samples by easy-to-hard progressive learning. GSAM further augments DRPL with a structure-aligned memory mechanism, which maintains and models the feature distribution of each ID. Experiments on two commonly used datasets, CASIA-B and OU-MVLP, demonstrate the effectiveness of GaitMPL. On CASIA-B, we achieve the state-of-the-art performance, i.e., 88.0% on the most challenging condition (Clothing) and 93.3% on the average condition, which outperforms the other methods by at least 3.8% and 1.4%, respectively.
['Xi Li', 'Zequn Qin', 'Lin Dong', 'Yuhan Zhao', 'Pengyi Zhang', 'Huanzhang Dou']
2023-06-06
null
null
null
null
['gait-recognition']
['computer-vision']
[-1.33464977e-01 -6.74766958e-01 -6.09867908e-02 -1.49000823e-01 -4.39290464e-01 -6.93648914e-03 2.61253834e-01 -5.47477938e-02 -4.27649021e-01 8.41653347e-01 -1.44274577e-01 2.03078136e-01 5.80133498e-03 -7.26766407e-01 -6.27727628e-01 -1.00819647e+00 -3.30218196e-01 3.44608992e-01 5.57004094e-01 -1.23822158e-02 1.09806508e-01 2.59493142e-01 -1.63871753e+00 1.44877017e-01 9.81652677e-01 1.20206130e+00 2.49598175e-01 5.45571983e-01 1.87365308e-01 5.94857574e-01 -3.47868502e-01 -6.10363483e-01 2.61263877e-01 5.23806885e-02 -2.65462875e-01 1.33225784e-01 7.32533157e-01 -2.86408633e-01 -5.81174850e-01 1.16278148e+00 5.63272417e-01 2.33668357e-01 5.89227915e-01 -1.55090380e+00 -7.07989335e-01 -1.77669525e-02 -1.09227490e+00 1.64737195e-01 1.07675888e-01 5.12888610e-01 6.69920564e-01 -1.00690615e+00 1.46778777e-01 1.59169424e+00 8.07938397e-01 3.89594257e-01 -1.06293881e+00 -7.51697958e-01 2.82754540e-01 8.74597430e-01 -1.70977199e+00 -2.83488214e-01 6.60919428e-01 -4.17930365e-01 6.76226974e-01 1.83192417e-01 6.17431223e-01 1.04634714e+00 3.42859745e-01 8.67265999e-01 1.26187694e+00 -2.96459403e-02 8.11090600e-03 -2.28331596e-01 3.42902184e-01 8.25461030e-01 6.88806891e-01 1.80460662e-01 -4.14125025e-01 8.91935006e-02 8.71960819e-01 5.92963099e-02 -8.20676312e-02 -3.81992698e-01 -9.56245363e-01 3.50310087e-01 3.06484848e-01 -2.33704284e-01 -3.70523632e-02 -1.93860903e-01 5.53708494e-01 1.62606180e-01 2.81754620e-02 -4.42216843e-01 -2.21455216e-01 -1.05933450e-01 -7.43821084e-01 2.94802338e-01 5.12312412e-01 9.76144373e-01 6.63070023e-01 1.35799006e-01 -1.23789601e-01 1.02568662e+00 7.25387931e-01 1.00317335e+00 5.34344137e-01 -5.42631686e-01 8.55390906e-01 4.26837534e-01 1.46321282e-01 -1.44128084e+00 -3.03648829e-01 -3.89556468e-01 -1.37389708e+00 1.11561358e-01 6.73511267e-01 3.74120660e-02 -8.87210846e-01 1.65121794e+00 2.12538972e-01 4.80641901e-01 -1.94780871e-01 9.32446301e-01 7.78440237e-01 7.38012135e-01 3.91427487e-01 -7.19089061e-02 1.29690158e+00 -1.09533584e+00 -3.74445379e-01 -4.40109849e-01 1.39495641e-01 -7.29333937e-01 1.14797127e+00 3.38599443e-01 -9.17393625e-01 -1.09919214e+00 -1.29755139e+00 2.46039122e-01 -2.49918863e-01 2.94209272e-01 1.23391286e-01 7.21217692e-01 -6.84695721e-01 5.60232818e-01 -8.72765303e-01 -2.82383263e-01 4.31914717e-01 3.56792122e-01 -2.81550169e-01 -1.77982911e-01 -1.05066812e+00 7.26434231e-01 3.69946897e-01 2.70021319e-01 -5.70964277e-01 -5.06173790e-01 -8.41682851e-01 -6.41803071e-02 1.51327610e-01 -4.72777039e-01 5.06883979e-01 -4.02789861e-01 -1.21544886e+00 7.87834704e-01 -1.31481975e-01 -3.47488970e-01 8.46292317e-01 -3.20181102e-01 -7.64938772e-01 8.13927501e-02 2.01836795e-01 4.71070886e-01 1.11613297e+00 -1.09627807e+00 -6.85959220e-01 -4.19798851e-01 -4.51564133e-01 -7.85554349e-02 -2.96060175e-01 -1.83381125e-01 -6.21172905e-01 -8.81033123e-01 -6.21232986e-02 -9.31057930e-01 -3.25136781e-02 1.16754368e-01 -2.72060156e-01 -6.74065948e-02 9.30983722e-01 -1.06242168e+00 1.40526676e+00 -2.09358478e+00 -1.36162443e-02 2.00391248e-01 1.76543191e-01 7.48438597e-01 -2.62415200e-01 -1.75342649e-01 -2.43034959e-02 -2.55420417e-01 -5.24511077e-02 -3.53657842e-01 9.47879329e-02 2.57722437e-01 3.60955112e-02 4.70844060e-01 3.12964112e-01 8.22261751e-01 -9.05370235e-01 -7.68134117e-01 3.80278498e-01 3.57069880e-01 -1.68010145e-01 7.37813562e-02 3.89881819e-01 2.44144753e-01 -1.13159768e-01 9.66777503e-01 1.15907371e+00 -1.22159079e-01 4.51588407e-02 -5.78312457e-01 3.96684557e-02 -3.08355212e-01 -1.60424185e+00 1.09740567e+00 -4.92429510e-02 3.16386968e-01 -1.80107653e-01 -9.60166633e-01 1.08097732e+00 -1.26754671e-01 3.24886799e-01 -9.12710011e-01 -8.07828605e-02 2.27575645e-01 -1.74777638e-02 -5.42535484e-01 3.76161516e-01 1.58639982e-01 -1.09747604e-01 9.77614373e-02 1.61185972e-02 8.52581620e-01 3.81956011e-01 -2.70335943e-01 6.72184169e-01 -1.99597236e-03 8.35723877e-02 -4.03712302e-01 8.89244020e-01 -6.73064768e-01 9.44349527e-01 7.28488088e-01 -7.62408078e-01 5.74078679e-01 1.12026580e-01 -5.84910154e-01 -1.07662833e+00 -1.54060113e+00 -1.10494494e-02 7.97623634e-01 5.28970957e-01 -1.70407355e-01 -5.32540500e-01 -5.54615319e-01 8.93289447e-02 6.75565749e-02 -3.97382259e-01 -2.82111168e-01 -8.84436846e-01 -1.11403489e+00 5.09404302e-01 8.21291268e-01 1.31465721e+00 -6.18573666e-01 -3.54516625e-01 7.18095452e-02 -2.80399114e-01 -1.11130083e+00 -7.41838157e-01 -3.09613138e-01 -6.29413188e-01 -1.01605380e+00 -9.90121186e-01 -9.77866054e-01 6.37347460e-01 4.54701722e-01 9.97861564e-01 6.80187121e-02 -2.78390855e-01 5.22372760e-02 -3.59351672e-02 2.40314513e-01 1.81908458e-01 -2.22731568e-02 4.27556574e-01 5.03419757e-01 4.05203938e-01 -4.97625977e-01 -6.69460773e-01 7.52937078e-01 -3.10711384e-01 7.48188049e-02 7.07059860e-01 1.07867742e+00 6.12952411e-01 3.69696110e-01 4.24316674e-01 -1.93365633e-01 2.52288938e-01 -5.00738919e-01 -4.24407393e-01 4.39848393e-01 -3.85816753e-01 -8.78291875e-02 5.26324689e-01 -7.88225591e-01 -8.72689962e-01 -1.52477935e-01 -4.96209562e-02 -4.00158852e-01 -1.85528859e-01 2.07236886e-01 -6.81730390e-01 -3.14657003e-01 2.14724422e-01 4.46905732e-01 3.15450467e-02 -4.44466263e-01 -6.18286617e-02 4.42073524e-01 7.48510242e-01 -8.93674970e-01 1.14707351e+00 2.28442982e-01 1.90757498e-01 -8.47562253e-01 -3.35839987e-01 -2.65506506e-01 -7.39267528e-01 -4.83746946e-01 6.47633731e-01 -9.89708424e-01 -7.78399825e-01 1.17489874e+00 -6.27988398e-01 -2.79311925e-01 -5.06202094e-02 3.14378291e-01 -3.28726053e-01 7.91428328e-01 -9.12273824e-01 -6.92830682e-01 -4.00936484e-01 -1.07947290e+00 7.12851524e-01 4.60056782e-01 -4.93442453e-02 -7.37984121e-01 -4.16790575e-01 4.00627941e-01 3.80331963e-01 3.29872012e-01 8.05301368e-01 -7.42755085e-02 -5.43202102e-01 -1.05320632e-01 -6.75124526e-01 3.30261648e-01 1.32281125e-01 -6.34915195e-03 -7.66735017e-01 -6.55876756e-01 -4.45957482e-01 -2.66695023e-01 8.99242103e-01 3.35306644e-01 7.46328831e-01 -3.33906978e-01 -1.75121576e-01 4.26776558e-01 1.25870109e+00 3.20179820e-01 8.53654504e-01 4.18064713e-01 9.47268963e-01 4.86321330e-01 7.39586353e-01 3.93635362e-01 6.61727250e-01 7.98160255e-01 9.74080637e-02 -1.78460814e-02 -2.99550027e-01 -2.98155516e-01 5.61130583e-01 9.76072788e-01 -2.68938571e-01 1.29560739e-01 -8.38717639e-01 4.61158842e-01 -2.16959667e+00 -1.02418137e+00 -1.93780303e-01 2.39919257e+00 5.62649012e-01 5.30582428e-01 5.83158970e-01 3.25167179e-01 1.07458580e+00 1.15662731e-01 -6.23188734e-01 5.11630960e-02 -2.99300641e-01 8.03313125e-03 6.61275029e-01 3.78099650e-01 -1.50138605e+00 6.80323899e-01 5.18273354e+00 1.27687013e+00 -9.93874013e-01 -5.92209632e-03 8.31799388e-01 1.33255497e-02 6.10451937e-01 -5.44805050e-01 -1.11829507e+00 1.18843246e+00 7.74828196e-01 2.10896935e-02 2.04635620e-01 7.25084186e-01 2.02450752e-01 -4.77398559e-02 -7.83398271e-01 1.37027359e+00 1.55931357e-02 -9.41864669e-01 2.12870657e-01 -1.32235289e-01 5.14226317e-01 -3.41536373e-01 3.91966432e-01 4.75280643e-01 -6.42770575e-03 -7.84545481e-01 6.76413178e-01 6.08070314e-01 7.46577680e-01 -9.20479655e-01 8.22859466e-01 8.52471367e-02 -2.10184908e+00 -2.05477193e-01 -5.32594800e-01 -4.12586555e-02 2.12888718e-01 4.57080841e-01 5.58727831e-02 6.23183787e-01 1.04438937e+00 9.15393770e-01 -8.45669508e-01 1.43920231e+00 3.05997301e-02 4.42454398e-01 -2.64560819e-01 1.93369798e-02 2.56444197e-02 -2.22932339e-01 4.75955367e-01 1.40218771e+00 2.42640421e-01 -3.83852303e-01 5.65011561e-01 5.12825072e-01 1.84741050e-01 -1.53529003e-01 3.59020615e-03 2.57654905e-01 5.50603271e-01 9.63023365e-01 -5.33603728e-01 -3.98312092e-01 -5.26471078e-01 9.40904617e-01 1.82397693e-01 2.59908140e-01 -8.97260547e-01 -3.56554896e-01 8.75459194e-01 2.20067158e-01 4.36627030e-01 -4.31856900e-01 -1.61888480e-01 -1.26417553e+00 3.40416729e-01 -8.43774140e-01 4.80780482e-01 -3.65845054e-01 -1.82699192e+00 2.93646544e-01 -1.91662923e-01 -1.48473430e+00 2.15107098e-01 -7.32259691e-01 -6.55673981e-01 8.46766353e-01 -1.63348556e+00 -1.46686935e+00 -5.85387826e-01 7.85085857e-01 4.58839804e-01 -3.18477273e-01 4.10073817e-01 9.72475588e-01 -9.60154057e-01 1.11410880e+00 -5.23677133e-02 4.05303717e-01 8.64176989e-01 -1.01488769e+00 5.79749584e-01 1.01276076e+00 -5.20816207e-01 5.19343972e-01 4.14160907e-01 -7.72879839e-01 -1.43937218e+00 -1.49257660e+00 6.75828636e-01 -2.43967921e-02 7.49313772e-01 -2.13977233e-01 -1.04645610e+00 2.13924333e-01 -2.90304124e-01 1.32087052e-01 5.99284172e-01 -1.19201913e-01 -3.75396997e-01 -4.15879846e-01 -1.17415953e+00 7.19971120e-01 1.16893303e+00 -3.38564485e-01 -3.87934238e-01 -1.40028857e-02 2.18651533e-01 -2.66021997e-01 -9.37613189e-01 4.59183633e-01 8.59713376e-01 -7.23643601e-01 1.33372819e+00 -3.49981785e-01 1.41089723e-01 -8.13960195e-01 -3.83714080e-01 -9.12753463e-01 -7.78604984e-01 -2.30477616e-01 -6.09717607e-01 1.45496643e+00 -1.25736892e-01 -6.68226242e-01 8.07132900e-01 5.50850272e-01 2.22774789e-01 -7.37261832e-01 -1.00790071e+00 -1.23760223e+00 -3.14027034e-02 -1.31603524e-01 4.98952925e-01 7.01470673e-01 -4.04665053e-01 1.13570116e-01 -7.84409046e-01 2.99801856e-01 1.11046720e+00 3.04897614e-02 7.44524300e-01 -1.28572237e+00 -4.14218456e-01 -4.68867511e-01 -8.69744480e-01 -1.25267112e+00 -3.99462804e-02 -5.07950842e-01 -1.08163990e-01 -9.76749599e-01 2.71654159e-01 -6.05549991e-01 -5.37169397e-01 2.90098697e-01 -5.89018285e-01 4.31098878e-01 3.93273294e-01 2.12398350e-01 -8.95903826e-01 5.26462317e-01 1.09413683e+00 -5.73136687e-01 -8.67780000e-02 5.96722923e-02 -3.52713555e-01 6.96326733e-01 8.18767011e-01 4.03978117e-02 -2.04432473e-01 -2.71581620e-01 -2.95724690e-01 -3.23510796e-01 6.41851664e-01 -1.59081745e+00 2.64982224e-01 -1.72440149e-02 8.09444785e-01 -7.58466303e-01 3.96238267e-01 -6.28324986e-01 8.94554481e-02 8.14308703e-01 1.78343982e-01 1.96542665e-01 3.00936550e-01 6.67800426e-01 -4.21479829e-02 1.58239409e-01 9.91029620e-01 -4.92043085e-02 -1.06157660e+00 6.42214894e-01 -3.48213822e-01 1.39339998e-01 8.86082768e-01 -4.90863323e-01 -4.02708232e-01 6.49867058e-02 -5.74992299e-01 3.83994073e-01 3.00903380e-01 4.89813924e-01 8.44915152e-01 -1.78690159e+00 -7.27498055e-01 4.72500652e-01 5.26394993e-02 -3.68158937e-01 6.51095629e-01 8.19705129e-01 -2.64368623e-01 1.38954297e-01 -6.25854075e-01 -7.85835087e-01 -1.55309296e+00 4.97350097e-01 1.60590202e-01 -4.69368756e-01 -5.07581592e-01 6.84585333e-01 -6.95898384e-02 -1.65324166e-01 3.16235930e-01 7.68780010e-03 -1.62655011e-01 1.25924751e-01 7.21476197e-01 9.63246047e-01 -1.78361088e-01 -1.06051147e+00 -3.85733217e-01 9.53632891e-01 -2.83683866e-01 4.43001300e-01 9.54300702e-01 -1.85430706e-01 1.45841166e-01 2.38198861e-01 1.08438897e+00 -3.86230528e-01 -1.66161513e+00 -3.48679394e-01 -3.84411551e-02 -8.17361712e-01 -4.15428877e-01 -5.39865613e-01 -1.16708684e+00 8.41512859e-01 1.14815176e+00 -1.94243446e-01 9.79366124e-01 -6.04269981e-01 1.30020595e+00 2.86622286e-01 6.02183104e-01 -1.18383503e+00 3.10895741e-01 5.17374277e-01 5.74228168e-01 -1.30328107e+00 -5.82168773e-02 -4.88453001e-01 -4.85339075e-01 9.26926076e-01 8.26339245e-01 -2.20824331e-01 5.95944047e-01 2.06620187e-01 -3.02908510e-01 4.40842777e-01 -3.20591062e-01 -1.27198428e-01 3.50171953e-01 9.27290916e-01 -2.52707414e-02 3.30050945e-01 -1.78649262e-01 6.68517292e-01 3.90589498e-02 -8.97052214e-02 -4.26973365e-02 9.11647797e-01 -4.98286217e-01 -1.02297759e+00 -5.70014775e-01 4.12591815e-01 2.84559093e-02 1.35983318e-01 6.47431910e-02 5.17387092e-01 3.75156641e-01 9.59685147e-01 -2.98538040e-02 -8.16356838e-01 2.61774838e-01 -1.94844246e-01 5.54928660e-01 4.32320945e-02 -1.07457861e-01 -1.37662306e-01 3.24066766e-02 -4.97729003e-01 -2.65030414e-01 -7.45305061e-01 -9.07461643e-01 -8.36220324e-01 1.45903816e-02 -2.46248424e-01 -1.55931547e-01 8.75014484e-01 2.92775512e-01 3.90122890e-01 5.54948866e-01 -1.02947891e+00 -4.76443261e-01 -6.79838002e-01 -6.17387056e-01 5.31633377e-01 1.75653532e-01 -1.08002555e+00 -7.23138358e-03 1.98090941e-01]
[14.376871109008789, 1.3328334093093872]
2cd61c65-f073-49ae-a8ca-5d27f5f95fa4
conditioned-and-composed-image-retrieval
null
null
https://openaccess.thecvf.com/content/CVPR2022W/ODRUM/html/Baldrati_Conditioned_and_Composed_Image_Retrieval_Combining_and_Partially_Fine-Tuning_CLIP-Based_CVPRW_2022_paper.html
https://openaccess.thecvf.com/content/CVPR2022W/ODRUM/papers/Baldrati_Conditioned_and_Composed_Image_Retrieval_Combining_and_Partially_Fine-Tuning_CLIP-Based_CVPRW_2022_paper.pdf
Conditioned and Composed Image Retrieval Combining and Partially Fine-Tuning CLIP-Based Features
In this paper, we present an approach for conditioned and composed image retrieval based on CLIP features. In this extension of content-based image retrieval (CBIR), an image is combined with a text that provides information regarding user intentions and is relevant for application domains like e-commerce. The proposed method is based on an initial training stage where a simple combination of visual and textual features is used, to fine-tune the CLIP text encoder. Then in a second training stage, we learn a more complex combiner network that merges visual and textual features. Contrastive learning is used in both stages. The proposed approach obtains state-of-the-art performance for conditioned CBIR on the FashionIQ dataset and for composed CBIR on the more recent CIRR dataset.
['Alberto del Bimbo', 'Tiberio Uricchio', 'Marco Bertini', 'Alberto Baldrati']
2022-06-19
null
null
null
cvprw-2022-6
['composed-image-retrieval', 'content-based-image-retrieval']
['computer-vision', 'computer-vision']
[ 5.11196971e-01 -5.69268823e-01 -3.63080919e-01 -4.39591914e-01 -1.22308755e+00 -4.62525487e-01 9.21020508e-01 3.48297566e-01 -6.85961485e-01 3.27559412e-01 2.57346153e-01 3.14954445e-02 -3.55556875e-01 -5.65908551e-01 -6.09734476e-01 -5.60365677e-01 2.41015658e-01 3.51617783e-01 1.79614633e-01 -3.44213516e-01 2.91481316e-01 3.79106820e-01 -1.56975198e+00 8.80632818e-01 3.52298468e-01 1.46714473e+00 5.89259684e-01 6.76175594e-01 -8.80153179e-02 1.16669762e+00 -5.02852321e-01 -5.30202746e-01 2.84779221e-01 -4.04295921e-01 -5.63010335e-01 2.98024386e-01 6.22146785e-01 -4.59237158e-01 -4.14846361e-01 8.66297185e-01 4.02357489e-01 3.03344607e-01 8.95200074e-01 -1.03743804e+00 -7.83791780e-01 6.00324452e-01 -6.60294712e-01 6.59184828e-02 4.51494545e-01 -2.33574361e-01 1.20993924e+00 -9.80443299e-01 8.68292987e-01 1.14871287e+00 1.75134689e-01 5.84847853e-02 -1.06435061e+00 -5.56473434e-01 1.67205542e-01 3.59936595e-01 -1.49813902e+00 -3.20407957e-01 9.66405332e-01 -4.75188911e-01 6.00638390e-01 2.96390980e-01 5.23410022e-01 8.44602168e-01 2.29384571e-01 1.09838998e+00 1.04786229e+00 -8.18946123e-01 1.58125028e-01 4.34163898e-01 1.79592699e-01 5.34609079e-01 -1.31391585e-01 1.10987760e-02 -4.28941518e-01 3.18552293e-02 4.99288231e-01 2.26142839e-01 -2.78036594e-02 -3.16177428e-01 -8.87583315e-01 1.16081762e+00 6.21695817e-01 8.50091875e-01 -5.49940467e-01 8.07739124e-02 3.01347703e-01 4.84493405e-01 4.29420620e-01 2.84092128e-01 -4.09128368e-02 4.65193123e-01 -1.35751116e+00 1.51531249e-01 6.69879079e-01 9.27494526e-01 5.43128967e-01 -3.06956947e-01 -6.84022963e-01 1.14625084e+00 4.06642288e-01 6.11163795e-01 4.98688579e-01 -6.08000636e-01 2.33689621e-01 2.37292334e-01 1.75425466e-02 -1.13008034e+00 -1.03601992e-01 -4.98328537e-01 -8.13860178e-01 1.58896565e-01 7.53026232e-02 3.10028404e-01 -1.18281031e+00 1.26129067e+00 -1.59844294e-01 -5.64379208e-02 6.28955849e-03 1.09720445e+00 1.03126502e+00 8.46722126e-01 1.30970299e-01 -1.06562540e-01 1.59758222e+00 -1.08422446e+00 -7.65759587e-01 -7.56087452e-02 9.53107849e-02 -1.13362491e+00 7.48659968e-01 5.77134430e-01 -1.14701045e+00 -8.05026114e-01 -1.25224268e+00 -4.31127995e-01 -6.35852933e-01 6.42522752e-01 2.74673909e-01 4.53443080e-01 -1.21569860e+00 1.55095413e-01 -1.50811121e-01 -3.53030741e-01 1.57058731e-01 2.76493013e-01 -3.35727066e-01 -4.46955174e-01 -1.07535982e+00 7.53071010e-01 3.72766435e-01 9.59372055e-03 -8.15327168e-01 -3.08097512e-01 -8.31007540e-01 2.70288020e-01 5.29869735e-01 -4.41356182e-01 1.03370547e+00 -1.34900415e+00 -1.50361550e+00 8.78365934e-01 9.99283344e-02 -5.22921026e-01 3.89630675e-01 -1.27035946e-01 -4.20622081e-01 6.34426534e-01 -6.52753860e-02 8.17105830e-01 1.27458763e+00 -1.53427422e+00 -7.27993786e-01 -2.71157593e-01 2.88838118e-01 2.14712754e-01 -2.96490461e-01 2.13546529e-01 -1.32200134e+00 -6.95614159e-01 -1.59418359e-01 -8.75057101e-01 5.99777214e-02 -9.04799532e-03 -1.69553071e-01 -6.27497137e-02 7.99043536e-01 -8.56927931e-01 9.84551251e-01 -2.13705015e+00 3.01906109e-01 5.42087972e-01 5.41697480e-02 3.22650164e-01 -6.23345792e-01 5.47252893e-01 7.75397122e-02 -3.80505621e-02 2.02978685e-01 -3.10734868e-01 -2.78438702e-02 -2.96432167e-01 -1.86001226e-01 1.91197753e-01 1.86962754e-01 9.93368685e-01 -5.31452835e-01 -7.41119862e-01 5.81968129e-01 7.25406766e-01 -4.63341653e-01 7.42627978e-02 -1.66200906e-01 2.35539213e-01 -4.55837458e-01 5.45050442e-01 6.96372211e-01 -3.14417332e-01 3.60654384e-01 -6.19655013e-01 -7.53080025e-02 -3.16714227e-01 -8.45168471e-01 1.96869540e+00 -7.39116371e-01 8.74968708e-01 1.53340166e-02 -1.17426813e+00 9.46962297e-01 3.59674752e-01 6.42249465e-01 -1.21736276e+00 4.18560028e-01 8.97704531e-03 -4.22500134e-01 -4.10368949e-01 8.26550603e-01 1.52017623e-01 -2.82919854e-01 1.93871379e-01 4.08000886e-01 -6.13432601e-02 4.29255426e-01 2.65011519e-01 8.30384493e-01 2.08048403e-01 2.27145031e-01 2.60293540e-02 9.38843727e-01 7.88093060e-02 -1.12512801e-02 8.53489101e-01 2.04251602e-01 6.76289439e-01 1.39013782e-01 -2.34725237e-01 -1.12796354e+00 -8.18269193e-01 -5.73578067e-02 1.14812958e+00 2.72338927e-01 -3.38619381e-01 -4.00950491e-01 -5.54985702e-01 -1.98845893e-01 5.85419238e-01 -4.93555546e-01 -1.03538565e-01 -4.22093034e-01 -4.14604276e-01 5.96497618e-02 3.00522774e-01 6.02638066e-01 -1.03916419e+00 -2.13749647e-01 6.94233775e-02 -3.61368805e-01 -1.13212550e+00 -5.43050230e-01 3.88214104e-02 -4.62175965e-01 -7.44702458e-01 -1.06903505e+00 -1.16040552e+00 5.53520322e-01 5.02943456e-01 9.32547331e-01 1.22186661e-01 -6.11274421e-01 5.86092234e-01 -7.56522417e-01 -2.17196777e-01 -3.70743245e-01 -8.00601766e-03 -6.21320724e-01 2.81551629e-01 2.66847610e-01 -1.30122229e-02 -6.13500178e-01 8.32873136e-02 -1.29210699e+00 -1.15340330e-01 1.09073663e+00 9.40534472e-01 5.93530595e-01 7.33311027e-02 2.93590158e-01 -7.02957988e-01 5.95925152e-01 -2.07754716e-01 -7.89991915e-01 5.87633550e-01 -3.92727643e-01 5.49436826e-03 4.47989970e-01 -6.31253183e-01 -1.24558342e+00 2.17811778e-01 -1.90526247e-01 -6.12900674e-01 2.32168008e-02 8.39710832e-01 1.70911342e-01 4.49409522e-02 4.15011823e-01 1.75914258e-01 -2.03470424e-01 -5.98608434e-01 5.22065461e-01 1.11127031e+00 3.60083193e-01 -2.01750606e-01 6.68123186e-01 2.45398834e-01 -1.85087904e-01 -9.91693258e-01 -7.68328309e-01 -9.83262062e-01 -7.32621551e-01 -4.05051559e-01 1.04873502e+00 -1.20118701e+00 -6.38647318e-01 1.81983814e-01 -9.22996104e-01 -3.13126445e-02 -8.36587399e-02 6.98069751e-01 -4.90812540e-01 2.31608853e-01 -7.40974724e-01 -7.41617858e-01 -5.24250329e-01 -1.17290938e+00 1.27465522e+00 5.27244098e-02 3.07170928e-01 -5.73887467e-01 -7.46929944e-02 6.91252768e-01 4.21455085e-01 -2.07461745e-01 7.47631967e-01 -7.13304818e-01 -6.31467462e-01 -4.88257617e-01 -5.60566902e-01 5.05336165e-01 -1.14956260e-01 -2.24146768e-01 -8.53771508e-01 -4.50463265e-01 -2.84099549e-01 -6.15640402e-01 1.21247566e+00 3.28414470e-01 1.00046968e+00 -1.08484522e-01 -2.68239379e-01 2.08245471e-01 1.93463719e+00 5.83033860e-01 8.49871099e-01 2.11335286e-01 3.16436499e-01 5.12115538e-01 8.51294279e-01 3.53677601e-01 1.32049486e-01 1.03371322e+00 3.33025046e-02 -2.60010391e-01 -3.66467118e-01 -9.77788940e-02 1.92855880e-01 6.79146647e-01 4.90782186e-02 -4.63230222e-01 -4.43363339e-01 3.79610062e-01 -1.84119105e+00 -1.10914218e+00 1.74548477e-01 2.14161086e+00 5.12359619e-01 -1.85762420e-01 -9.03338380e-03 9.77848023e-02 6.70954466e-01 1.81496173e-01 -1.76992655e-01 -1.54535532e-01 -1.09886974e-02 3.40640336e-01 4.84626234e-01 4.65130985e-01 -1.40219414e+00 9.40470874e-01 5.97112513e+00 1.14519048e+00 -1.32927096e+00 2.52275944e-01 6.38733029e-01 1.13680892e-01 1.49261095e-02 -2.76987344e-01 -5.18604934e-01 1.60447583e-01 6.04831040e-01 2.11436495e-01 4.89373237e-01 6.77453876e-01 -8.77505913e-02 -2.81827271e-01 -8.70004535e-01 1.06476235e+00 7.32315719e-01 -1.17616558e+00 1.03352129e-01 -1.17770590e-01 6.16449714e-01 -2.18515098e-01 3.11820894e-01 3.60625744e-01 -6.33624792e-02 -6.75895989e-01 8.09770346e-01 6.94730401e-01 7.64705420e-01 -6.48408771e-01 9.41403925e-01 -7.54041411e-03 -1.18954277e+00 -2.93295950e-01 -3.40564519e-01 4.95856643e-01 6.85712770e-02 2.90503591e-01 -5.28116167e-01 7.35888243e-01 5.57565153e-01 6.82704687e-01 -6.47979081e-01 1.10865891e+00 5.74644543e-02 2.51559317e-01 -1.82702243e-02 -9.22563002e-02 2.85547316e-01 -2.53659517e-01 2.92531490e-01 1.41485274e+00 2.93957233e-01 -9.17619988e-02 2.76437759e-01 7.48053253e-01 -1.80553302e-01 6.08195007e-01 -4.73744452e-01 -3.29346061e-02 -4.44575325e-02 1.63399518e+00 -8.40211987e-01 -6.45111382e-01 -6.17103338e-01 1.25417817e+00 9.11521539e-02 3.55362535e-01 -7.71859705e-01 -3.05216283e-01 -2.00997904e-01 -2.27364004e-01 9.34074581e-01 5.40669896e-02 4.70451653e-01 -1.32541835e+00 -1.67984366e-01 -9.06589687e-01 3.58121932e-01 -1.21500278e+00 -1.14462793e+00 6.81653738e-01 4.24475148e-02 -1.38081610e+00 -3.52685481e-01 -6.06909215e-01 -1.36212274e-01 5.84616542e-01 -1.72192049e+00 -1.65493619e+00 -2.20829010e-01 8.04381371e-01 7.39710867e-01 -2.28595838e-01 6.51177704e-01 6.59146547e-01 -1.79367810e-01 4.75544453e-01 2.63182610e-01 2.41468817e-01 8.15666974e-01 -9.49431181e-01 -3.50727409e-01 5.33865690e-01 5.64879000e-01 2.69616157e-01 5.18411100e-01 -5.37756920e-01 -1.49174869e+00 -1.07933831e+00 8.11996758e-01 2.11962670e-01 3.60655993e-01 -2.96883106e-01 -2.50394970e-01 6.24698818e-01 5.60368180e-01 -1.28640324e-01 5.80119669e-01 -6.59561530e-02 -4.12734210e-01 -5.17001152e-01 -1.09781432e+00 2.73438931e-01 2.95804530e-01 -5.67098498e-01 -4.43810135e-01 3.34342569e-01 5.28153062e-01 -5.23781516e-02 -9.54123974e-01 1.34460479e-01 9.21356976e-01 -5.96389890e-01 1.24262488e+00 -2.07286492e-01 5.52126467e-01 -1.56506985e-01 -4.62404579e-01 -9.67997551e-01 -4.75570589e-01 -1.16557516e-01 5.36445796e-01 1.22362053e+00 3.76399308e-01 5.81332780e-02 3.53465497e-01 2.08757222e-02 2.40846589e-01 -1.56050578e-01 -4.11923230e-01 -4.88659739e-01 -2.90285736e-01 -2.89524764e-01 7.33871013e-02 6.14976764e-01 -1.04632795e-01 6.36821866e-01 -7.75414050e-01 -1.46945983e-01 5.40258527e-01 4.51242983e-01 5.74336231e-01 -1.15445495e+00 -4.75829691e-01 -2.11782798e-01 -4.57541794e-01 -1.08426774e+00 -6.11724108e-02 -1.01064324e+00 1.69003531e-01 -1.45408285e+00 7.38348007e-01 -2.20581770e-01 -6.44593298e-01 3.00153971e-01 8.62167031e-02 8.72142673e-01 8.63308251e-01 1.36038452e-01 -9.65469778e-01 1.56288907e-01 1.11484838e+00 -6.59471929e-01 -2.09973752e-01 -2.36275017e-01 -5.02386749e-01 2.63979137e-01 5.03450811e-01 -1.97056293e-01 -4.39938903e-01 -9.37410742e-02 3.34286951e-02 4.89561617e-01 3.66498351e-01 -9.31420326e-01 2.62565404e-01 2.50097513e-01 7.31615305e-01 -1.00700355e+00 7.72807300e-01 -1.15336668e+00 1.05636939e-01 3.84765416e-01 -7.46050298e-01 -1.74201623e-01 -1.38477348e-02 5.56612849e-01 -5.89700639e-01 -4.51624691e-01 4.98297632e-01 -1.56186566e-01 -6.54337108e-01 5.36115095e-02 -5.47358632e-01 -5.19436181e-01 8.95978689e-01 2.30771095e-01 -1.28877148e-01 -7.39278674e-01 -8.57711792e-01 -1.18099362e-01 3.61488238e-02 6.81172609e-01 7.83072352e-01 -1.32970941e+00 -6.74740255e-01 1.41421005e-01 3.67986917e-01 -7.96222687e-01 2.93946326e-01 7.88496435e-01 -4.39279765e-01 6.66300774e-01 -3.10743123e-01 -6.24760091e-01 -1.56957173e+00 1.00570345e+00 -1.45714417e-01 -5.67172050e-01 -4.12930220e-01 3.54292184e-01 2.63500541e-01 5.94751090e-02 2.39388004e-01 -1.05442166e-01 -6.22256815e-01 4.81751472e-01 7.47098863e-01 5.09225056e-02 5.45662120e-02 -9.96145189e-01 -9.88174882e-03 6.88376963e-01 -2.88190246e-01 -5.47087669e-01 1.23343849e+00 -3.63849550e-01 -4.53708731e-02 3.09327871e-01 1.59743750e+00 -1.65098339e-01 -5.45441747e-01 -4.74046260e-01 -1.50222912e-01 -4.06023145e-01 7.08622932e-01 -1.06868660e+00 -1.38592732e+00 7.31536090e-01 9.89992499e-01 1.15301661e-01 1.53633142e+00 -3.62173468e-02 5.41496694e-01 4.05337989e-01 2.53612190e-01 -1.13868165e+00 3.59236240e-01 6.22540116e-02 1.09677804e+00 -1.31729925e+00 1.64027348e-01 -2.47699097e-01 -7.27326453e-01 1.14352202e+00 4.28897031e-02 -1.79001167e-01 8.10681105e-01 5.87702170e-02 1.32437408e-01 -1.64297044e-01 -6.73048854e-01 -6.01536214e-01 7.66636014e-01 2.11210355e-01 3.85583490e-01 -6.76884875e-02 -8.33479464e-01 4.36107844e-01 3.41865242e-01 2.65549481e-01 1.05251364e-01 1.00520730e+00 -4.27585274e-01 -1.25309181e+00 -3.86759430e-01 4.37980115e-01 -7.06753969e-01 -1.71412826e-01 -2.64596552e-01 7.77636528e-01 2.49889530e-02 1.25197566e+00 1.11164071e-01 -4.26164091e-01 1.59715459e-01 -5.08686341e-02 6.19369686e-01 -2.49102756e-01 -7.63887942e-01 7.69756973e-01 7.40355924e-02 -5.42353988e-01 -8.00713480e-01 -4.33214813e-01 -5.65909922e-01 1.80105105e-01 -5.63530624e-01 -7.74596334e-02 1.07458925e+00 7.11433828e-01 7.81775340e-02 5.21882176e-01 7.48369634e-01 -9.58195031e-01 -1.45617381e-01 -9.29315984e-01 -6.91551864e-01 5.76693535e-01 2.27837175e-01 -3.64689142e-01 -5.56037314e-02 3.55187654e-01]
[10.814242362976074, 1.1254209280014038]
673f00a9-42d8-44f7-bd6d-2100cd499caf
multi-video-moment-ranking-with-multimodal
2301.13606
null
https://arxiv.org/abs/2301.13606v1
https://arxiv.org/pdf/2301.13606v1.pdf
Multi-video Moment Ranking with Multimodal Clue
Video corpus moment retrieval~(VCMR) is the task of retrieving a relevant video moment from a large corpus of untrimmed videos via a natural language query. State-of-the-art work for VCMR is based on two-stage method. In this paper, we focus on improving two problems of two-stage method: (1) Moment prediction bias: The predicted moments for most queries come from the top retrieved videos, ignoring the possibility that the target moment is in the bottom retrieved videos, which is caused by the inconsistency of Shared Normalization during training and inference. (2) Latent key content: Different modalities of video have different key information for moment localization. To this end, we propose a two-stage model \textbf{M}ult\textbf{I}-video ra\textbf{N}king with m\textbf{U}l\textbf{T}imodal clu\textbf{E}~(MINUTE). MINUTE uses Shared Normalization during both training and inference to rank candidate moments from multiple videos to solve moment predict bias, making it more efficient to predict target moment. In addition, Mutilmdaol Clue Mining~(MCM) of MINUTE can discover key content of different modalities in video to localize moment more accurately. MINUTE outperforms the baselines on TVR and DiDeMo datasets, achieving a new state-of-the-art of VCMR. Our code will be available at GitHub.
['Xueqi Cheng', 'HuaWei Shen', 'Yanyan Lan', 'Liang Pang', 'Danyang Hou']
2023-01-29
null
null
null
null
['moment-retrieval']
['computer-vision']
[-2.15616520e-03 -4.73256290e-01 -8.39449406e-01 -2.14783043e-01 -1.10154748e+00 -6.19137466e-01 5.35013795e-01 -3.13790321e-01 -4.79694635e-01 4.21242893e-01 4.74909842e-01 7.03524724e-02 -1.75731957e-01 -2.17100844e-01 -9.91962850e-01 -6.71771646e-01 -3.30409378e-01 3.11481148e-01 2.01913387e-01 1.76167518e-01 5.45413733e-01 4.88850474e-01 -1.74934745e+00 7.58435667e-01 3.68346095e-01 1.31060326e+00 4.64907527e-01 9.73075807e-01 -1.10891782e-01 1.38722277e+00 -4.57263857e-01 -3.33431512e-01 1.83083341e-01 -4.02724713e-01 -7.87954926e-01 -2.57645577e-01 6.87629521e-01 -7.50286102e-01 -8.33806157e-01 1.06005764e+00 5.52884459e-01 4.64080602e-01 8.00146043e-01 -1.48686731e+00 -1.64220050e-01 6.66306078e-01 -8.83986294e-01 9.73362803e-01 7.71947086e-01 -1.46030664e-01 9.99329925e-01 -1.09359598e+00 1.13344634e+00 1.36263871e+00 1.59162357e-01 4.49093759e-01 -5.87443411e-01 -8.65361691e-01 3.70059729e-01 7.19628453e-01 -1.58308911e+00 -6.63664043e-01 8.49099696e-01 -3.40773940e-01 9.10204351e-01 5.08596778e-01 2.67756641e-01 1.13875043e+00 3.97364318e-01 1.45907843e+00 6.20370090e-01 -5.50953299e-02 -2.24565014e-01 -1.00004919e-01 -3.92362237e-01 6.10394597e-01 -5.04093468e-01 -3.99878740e-01 -1.02084959e+00 3.77379805e-02 8.21765661e-01 1.81816176e-01 -3.21794331e-01 6.78767487e-02 -1.77016950e+00 6.21018887e-01 -1.64852902e-01 3.84874254e-01 -9.78932828e-02 3.16314816e-01 4.80788320e-01 4.32316005e-01 5.53723752e-01 4.83872704e-02 -6.37199342e-01 -5.67964256e-01 -1.35945797e+00 3.36529195e-01 5.87527156e-01 1.04529488e+00 5.75187504e-01 -3.25067878e-01 -3.26752990e-01 7.06894100e-01 2.96048790e-01 8.17168713e-01 6.85263634e-01 -1.20867956e+00 8.97618651e-01 6.10162579e-02 -1.16807045e-02 -1.27203131e+00 -7.80143365e-02 4.73049693e-02 -5.75805664e-01 -4.71623391e-01 1.73393145e-01 -4.38614115e-02 -5.75261474e-01 1.65128326e+00 2.72138119e-01 7.13601649e-01 -1.89443856e-01 1.03577173e+00 9.45158064e-01 1.12471318e+00 -1.88121617e-01 -5.39721549e-01 1.07490492e+00 -8.71005833e-01 -6.97879076e-01 1.07310317e-01 4.31206197e-01 -1.07547200e+00 8.41388047e-01 5.70693970e-01 -1.17280793e+00 -4.78333831e-01 -5.26037276e-01 -2.33690500e-01 -1.64819732e-01 2.80482709e-01 4.64368612e-01 -1.83973595e-01 -9.28117335e-01 4.80004013e-01 -6.11525059e-01 -2.56913364e-01 1.28713623e-01 2.39600018e-01 -4.55620617e-01 -3.24693501e-01 -1.11356890e+00 3.84980083e-01 4.54569429e-01 2.72450354e-02 -1.13089621e+00 -6.36840880e-01 -8.45956206e-01 -1.99618101e-01 7.65182078e-01 -2.90696144e-01 1.24900007e+00 -7.45290279e-01 -1.18389952e+00 5.43160260e-01 -6.59412026e-01 -2.76256323e-01 7.03077793e-01 -4.91744190e-01 -6.86075926e-01 8.19488049e-01 4.56651300e-01 9.31453884e-01 1.13919830e+00 -7.70202637e-01 -1.00126469e+00 1.50139570e-01 -4.66810502e-02 2.93368727e-01 -2.03522760e-02 3.69403422e-01 -1.10979033e+00 -1.07060778e+00 1.70834586e-01 -7.06273437e-01 2.69031763e-01 -2.66498119e-01 -2.50998765e-01 -3.17860305e-01 9.84268665e-01 -6.17909014e-01 1.73049927e+00 -2.15415311e+00 2.41056830e-01 1.97600693e-01 8.79552886e-02 -2.28000417e-01 -3.19694817e-01 4.75596040e-01 -1.35241583e-01 1.14888959e-01 5.73645890e-01 -1.13838322e-01 -1.04916140e-01 1.17711350e-01 -7.89505839e-01 5.92505276e-01 -1.50659472e-01 7.10917056e-01 -9.03419137e-01 -1.07386065e+00 2.76112318e-01 2.02498332e-01 -5.28279841e-01 1.24474667e-01 -3.53072047e-01 4.76162434e-01 -3.89742553e-01 9.90890205e-01 2.83167094e-01 -2.23007634e-01 -1.72379702e-01 -5.96810341e-01 -8.84254053e-02 5.32740802e-02 -1.34469211e+00 1.89684772e+00 7.60188922e-02 8.46188486e-01 -2.24337041e-01 -7.21061230e-01 2.79940814e-01 5.34659147e-01 1.20827127e+00 -6.24654591e-01 1.19842783e-01 -1.80599522e-02 -4.35847610e-01 -7.40904450e-01 5.24184167e-01 4.36012477e-01 -4.80154119e-02 4.58361894e-01 2.39771917e-01 1.79990500e-01 6.40260458e-01 5.86062133e-01 9.80259776e-01 2.29437292e-01 -2.05904856e-01 1.19749576e-01 4.75727350e-01 -1.58260405e-01 6.35440469e-01 9.79999900e-01 -1.40560985e-01 4.67642665e-01 6.20850921e-01 -1.35503009e-01 -8.59606683e-01 -1.00614882e+00 4.27506715e-02 1.52943993e+00 2.59673566e-01 -5.18841088e-01 -4.25981581e-01 -5.77522755e-01 -3.13485414e-01 3.84378105e-01 -4.43830639e-01 1.85405388e-01 -7.71029651e-01 -4.64506686e-01 5.17205060e-01 2.64667839e-01 3.40419531e-01 -9.88254964e-01 -3.71826857e-01 1.33827791e-01 -9.36178863e-01 -1.25247598e+00 -9.25032675e-01 -2.66488135e-01 -7.93654084e-01 -1.13506675e+00 -7.40842342e-01 -4.41073596e-01 5.11292100e-01 5.11643827e-01 1.09754038e+00 -1.23001657e-01 -1.81671768e-01 6.57288909e-01 -6.30662739e-01 -1.99999347e-01 1.75728098e-01 -4.04668823e-02 3.58482242e-01 -1.00132160e-01 2.05299735e-01 -2.96170056e-01 -8.44339907e-01 5.61775446e-01 -1.08317661e+00 8.00847784e-02 5.76323271e-01 3.12995791e-01 9.11244750e-01 1.28722832e-01 2.90520817e-01 -3.12786251e-01 3.05678874e-01 -7.74709463e-01 -3.76185626e-01 2.60587245e-01 -2.72074401e-01 -1.02921017e-01 3.31597209e-01 -9.49791610e-01 -8.89780819e-01 -1.40459582e-01 -1.52241504e-02 -9.98583257e-01 6.38557971e-02 5.97178042e-01 9.56753734e-03 2.60068744e-01 1.94121540e-01 2.83432692e-01 -3.20712566e-01 -5.35456300e-01 2.66502082e-01 3.39176357e-01 7.59694815e-01 -5.89676023e-01 6.28713250e-01 5.53948641e-01 -1.11516006e-01 -5.76835990e-01 -8.56635511e-01 -1.02558482e+00 -5.03339827e-01 -6.15209103e-01 7.37390697e-01 -1.29235947e+00 -7.65939057e-01 2.23951757e-01 -8.98597360e-01 -2.46803835e-01 3.66033539e-02 8.78231227e-01 -6.47880077e-01 7.31785774e-01 -9.09025967e-01 -4.24221426e-01 -2.85826892e-01 -1.04910278e+00 1.19540179e+00 5.85994609e-02 -4.89669889e-02 -5.96293747e-01 3.51779647e-02 3.42723966e-01 -2.41709962e-01 1.02749802e-01 5.19705415e-01 -6.13135338e-01 -7.90497899e-01 -2.92584360e-01 -2.44312406e-01 6.33413494e-02 -2.62825429e-01 3.23047101e-01 -6.38762534e-01 -3.71606678e-01 -1.13054261e-01 -2.55874723e-01 7.97436476e-01 6.14957690e-01 1.30280268e+00 -5.07370591e-01 -2.70388544e-01 6.15951955e-01 1.17511451e+00 2.60111362e-01 6.72327399e-01 2.45371684e-01 6.38403237e-01 1.33186549e-01 1.31618047e+00 6.61181331e-01 4.11589354e-01 4.47196722e-01 2.42297947e-01 2.75382757e-01 1.31800547e-01 -1.53048128e-01 8.53805244e-01 1.04355669e+00 -1.97803617e-01 -4.57781762e-01 -6.60747170e-01 5.91714382e-01 -2.13660026e+00 -1.50655627e+00 1.44551218e-01 1.94543898e+00 6.70448899e-01 4.33703959e-02 9.22200754e-02 -1.84806705e-01 6.63214028e-01 5.01422882e-01 -4.93318170e-01 7.54134357e-02 -1.03716590e-01 -3.45232666e-01 2.85688579e-01 1.06064983e-01 -1.20045972e+00 8.67949843e-01 5.73859501e+00 1.27726817e+00 -1.12326884e+00 1.84348911e-01 6.09625816e-01 -7.37783968e-01 -1.21106558e-01 -2.17781350e-01 -9.94480550e-01 6.83050036e-01 7.80367613e-01 9.42114741e-02 6.15036845e-01 9.00481582e-01 3.35830480e-01 -5.77175796e-01 -1.22042572e+00 1.42064989e+00 5.09155273e-01 -1.40125203e+00 2.68205762e-01 -1.26958102e-01 8.28969181e-01 3.57262105e-01 1.30105183e-01 4.21935290e-01 -3.58181626e-01 -6.07479513e-01 1.02393603e+00 9.31141257e-01 1.03155541e+00 -8.29569638e-01 5.75787485e-01 4.83165830e-01 -1.63684142e+00 -3.60987559e-02 -5.01048923e-01 2.79617757e-01 1.39683560e-01 4.19037849e-01 -6.29165649e-01 5.06864965e-01 1.08064425e+00 8.44688594e-01 -6.13962352e-01 8.84893060e-01 4.67792265e-02 3.09409499e-01 -3.13737988e-01 2.13741273e-01 1.79473028e-01 3.99318635e-02 6.78841054e-01 1.31527877e+00 4.99964982e-01 1.81981295e-01 3.14948618e-01 2.52032191e-01 -2.62027264e-01 5.70839755e-02 -3.66759181e-01 -2.28842318e-01 4.16785121e-01 1.09273076e+00 -8.43821168e-01 -4.98916328e-01 -5.03957927e-01 9.95803535e-01 -1.09131023e-01 4.51513231e-01 -1.23257494e+00 -3.21825087e-01 4.14901942e-01 -3.88847925e-02 4.13242310e-01 -5.46959117e-02 6.15973771e-01 -1.77253270e+00 4.30709757e-02 -9.14541006e-01 6.66047275e-01 -9.72492576e-01 -1.19037473e+00 1.41008884e-01 4.97665465e-01 -1.50457788e+00 -6.80864632e-01 -3.88178021e-01 -3.30403924e-01 4.59528029e-01 -1.27986479e+00 -8.73044550e-01 3.11436784e-02 1.25766313e+00 8.87256742e-01 -2.10521758e-01 1.69882804e-01 6.89328015e-01 -3.61773372e-01 3.97822469e-01 2.30118915e-01 2.41125390e-01 1.12912655e+00 -9.71909702e-01 4.34561260e-03 1.10395193e+00 2.57920623e-01 5.03055751e-01 6.56512558e-01 -9.20756757e-01 -1.71564090e+00 -9.97661650e-01 7.86279976e-01 -5.65853715e-01 7.83502758e-01 -9.81521532e-02 -4.86018509e-01 7.53353715e-01 -2.05052599e-01 5.57817183e-02 6.62885070e-01 -2.07069293e-01 -2.74775982e-01 -2.39182457e-01 -7.10439086e-01 7.12283254e-01 8.93250525e-01 -9.53113258e-01 -4.22511309e-01 4.71610457e-01 6.47811592e-01 -4.69684631e-01 -1.09118736e+00 4.55222666e-01 7.50218332e-01 -7.70777404e-01 1.11149359e+00 -2.53802747e-01 6.19645298e-01 -4.34377104e-01 -2.37521753e-01 -4.72973764e-01 2.37612143e-01 -6.24278367e-01 -8.28846514e-01 1.18321359e+00 5.18566407e-02 5.72721735e-02 4.78298068e-01 3.58574867e-01 8.43967199e-02 -6.36991978e-01 -1.07704091e+00 -4.78316844e-01 -6.21357679e-01 -8.71809185e-01 3.24129939e-01 9.26094115e-01 -1.45116867e-03 6.38875291e-02 -8.73381138e-01 1.02396816e-01 4.68740851e-01 3.21091264e-01 8.46789956e-01 -7.73208559e-01 -2.71256775e-01 -2.27459565e-01 -1.37768999e-01 -1.50720119e+00 7.98211321e-02 -5.57904780e-01 1.05962865e-01 -1.23322523e+00 6.83570683e-01 8.05986077e-02 -4.86638665e-01 1.91379264e-01 1.37203589e-01 3.51971567e-01 4.70935494e-01 7.22392797e-01 -1.44393170e+00 7.91092589e-02 1.27570105e+00 2.13109665e-02 -6.43013865e-02 -1.20545015e-01 -1.63933158e-01 8.45337749e-01 3.97533000e-01 -5.68177402e-01 -5.31746149e-01 -4.46929008e-01 5.64739645e-01 6.82646334e-01 3.24415743e-01 -9.98941779e-01 4.33545947e-01 -3.04855704e-01 7.43781090e-01 -1.30217838e+00 4.21455383e-01 -8.10268879e-01 2.67827541e-01 6.50527328e-02 -2.48104379e-01 3.09971601e-01 -8.85419622e-02 7.30117619e-01 -4.16133136e-01 -2.09540114e-01 2.54110277e-01 -2.63038069e-01 -1.00168097e+00 6.71817005e-01 -4.67973530e-01 9.90267191e-03 6.67321980e-01 -9.96094793e-02 -4.00945604e-01 -7.15792596e-01 -6.18382990e-01 2.56840587e-01 2.51119941e-01 6.28573477e-01 7.88083434e-01 -1.31180108e+00 -3.29640150e-01 -3.20838183e-01 -4.25764062e-02 -1.22255236e-01 3.80215704e-01 1.15662026e+00 -4.02284145e-01 3.94772619e-01 7.17195645e-02 -8.02775383e-01 -1.23914731e+00 7.42548585e-01 -1.32231429e-01 -2.35285327e-01 -3.23652178e-01 7.66848922e-01 1.37737155e-01 4.40011844e-02 4.81020242e-01 -4.42400604e-01 -2.12005064e-01 4.56843078e-01 7.80048192e-01 5.44449270e-01 -3.44097376e-01 -1.01346517e+00 -3.20917904e-01 5.39499402e-01 -2.58979946e-01 -3.30116063e-01 1.17245388e+00 -4.76345837e-01 -1.74512982e-01 6.48834348e-01 1.42836452e+00 1.80724353e-01 -1.26429069e+00 -2.18096450e-01 -2.04616174e-01 -6.67179167e-01 -4.81424183e-02 -4.96174991e-01 -9.40822244e-01 5.00724852e-01 6.33164287e-01 -3.50277275e-01 1.22596002e+00 1.07638188e-01 9.30970371e-01 6.58787191e-01 4.80758756e-01 -1.42938221e+00 4.12530243e-01 4.67848420e-01 7.30695128e-01 -1.39891911e+00 3.89333576e-01 -3.52770276e-02 -6.73309863e-01 1.15345716e+00 3.66314620e-01 1.34978652e-01 6.13274753e-01 -2.18862012e-01 -1.99191511e-01 -2.09741086e-01 -9.00525033e-01 -1.29838526e-01 7.07722008e-01 -3.88683788e-02 2.70790040e-01 -3.53763491e-01 -2.37600461e-01 4.55704480e-01 1.40121862e-01 4.76274937e-02 1.78119808e-01 1.05406940e+00 -3.56598973e-01 -7.43602991e-01 -4.24717218e-01 5.97692430e-01 -1.01861000e+00 -1.75330803e-01 7.56482109e-02 7.88189292e-01 1.59725264e-01 1.06262445e+00 1.00349896e-01 -3.26008469e-01 -1.21111266e-01 -1.24758892e-01 3.40855658e-01 -1.66119322e-01 -1.65132374e-01 9.58934069e-01 -2.09911272e-01 -1.05067813e+00 -8.41571391e-01 -8.85767341e-01 -1.14983523e+00 -4.75271165e-01 -3.08511019e-01 1.42154351e-01 4.04201090e-01 8.95726800e-01 2.50285774e-01 2.08148658e-01 6.87578440e-01 -9.99287486e-01 -9.67103913e-02 -7.63127446e-01 -5.74340522e-01 4.64415193e-01 3.02695870e-01 -6.32045925e-01 -3.68630052e-01 4.85258102e-01]
[10.13213062286377, 0.7608920931816101]
7a1920c7-4185-4c86-90ed-12106de0c5b5
deep-event-stereo-leveraged-by-event-to-image
null
null
https://www.researchgate.net/publication/346617073_Deep_Event_Stereo_Leveraged_by_Event-to-Image_Translation
https://www.researchgate.net/publication/346617073_Deep_Event_Stereo_Leveraged_by_Event-to-Image_Translation
Deep Event Stereo Leveraged by Event-to-Image Translation
Depth estimation in real-world applications requires precise responses to fast motion and challenging lighting conditions. Event cameras use bio-inspired event-driven sensors that provide instantaneous and asynchronous information of pixel-level log intensity changes, which makes them suitable for depth estimation in such challenging conditions. However, as the event cameras primarily provide asynchronous and spatially sparse event data, it is hard to provide accurate dense disparity map in stereo event camera setups - especially in estimating disparities on local structures or edges. In this study, we develop a novel deep event stereo network that reconstructs spatial intensity image features from embedded event streams and leverages the event features using the reconstructed image features to compute dense disparity maps. To this end, we propose a novel event-to-image translation network with a cross-semantic attention mechanism that calculates the global semantic context of the event features for the intensity image reconstruction. In addition, a feature aggregation module is developed for accurate disparity estimation, which modulates the event features with the reconstructed image features by a stacked dilated spatially-adaptive denormalization mechanism. Experimental results reveal that our method can outperform the state-of-the-art methods by significant margins both in quantitative and qualitative measures.
['Yong Ju Jung', 'S. M. Nadim Uddin', 'Hae Woong Jang', 'Soikat Hasan Ahmed']
2021-02-02
null
null
null
null
['event-based-vision']
['computer-vision']
[ 6.23124242e-01 -4.75795507e-01 1.22852206e-01 -5.78579426e-01 -7.05491066e-01 -1.96069822e-01 4.60927725e-01 1.56180263e-01 -5.48834264e-01 7.30169058e-01 4.33819771e-01 4.71805215e-01 2.68405639e-02 -1.07461751e+00 -1.03436172e+00 -7.67885685e-01 2.15536267e-01 -1.57723367e-01 4.52395141e-01 1.75989175e-03 5.80150247e-01 3.99679393e-01 -2.11677098e+00 4.50411886e-01 6.97574496e-01 1.40096521e+00 4.31076169e-01 8.00077021e-01 -1.15066230e-01 1.25207210e+00 -4.11425024e-01 -1.61082670e-02 1.64859802e-01 -4.97974873e-01 -1.45731241e-01 6.91399053e-02 5.20406723e-01 -8.55744421e-01 -5.82513273e-01 8.66495848e-01 6.83792412e-01 1.12575009e-01 2.85504967e-01 -1.06321788e+00 -1.33994415e-01 5.16806319e-02 -6.60184622e-01 5.66699505e-01 6.48755074e-01 3.76671702e-01 6.06023490e-01 -8.93718183e-01 7.56523967e-01 9.33809161e-01 3.77449512e-01 2.91747421e-01 -9.82576072e-01 -7.31762469e-01 2.82460172e-02 6.22887969e-01 -1.32992387e+00 -6.66541696e-01 9.10531938e-01 -2.00784549e-01 1.10522687e+00 2.67305635e-02 9.18321431e-01 9.73041594e-01 5.20707846e-01 5.51331997e-01 1.00536966e+00 -8.35091472e-02 5.29707074e-01 -4.15728509e-01 -4.27133560e-01 4.57264155e-01 1.15897888e-02 3.58526826e-01 -1.38167465e+00 3.17664832e-01 1.12478852e+00 3.89646769e-01 -5.56116164e-01 6.69848844e-02 -1.38073623e+00 4.07382429e-01 5.87931335e-01 7.52601656e-04 -7.57547021e-01 5.72362661e-01 2.42612347e-01 6.55008554e-02 4.85601008e-01 -1.26687884e-01 -3.78865659e-01 -3.68549883e-01 -8.40792358e-01 -1.69469006e-02 4.15386379e-01 6.71372414e-01 1.14758968e+00 -9.36574265e-02 -6.62685335e-02 3.29401851e-01 7.25911483e-02 4.44014907e-01 5.15108407e-01 -1.22521758e+00 4.55979288e-01 5.86691260e-01 1.07887730e-01 -1.06170332e+00 -2.69578755e-01 -3.00042308e-03 -7.52772748e-01 2.81574875e-01 2.81685531e-01 1.77764490e-01 -6.32867634e-01 1.62539065e+00 4.62151378e-01 9.33012962e-01 -6.89808354e-02 1.17564416e+00 7.39215076e-01 7.34678447e-01 -5.63248992e-02 -4.02801126e-01 1.22599602e+00 -4.47332919e-01 -7.65679657e-01 -3.20485145e-01 1.96569506e-02 -5.43624759e-01 8.77122641e-01 3.08967859e-01 -1.22909284e+00 -4.99357581e-01 -1.12798309e+00 -4.90329832e-01 -1.84955627e-01 -3.37963849e-01 7.67430246e-01 6.40689135e-02 -9.81004894e-01 4.53623325e-01 -1.03021157e+00 -2.17072293e-01 3.66927207e-01 3.22924525e-01 -2.12623149e-01 -2.35403776e-01 -1.03048217e+00 4.13288593e-01 2.24604115e-01 1.60597414e-01 -8.88293266e-01 -8.80892456e-01 -1.14764178e+00 -2.09693983e-02 -6.38175234e-02 -8.62324834e-01 1.02271092e+00 -9.80632484e-01 -1.70284951e+00 8.27470601e-01 -5.82504928e-01 -3.36447597e-01 4.02337372e-01 2.47821081e-02 -6.19212016e-02 5.76661944e-01 2.26429120e-01 9.41778958e-01 6.95739985e-01 -9.12659943e-01 -1.09820068e+00 -4.06072617e-01 3.00515220e-02 4.97340620e-01 -2.83093065e-01 -2.34879360e-01 -5.55270672e-01 -4.15148079e-01 2.86674291e-01 -2.57978439e-01 -1.35837749e-01 5.38576066e-01 6.34131506e-02 3.18209499e-01 5.65379322e-01 -3.35841298e-01 1.02300549e+00 -2.25929189e+00 1.84719136e-03 -4.47496437e-02 2.94516593e-01 -4.87416834e-01 1.74843892e-01 1.67370215e-01 2.25279033e-01 -6.66546583e-01 -2.30005682e-01 -4.70841169e-01 -3.60165417e-01 1.97650686e-01 -3.69558424e-01 6.44182503e-01 2.34604865e-01 8.42553616e-01 -1.12677073e+00 -6.37304187e-01 9.10255313e-01 8.59237373e-01 -5.83747208e-01 3.48282039e-01 -1.90770924e-01 7.98750639e-01 -4.05710936e-01 7.04071939e-01 6.95149064e-01 -2.79043943e-01 -1.38636604e-01 -4.47118521e-01 -3.96629453e-01 2.23523453e-01 -1.06799245e+00 2.26336026e+00 -6.91587329e-01 9.17538822e-01 -2.04719290e-01 -6.18364632e-01 7.73189783e-01 8.61010998e-02 7.76798964e-01 -1.25486779e+00 3.24233890e-01 2.74767637e-01 -7.01219857e-01 -4.64145303e-01 6.71323776e-01 2.34302785e-02 -7.24349171e-02 2.27655128e-01 -2.88609952e-01 -3.14887166e-01 -2.03509405e-02 5.88439815e-02 1.24403417e+00 1.17602326e-01 2.02883005e-01 2.38237485e-01 4.11766022e-01 -2.78521955e-01 9.08738673e-01 2.63160050e-01 -2.69595236e-01 1.02668810e+00 3.83491926e-02 -5.13065517e-01 -8.62203300e-01 -1.28925872e+00 -1.70006692e-01 7.39577234e-01 9.19988871e-01 -1.35859564e-01 -5.68040133e-01 -1.22470381e-02 -1.51002586e-01 3.94530982e-01 -5.56863248e-01 -2.31040806e-01 -4.93967801e-01 -6.88216746e-01 6.93832859e-02 7.35536516e-01 9.41579759e-01 -9.83668089e-01 -1.07070959e+00 5.57518482e-01 -5.24758935e-01 -1.40204394e+00 -6.08934224e-01 1.42520979e-01 -8.53076577e-01 -9.87369061e-01 -3.74905467e-01 -7.65641451e-01 5.72725296e-01 5.09503245e-01 9.56086397e-01 -2.91034192e-01 -4.37996656e-01 2.87107736e-01 -2.72663295e-01 -2.87348032e-01 2.81632572e-01 -4.04864252e-01 -3.93327594e-01 3.68506730e-01 4.37028289e-01 -9.99946535e-01 -1.42184389e+00 2.45398074e-01 -1.21395886e+00 3.76239479e-01 3.66146713e-01 6.18918657e-01 1.01006234e+00 7.98064321e-02 3.16560745e-01 -2.64283329e-01 -9.91654992e-02 -3.79147321e-01 -8.06551993e-01 -2.71408081e-01 -2.42780820e-01 -1.48170039e-01 4.78387743e-01 -2.28172079e-01 -1.33537865e+00 2.48158023e-01 -5.10745449e-03 -4.22661632e-01 2.80711167e-02 2.22067729e-01 -1.53657585e-01 -5.71918152e-02 5.76715052e-01 4.22659278e-01 -2.75589108e-01 4.67925705e-02 1.68492645e-02 6.35975063e-01 9.92935240e-01 -2.62087762e-01 3.32953483e-01 1.33099544e+00 5.50689325e-02 -5.08483946e-01 -7.84600139e-01 -5.55242419e-01 -2.03872144e-01 -5.26505709e-01 9.80641246e-01 -1.45488548e+00 -8.32591593e-01 9.75458503e-01 -1.33610058e+00 -4.98938859e-01 -3.67311031e-01 5.48653364e-01 -7.46147990e-01 1.92378610e-01 -9.53708231e-01 -5.24069726e-01 -2.52172858e-01 -1.09883845e+00 1.77146149e+00 5.61764002e-01 3.77953835e-02 -8.55662405e-01 4.25495282e-02 2.63110399e-01 4.22008187e-01 5.57112098e-01 2.54864812e-01 5.55482209e-01 -1.36278248e+00 9.84133333e-02 -5.25263667e-01 -3.64900241e-03 1.77246764e-01 -1.36272326e-01 -1.31612372e+00 1.22951016e-01 1.27727807e-01 -5.06742448e-02 8.20836723e-01 8.83844614e-01 1.37816560e+00 2.65632778e-01 -1.78655699e-01 1.17675161e+00 1.74275994e+00 6.61161020e-02 1.16677439e+00 3.80879968e-01 7.59089649e-01 4.34233010e-01 7.46974170e-01 1.02795815e+00 8.32478285e-01 6.63806677e-01 6.49960399e-01 -1.01423144e-01 -2.88732618e-01 -3.85443151e-01 4.37377751e-01 6.43275440e-01 1.57588854e-01 -3.29015195e-01 -4.63637233e-01 6.60618305e-01 -1.79558992e+00 -9.81613994e-01 -2.07101062e-01 2.21678352e+00 9.57976639e-01 7.21403724e-03 -5.63563645e-01 1.72197536e-01 8.53293180e-01 3.11820090e-01 -7.55083323e-01 3.90122943e-02 -5.15529037e-01 3.98701102e-01 5.50544679e-01 4.34482694e-01 -7.95184970e-01 7.19975173e-01 5.24489737e+00 6.32407486e-01 -1.34470117e+00 7.91769475e-02 7.27897108e-01 -4.44949746e-01 -4.28348690e-01 -8.71822909e-02 -5.50558984e-01 8.05353880e-01 8.39117706e-01 -1.26179591e-01 4.17912275e-01 3.74214023e-01 7.67794371e-01 -7.07188308e-01 -1.15910327e+00 1.60786164e+00 4.55194190e-02 -1.50911546e+00 -1.50743842e-01 8.47454816e-02 9.80036259e-01 1.05709217e-01 -1.56125888e-01 -4.97093797e-01 1.48464749e-02 -6.96351528e-01 8.68171573e-01 7.32721448e-01 9.72856164e-01 -7.12746322e-01 4.43205565e-01 3.91859822e-02 -1.51552188e+00 -5.17788418e-02 -2.85407066e-01 -3.44335109e-01 5.30827165e-01 1.03558302e+00 -2.38169238e-01 2.30679631e-01 9.43667114e-01 1.25346136e+00 -1.47166893e-01 9.42754507e-01 -1.44287243e-01 9.05955806e-02 -3.87300253e-01 3.15064132e-01 -2.01265767e-01 -1.20772213e-01 2.39792183e-01 8.03043962e-01 5.11323094e-01 1.96095869e-01 -3.35623026e-01 7.31548846e-01 -1.22123674e-01 -2.63598531e-01 -4.89422232e-01 4.78957683e-01 6.25032663e-01 1.10760832e+00 -6.96905315e-01 -2.85438985e-01 -4.62158650e-01 1.47754419e+00 1.13441736e-01 3.12697917e-01 -1.04457700e+00 -3.93058836e-01 9.43310380e-01 9.05342251e-02 3.19653749e-01 -1.67018652e-01 -4.51284438e-01 -1.33225441e+00 3.30587804e-01 -2.25046322e-01 3.13196599e-01 -1.20586514e+00 -1.11662722e+00 2.36897618e-01 -3.94348860e-01 -1.44240654e+00 -2.49804989e-01 -2.99805194e-01 -6.62065208e-01 5.27338326e-01 -2.11931562e+00 -8.53460193e-01 -1.19647396e+00 9.27120805e-01 5.78258157e-01 3.86109799e-01 3.28877658e-01 6.14100397e-01 -4.39783037e-01 1.88301235e-01 1.10281315e-02 -1.33252263e-01 7.12251663e-01 -9.17562187e-01 2.61068851e-01 8.95742238e-01 -2.00977847e-01 -5.89042529e-02 5.34669101e-01 -4.53992128e-01 -1.80803418e+00 -1.25674677e+00 6.72844887e-01 -2.07178608e-01 3.67749095e-01 -3.00210059e-01 -6.04354501e-01 3.86510760e-01 -1.00744382e-01 5.68025827e-01 2.68701077e-01 -8.42678905e-01 -6.11014590e-02 -5.88320971e-01 -1.15851581e+00 4.43530530e-01 1.43165720e+00 -7.40522146e-01 -5.81682585e-02 1.79234236e-01 7.34858215e-01 -6.51450694e-01 -8.26327324e-01 3.51523608e-01 5.70765257e-01 -1.49237192e+00 1.04110801e+00 4.69455600e-01 7.51951098e-01 -7.16411114e-01 -3.52913231e-01 -8.14543843e-01 1.49628550e-01 -3.96154165e-01 -1.22747876e-01 1.13790751e+00 -1.38581693e-01 -7.72860706e-01 9.17091906e-01 4.66437697e-01 -1.87932834e-01 -4.10711497e-01 -1.25817847e+00 -3.11653346e-01 -7.35190034e-01 -7.02765048e-01 6.19594991e-01 6.19999588e-01 -3.21433157e-01 -9.60768387e-02 -6.86609512e-03 2.64689445e-01 8.52289021e-01 2.73133576e-01 5.79249561e-01 -7.50124693e-01 -1.87026590e-01 -1.81084499e-01 -9.96846020e-01 -1.28621852e+00 1.15657382e-01 -4.38568920e-01 2.84438461e-01 -1.39295900e+00 1.08080730e-01 -7.94428885e-02 -2.64228284e-01 -1.58244386e-01 -2.87134141e-01 6.13909423e-01 -3.58496726e-01 9.15277600e-02 -7.38022983e-01 8.08222711e-01 1.08920252e+00 9.23078060e-02 -1.79286659e-01 -5.89484811e-01 -3.33695352e-01 5.14132857e-01 3.57194453e-01 -2.85264313e-01 -4.65949357e-01 -7.22249687e-01 4.54256713e-01 2.87374258e-01 7.11430013e-01 -1.25548398e+00 6.38301671e-01 -2.75027484e-01 6.15242481e-01 -5.59980452e-01 5.79863608e-01 -8.08171034e-01 3.31220031e-01 1.22807927e-01 -1.82779171e-02 -6.34737462e-02 3.23480293e-02 8.28889012e-01 -5.93560755e-01 3.92766088e-01 6.63102984e-01 9.66370944e-03 -1.21482098e+00 4.92841721e-01 -1.75508022e-01 3.28723043e-02 1.16497731e+00 -7.40475953e-01 -3.46172392e-01 -4.59870011e-01 2.49682497e-02 9.03456137e-02 8.44587922e-01 7.74257779e-02 1.02239633e+00 -1.36411047e+00 -5.16668320e-01 4.05298859e-01 3.30857098e-01 5.31543016e-01 6.26800418e-01 8.77413690e-01 -8.32264781e-01 -8.98521468e-02 -3.03350717e-01 -1.02724385e+00 -8.12509596e-01 9.72930938e-02 3.84126484e-01 1.83061197e-01 -7.00950205e-01 7.83113897e-01 6.02926195e-01 2.26731777e-01 9.47672576e-02 -6.06874466e-01 1.84943065e-01 2.50558499e-02 7.99875021e-01 2.79176086e-01 1.59760922e-01 -4.01755035e-01 -2.74181455e-01 9.26783502e-01 3.22311074e-01 -2.80195028e-01 1.48406398e+00 -6.85307741e-01 -5.44730667e-03 4.07639444e-01 1.38389885e+00 -2.09747180e-01 -2.01309514e+00 -1.95861369e-01 -5.42586446e-01 -8.69688153e-01 4.84817892e-01 -2.61163235e-01 -1.36337876e+00 7.79659510e-01 7.61393070e-01 -4.57503796e-01 1.83971858e+00 -1.53531000e-01 1.25311661e+00 -1.04366541e-01 6.29304588e-01 -1.11295652e+00 2.89138079e-01 2.00044289e-01 4.83922333e-01 -1.32027888e+00 -1.40800387e-01 -5.28321505e-01 -2.72307992e-01 1.07480812e+00 5.59881091e-01 -3.91571075e-02 5.23885071e-01 4.90485609e-01 -6.44777790e-02 -2.64759123e-01 -9.43669796e-01 -2.14951038e-01 -1.92261860e-01 5.11403084e-01 1.16956174e-01 -2.87196577e-01 1.79559048e-02 8.99387076e-02 3.94695951e-03 4.11601633e-01 4.97123808e-01 1.01040423e+00 -4.13443029e-01 -6.58282876e-01 -3.08412582e-01 3.23882610e-01 -1.67719573e-01 -2.21417204e-01 1.85960159e-01 2.69656599e-01 2.05508456e-01 8.47799361e-01 6.89915121e-01 -3.78437310e-01 3.65440577e-01 -4.51236606e-01 5.37110150e-01 -2.97187954e-01 -3.39542538e-01 -1.37375489e-01 -1.56118572e-01 -1.12513638e+00 -6.73090279e-01 -7.84883499e-01 -1.58153737e+00 -4.20329213e-01 -2.37390343e-02 -6.25266314e-01 9.23480570e-01 8.18644762e-01 6.10498667e-01 5.81473172e-01 9.17396247e-01 -1.18223131e+00 3.04312468e-01 -5.03723383e-01 -5.30839503e-01 5.95627546e-01 4.30308580e-01 -5.76788187e-01 -7.10197270e-01 4.05032188e-01]
[9.096512794494629, -1.7228761911392212]
abcfee95-4cd6-469d-a4bc-dd8cb0857ddc
structural-segmentation-and-labeling-of-tabla
2211.0879
null
https://arxiv.org/abs/2211.08790v1
https://arxiv.org/pdf/2211.08790v1.pdf
Structural Segmentation and Labeling of Tabla Solo Performances
Tabla is a North Indian percussion instrument used as an accompaniment and an exclusive instrument for solo performances. Tabla solo is intricate and elaborate, exhibiting rhythmic evolution through a sequence of homogeneous sections marked by shared rhythmic characteristics. Each section has a specific structure and name associated with it. Tabla learning and performance in the Indian subcontinent is based on stylistic schools called gharana-s. Several compositions by various composers from different gharana-s are played in each section. This paper addresses the task of segmenting the tabla solo concert into musically meaningful sections. We then assign suitable section labels and recognize gharana-s from the sections. We present a diverse collection of over 38 hours of solo tabla recordings for the task. We motivate the problem and present different challenges and facets of the tasks. Inspired by the distinct musical properties of tabla solo, we compute several rhythmic and timbral features for the segmentation task. This work explores the approach of automatically locating the significant changes in the rhythmic structure by analyzing local self-similarity in an unsupervised manner. We also explore supervised random forest and a convolutional neural network trained on hand-crafted features. Both supervised and unsupervised approaches are also tested on a set of held-out recordings. Segmentation of an audio piece into its structural components and labeling is crucial to many music information retrieval applications like repetitive structure finding, audio summarization, and fast music navigation. This work helps us obtain a comprehensive musical description of the tabla solo concert.
['Hema A Murthy', 'R Aravind', 'Gowriprasad R']
2022-11-16
null
null
null
null
['music-information-retrieval']
['music']
[ 4.85459507e-01 -3.57955337e-01 -1.80203885e-01 4.80250418e-02 -9.78646576e-01 -1.23148727e+00 9.64431912e-02 9.49517265e-02 -5.32087013e-02 5.19384623e-01 5.07500708e-01 1.74254730e-01 -7.26144910e-01 -4.38965708e-01 -2.42079496e-01 -7.49341786e-01 -4.61712480e-01 7.22950220e-01 -9.47202891e-02 -5.09282589e-01 6.27243638e-01 5.37412524e-01 -1.57906318e+00 6.68826997e-01 4.81252164e-01 1.04906452e+00 3.87619466e-01 9.33378279e-01 -1.06499158e-01 5.59461236e-01 -1.08836472e+00 3.24090086e-02 1.98072895e-01 -1.00779784e+00 -1.05648029e+00 3.26297104e-01 4.35580313e-01 3.22898835e-01 1.46796122e-01 6.38963759e-01 5.69662869e-01 1.96926087e-01 7.31883585e-01 -8.18850994e-01 1.85450897e-01 1.45990813e+00 -5.15809596e-01 2.54456967e-01 3.97794753e-01 -4.73165005e-01 1.65082395e+00 -5.05694628e-01 6.91613615e-01 6.85766518e-01 1.17193031e+00 1.39061481e-01 -1.27809036e+00 -7.18554020e-01 -6.09391034e-01 2.29176730e-01 -1.35064232e+00 -2.23356619e-01 1.17692697e+00 -6.86864674e-01 5.67700863e-01 6.30842388e-01 1.11084139e+00 6.14348769e-01 -1.82425708e-01 9.74635482e-01 8.14456761e-01 -5.86781144e-01 1.33882329e-01 -6.28450453e-01 -6.63712323e-02 4.75090623e-01 -2.52846777e-01 -4.99207139e-01 -1.07828736e+00 -1.43427819e-01 6.30557418e-01 -5.10578215e-01 -3.00946176e-01 5.54237477e-02 -1.24455655e+00 6.55848026e-01 8.09454247e-02 6.49177551e-01 -1.70459464e-01 6.45079538e-02 9.05691147e-01 3.68414938e-01 -1.27140552e-01 1.14712715e+00 -4.71997440e-01 -4.81889904e-01 -1.67463315e+00 5.39356649e-01 8.05061996e-01 4.84456211e-01 6.01578057e-01 1.05340146e-01 7.41016120e-02 1.28047526e+00 -2.92957276e-01 -1.09069727e-01 7.88548887e-01 -1.29326212e+00 2.48488262e-01 3.49354982e-01 -3.67065102e-01 -7.07192302e-01 -5.54104328e-01 -6.45207405e-01 -6.91046059e-01 3.82924639e-02 5.51899433e-01 4.10493128e-02 -4.35783029e-01 1.50042653e+00 -6.49361908e-02 -1.62982494e-01 -3.33063185e-01 7.13476121e-01 9.44393277e-01 6.05552077e-01 -6.03033662e-01 -4.51673567e-01 1.47994006e+00 -8.98003519e-01 -5.94047785e-01 3.83644849e-01 2.98114121e-01 -1.31399357e+00 1.17538166e+00 9.29542422e-01 -1.19143796e+00 -6.93058848e-01 -1.23555160e+00 1.55860577e-02 3.76462072e-01 3.37026894e-01 4.28361386e-01 4.92114723e-01 -3.98168355e-01 1.12964189e+00 -4.72781837e-01 -3.23384285e-01 2.74698019e-01 3.36978197e-01 -1.59300476e-01 8.46677542e-01 -7.28538156e-01 4.82192039e-02 6.46154106e-01 -1.99431530e-03 -5.21613836e-01 -7.04502761e-01 -5.23941457e-01 -9.49902926e-03 3.55613202e-01 -2.25652412e-01 1.57089984e+00 -1.09686935e+00 -1.69050646e+00 1.16888297e+00 3.08420897e-01 -5.56624472e-01 2.39544421e-01 -2.61611253e-01 -3.04953635e-01 3.88379484e-01 1.92406312e-01 2.05953702e-01 8.04459155e-01 -7.84432828e-01 -7.61326313e-01 -1.73420623e-01 -4.48358893e-01 2.46550605e-01 -1.24496169e-01 2.75951922e-01 -3.42804164e-01 -1.32537055e+00 6.49446130e-01 -1.05943036e+00 2.08629131e-01 -7.87542820e-01 -8.40372324e-01 -1.85132965e-01 7.20564067e-01 -7.19043970e-01 1.64458919e+00 -1.97384596e+00 1.75715029e-01 4.06024694e-01 -5.11004888e-02 -9.89926606e-02 1.41772777e-01 7.34717250e-01 -9.84530821e-02 -9.26739722e-02 -4.30227399e-01 1.60210669e-01 -2.14737467e-02 5.12896292e-02 -6.32842422e-01 1.97797269e-01 -3.71761709e-01 7.67287016e-01 -5.40123761e-01 -4.15817022e-01 -3.46638352e-01 -3.65852788e-02 -3.92571896e-01 6.09681383e-02 -1.65919200e-01 8.05836141e-01 -1.88946784e-01 6.69385314e-01 1.76872328e-01 2.40213633e-01 2.98980653e-01 -2.33011171e-01 -4.29579765e-01 8.43639612e-01 -1.29249585e+00 2.20239210e+00 -5.03535122e-02 8.73756111e-01 -4.34299000e-02 -9.91967559e-01 1.27685845e+00 3.06604356e-01 8.46761227e-01 -1.81773305e-01 9.01796892e-02 7.44151652e-01 2.68323421e-01 -5.63106537e-01 9.55840409e-01 -3.82707834e-01 -5.05865514e-01 7.53213048e-01 2.31015235e-01 -5.34484446e-01 5.58560908e-01 -4.78497706e-02 9.87459540e-01 3.90875131e-01 6.91384971e-01 -3.91305298e-01 5.62546909e-01 1.17949218e-01 6.30659401e-01 5.65497398e-01 6.90654218e-02 1.15124393e+00 4.45041299e-01 -7.01135159e-01 -1.10065579e+00 -1.04436266e+00 -1.04933791e-01 1.44191480e+00 -3.14809561e-01 -9.87230480e-01 -7.31223881e-01 -1.02777608e-01 -2.69123703e-01 2.39974633e-02 -6.53765559e-01 2.71052390e-01 -9.65745509e-01 -4.95107442e-01 1.06225884e+00 3.56329024e-01 4.88082021e-01 -1.63623381e+00 -8.04915428e-01 3.99595708e-01 -5.65561354e-01 -5.70950329e-01 -7.59068549e-01 5.79821765e-01 -8.06041241e-01 -9.94051099e-01 -6.46770358e-01 -1.14367282e+00 -3.03824782e-01 -9.15546268e-02 1.22450149e+00 -3.89505684e-01 -5.30961394e-01 7.28479251e-02 -3.81108999e-01 -5.05223215e-01 -4.58831996e-01 6.60253525e-01 6.52585030e-02 3.27021144e-02 1.46633424e-02 -1.17365706e+00 -4.01968688e-01 2.86563843e-01 -6.88089490e-01 -1.49680257e-01 2.95467973e-01 5.24416864e-01 9.61736441e-01 1.48073370e-02 4.89440829e-01 -7.69202113e-01 6.44484818e-01 2.01456398e-02 -3.97791155e-02 -1.28803730e-01 7.33155385e-02 -9.47690457e-02 5.76283813e-01 -4.87420410e-01 -5.00885606e-01 1.92232832e-01 -9.78499055e-02 -1.07701138e-01 -5.41903637e-03 6.13005877e-01 -1.75009653e-01 3.07970047e-01 9.69033897e-01 2.00538874e-01 -3.12284917e-01 -8.60921621e-01 1.42503724e-01 8.89280677e-01 1.20517802e+00 -8.05663109e-01 7.49840319e-01 2.47178569e-01 -1.17449947e-02 -1.17702186e+00 -1.25893354e+00 -8.49050283e-01 -9.35230374e-01 -4.57575113e-01 7.59695768e-01 -4.72437352e-01 -8.13151538e-01 1.54379472e-01 -6.37612820e-01 -2.16275617e-01 -8.39616001e-01 2.89902717e-01 -1.09759092e+00 3.44509393e-01 -7.30208218e-01 -5.49735248e-01 -7.24496782e-01 -5.30111492e-01 1.02363229e+00 2.12187156e-01 -1.12907588e+00 -3.86734426e-01 7.94578552e-01 4.26513523e-01 -2.95285881e-01 4.21454281e-01 8.86233389e-01 -5.86224854e-01 -8.13468918e-02 2.34534755e-01 5.39959431e-01 2.07307026e-01 3.50899696e-01 -2.31795628e-02 -1.07470655e+00 -1.62167430e-01 -4.64036725e-02 -5.08477569e-01 1.15120029e+00 4.23131913e-01 9.95081902e-01 -2.26619691e-01 3.90521407e-01 7.86631107e-01 9.67189729e-01 3.07230979e-01 3.99137795e-01 7.79168189e-01 5.35822153e-01 7.23688185e-01 6.22802436e-01 5.70907235e-01 -4.61853355e-01 7.88812041e-01 -5.84135875e-02 4.40515488e-01 -2.36183479e-01 -3.29179943e-01 6.02396548e-01 1.41868472e+00 -6.76240742e-01 2.71062106e-01 -8.47942829e-01 6.59085393e-01 -1.71884668e+00 -1.32669604e+00 -2.13345081e-01 2.06357908e+00 1.09224200e+00 -1.02841547e-02 7.49145985e-01 1.07633042e+00 7.34291136e-01 2.36629784e-01 -2.54716098e-01 -5.57260633e-01 -4.99980360e-01 8.42832208e-01 1.32363141e-02 -4.32168059e-02 -1.44518101e+00 7.05117345e-01 6.12421703e+00 1.14029682e+00 -1.05369234e+00 -3.13310564e-01 5.41855320e-02 -2.99868584e-01 -4.04390367e-03 -4.54159565e-02 -3.36810797e-01 8.58662054e-02 6.34396851e-01 -8.89208615e-02 4.34179813e-01 5.67792892e-01 1.39773384e-01 1.08875729e-01 -8.32465053e-01 1.24387002e+00 1.40941128e-01 -1.67337048e+00 3.87878679e-02 -1.24033019e-01 9.85772669e-01 -5.10276556e-02 3.58400755e-02 2.32426785e-02 -3.89817119e-01 -1.02146065e+00 1.26283062e+00 2.24514678e-01 7.67697275e-01 -1.08115411e+00 2.64262199e-01 -8.75884725e-04 -1.54568589e+00 -9.92402732e-02 -7.56506994e-02 -1.74371168e-01 -1.25422418e-01 1.38313174e-01 -7.80866325e-01 6.72350705e-01 8.13422382e-01 7.84378290e-01 -4.51543897e-01 1.32705426e+00 -3.18511188e-01 1.25848699e+00 -2.61266291e-01 2.36242980e-01 2.17526764e-01 -6.04528368e-01 9.95321035e-01 1.37575114e+00 5.06894767e-01 -2.85753608e-01 2.68388301e-01 5.87399244e-01 -1.06005639e-01 6.37798548e-01 -7.36878365e-02 -1.53946564e-01 2.56720215e-01 1.30991566e+00 -1.22157538e+00 -2.94335634e-02 2.94842303e-01 8.11760724e-01 -1.33152768e-01 -2.00016648e-02 -3.46879154e-01 -8.21660340e-01 3.89212161e-01 3.24781924e-01 5.55334985e-01 -3.09408367e-01 -4.82175648e-01 -7.47189641e-01 -1.05640158e-01 -1.00417066e+00 6.66862130e-01 -5.58726549e-01 -1.09662271e+00 7.36410141e-01 -4.61036474e-01 -1.72141385e+00 -5.73836446e-01 -1.41171604e-01 -8.07962358e-01 5.04506528e-01 -7.13367760e-01 -1.01473486e+00 -1.60814926e-01 4.72764373e-01 9.05560553e-01 -5.91135859e-01 1.15066969e+00 9.13122594e-02 -1.53147832e-01 2.79614985e-01 3.46902758e-01 2.97441900e-01 8.76860917e-01 -1.49894261e+00 1.97179526e-01 2.34051839e-01 1.16414046e+00 3.09192657e-01 8.66802573e-01 -4.47060347e-01 -6.24728739e-01 -7.98721433e-01 9.52086210e-01 -8.13637525e-02 9.47997868e-01 -2.73818135e-01 -7.74034619e-01 4.42647904e-01 3.00718695e-01 -8.21831644e-01 1.25154698e+00 3.72993141e-01 -3.64746988e-01 -2.05971450e-01 -4.19025034e-01 4.41050768e-01 1.06390035e+00 -6.70576215e-01 -1.02017963e+00 2.19932705e-01 1.48014024e-01 -1.26879543e-01 -7.74175406e-01 2.03526020e-01 9.21659112e-01 -1.04073262e+00 6.09267235e-01 -3.89893502e-01 5.59849381e-01 -4.87932593e-01 -2.12645456e-02 -9.69376028e-01 -2.14562222e-01 -1.68736994e+00 2.32284442e-01 1.29428470e+00 2.36955836e-01 2.63039380e-01 1.05310810e+00 -8.01766396e-01 -4.41989213e-01 -2.26340830e-01 -8.26813221e-01 -8.44221711e-01 -9.39943567e-02 -7.16378391e-01 2.56252199e-01 9.04789627e-01 1.25587031e-01 6.82410419e-01 -3.38655412e-01 -4.75607395e-01 5.12271285e-01 1.00630939e+00 7.99539804e-01 -1.56465149e+00 -6.50826037e-01 -7.25973725e-01 -5.31831563e-01 -6.85527742e-01 -9.23552066e-02 -1.19960856e+00 2.86864601e-02 -1.04905915e+00 8.74631628e-02 -3.18074018e-01 -4.85926241e-01 4.03968543e-01 3.26712757e-01 1.03872108e+00 4.08744633e-01 7.58617461e-01 -5.21885514e-01 1.79368183e-01 9.91336346e-01 -3.41944307e-01 -8.41642022e-01 5.35154223e-01 -4.35254186e-01 1.06204462e+00 1.14961457e+00 -4.92788285e-01 -1.78502411e-01 1.86200991e-01 5.38946867e-01 -1.85316145e-01 -3.17383140e-01 -1.41827583e+00 1.15172647e-01 2.36777842e-01 6.20031394e-02 -1.04949307e+00 2.22546995e-01 -7.05581680e-02 2.69701421e-01 2.83415616e-01 -5.94514132e-01 -1.44953370e-01 -7.06286915e-03 8.32239836e-02 -5.64827025e-01 -4.67305332e-01 7.32104480e-01 -1.52210221e-01 -5.20934939e-01 -1.51318565e-01 -7.04518914e-01 8.56300443e-02 4.02305394e-01 -4.27232504e-01 4.46446270e-01 -3.87393862e-01 -1.14858294e+00 -5.08263588e-01 1.25028834e-01 2.87945688e-01 2.64704615e-01 -1.34866786e+00 -8.68726730e-01 1.11014836e-01 2.51211524e-01 -2.44149327e-01 2.13999793e-01 8.49928856e-01 -9.76498425e-01 1.02404729e-01 -5.60963750e-01 -5.96953869e-01 -1.75116181e+00 2.33065821e-02 6.44340813e-02 -2.80640811e-01 -8.39717567e-01 8.73474896e-01 2.52003465e-02 -3.42146993e-01 4.29306120e-01 -2.70561934e-01 -6.05430424e-01 6.95495963e-01 4.97553438e-01 4.23411667e-01 5.75372726e-02 -8.54046226e-01 -4.05800007e-02 8.73424411e-01 3.29769105e-01 -4.00100231e-01 1.47375667e+00 3.31913568e-02 -4.13620234e-01 1.18038321e+00 9.41859245e-01 8.12109590e-01 -6.92575514e-01 3.75193432e-02 4.14240539e-01 -5.26133291e-02 -5.42299926e-01 -7.77474523e-01 -7.19233155e-01 6.67888343e-01 1.47655725e-01 3.13608468e-01 1.24887133e+00 1.18473090e-01 9.92666006e-01 6.12859368e-01 1.65174827e-01 -1.30145085e+00 2.18619063e-01 7.72648096e-01 1.14545178e+00 -3.86102885e-01 1.66098475e-02 -1.72814175e-01 -6.57996416e-01 1.53257847e+00 -2.53207207e-01 -3.18949997e-01 2.58366555e-01 2.30806634e-01 2.74914831e-01 -8.08236003e-02 -2.03767106e-01 -3.84299368e-01 6.32581234e-01 3.38814914e-01 6.97097301e-01 1.25950918e-01 -6.04737520e-01 9.61885989e-01 -1.44413507e+00 -4.15499508e-01 4.27492708e-01 7.83194542e-01 -7.31111109e-01 -1.31422317e+00 -7.03678608e-01 1.78934976e-01 -7.49844968e-01 -4.84934496e-03 -8.93678010e-01 6.26653612e-01 5.48143864e-01 7.14908183e-01 1.54875174e-01 -5.75428843e-01 1.68478310e-01 2.01495424e-01 7.38015592e-01 -5.55594385e-01 -1.24110889e+00 9.87837434e-01 1.35760650e-01 -1.62596390e-01 -6.70848966e-01 -8.01927090e-01 -1.34079480e+00 3.90892401e-02 9.49819311e-02 7.30275095e-01 4.05781895e-01 6.16749823e-01 -3.77955228e-01 6.86439037e-01 7.35542178e-01 -1.00817895e+00 -8.82086437e-03 -1.20699430e+00 -1.26797378e+00 5.15813351e-01 1.63726613e-03 -1.10621199e-01 -1.66860651e-02 3.45442176e-01]
[15.900684356689453, 5.316169261932373]
62dd321a-0df8-4c1f-80c0-597a5cfbca9b
adapool-exponential-adaptive-pooling-for
2111.00772
null
https://arxiv.org/abs/2111.00772v3
https://arxiv.org/pdf/2111.00772v3.pdf
AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling
Pooling layers are essential building blocks of convolutional neural networks (CNNs), to reduce computational overhead and increase the receptive fields of proceeding convolutional operations. Their goal is to produce downsampled volumes that closely resemble the input volume while, ideally, also being computationally and memory efficient. Meeting both these requirements remains a challenge. To this end, we propose an adaptive and exponentially weighted pooling method: adaPool. Our method learns a regional-specific fusion of two sets of pooling kernels that are based on the exponent of the Dice-Sorensen coefficient and the exponential maximum, respectively. AdaPool improves the preservation of detail on a range of tasks including image and video classification and object detection. A key property of adaPool is its bidirectional nature. In contrast to common pooling methods, the learned weights can also be used to upsample activation maps. We term this method adaUnPool. We evaluate adaUnPool on image and video super-resolution and frame interpolation. For benchmarking, we introduce Inter4K, a novel high-quality, high frame-rate video dataset. Our experiments demonstrate that adaPool systematically achieves better results across tasks and backbones, while introducing a minor additional computational and memory overhead.
['Ronald Poppe', 'Alexandros Stergiou']
2021-11-01
null
null
null
null
['video-super-resolution']
['computer-vision']
[ 1.06189236e-01 -1.81868151e-01 -8.41756389e-02 -4.25931573e-01 -5.91159225e-01 -5.01165688e-01 6.45976126e-01 -9.54857841e-02 -7.90205956e-01 6.86998427e-01 2.08546937e-01 1.04132645e-01 2.48667374e-01 -7.66918063e-01 -9.40027893e-01 -6.22309625e-01 -1.61033317e-01 -3.42911273e-01 7.17487872e-01 3.05415988e-02 1.47399798e-01 7.92159617e-01 -1.56011367e+00 7.61246145e-01 6.12258613e-01 1.35190558e+00 2.32263267e-01 6.68762565e-01 3.52115147e-02 8.71715724e-01 -2.65337706e-01 -3.76938879e-01 4.88206863e-01 -1.43557131e-01 -8.25364590e-01 -1.02522619e-01 7.16279328e-01 -5.51401079e-01 -4.29018974e-01 7.95292616e-01 3.44195366e-01 3.44338685e-01 3.95996034e-01 -9.05032694e-01 -6.63812399e-01 4.15569901e-01 -6.50545895e-01 6.61838710e-01 -1.41443044e-01 1.37683630e-01 8.66353393e-01 -1.20581496e+00 6.38236701e-01 1.14736497e+00 9.65370297e-01 5.20165086e-01 -1.57089448e+00 -6.14941359e-01 2.79847495e-02 -5.50007336e-02 -1.37839425e+00 -5.96473098e-01 3.16650063e-01 -3.64997566e-01 9.76918221e-01 1.68249056e-01 5.15319109e-01 6.84774816e-01 3.50456715e-01 5.69802046e-01 1.09840381e+00 -2.76707672e-02 2.58448958e-01 9.06179622e-02 -2.48473302e-01 6.48550212e-01 1.09735064e-01 -1.52752474e-01 -8.50054622e-01 -4.53043683e-03 1.30306792e+00 -7.38594169e-03 -5.33563077e-01 -3.21072191e-01 -1.32790184e+00 6.35695338e-01 8.01322341e-01 1.59720495e-01 -5.22954404e-01 3.13124537e-01 3.39437217e-01 6.32267445e-02 6.29254222e-01 4.02745485e-01 -4.22023237e-01 1.19989038e-01 -1.21968174e+00 3.73319119e-01 3.91767651e-01 7.40847528e-01 8.71415257e-01 1.57181732e-02 -4.01520908e-01 8.18697870e-01 -7.31499568e-02 1.11904955e-02 5.06141365e-01 -1.13615632e+00 3.42526197e-01 5.17895043e-01 9.69768092e-02 -8.99517655e-01 -3.26341063e-01 -3.58037174e-01 -9.08443034e-01 5.34143806e-01 5.67134380e-01 3.19325514e-02 -8.32081914e-01 1.67789626e+00 1.84770852e-01 4.28347826e-01 -8.11036825e-02 1.02239728e+00 8.23638916e-01 7.22333372e-01 2.81732470e-01 8.53652880e-03 1.40615058e+00 -9.10338640e-01 -3.71753544e-01 -1.37879953e-01 3.28477651e-01 -6.37381554e-01 1.10754347e+00 1.15151636e-01 -1.51108229e+00 -6.22714937e-01 -1.14244568e+00 -4.69517946e-01 -2.45525002e-01 1.21048510e-01 5.37228286e-01 3.90318483e-01 -1.24632943e+00 1.00807238e+00 -9.46191967e-01 8.60909522e-02 8.21486294e-01 3.46636236e-01 -5.61371684e-01 5.20116314e-02 -8.97518814e-01 7.93619215e-01 5.19847333e-01 8.65377262e-02 -4.86625433e-01 -1.34306920e+00 -8.10717344e-01 2.50622332e-01 -1.17009886e-01 -4.94105667e-01 1.06213117e+00 -1.04109251e+00 -1.25297379e+00 8.57279360e-01 -5.14297113e-02 -8.09980333e-01 5.40244520e-01 -1.99225113e-01 5.79505004e-02 2.72587776e-01 -2.81110168e-01 1.13154447e+00 9.68029380e-01 -8.86579633e-01 -6.35581255e-01 -1.45369440e-01 -5.57843782e-02 1.28094763e-01 -5.62141836e-01 6.47001714e-02 -5.20998478e-01 -8.02520454e-01 -1.16559453e-01 -6.08296871e-01 -2.36689195e-01 4.71674323e-01 1.57345638e-01 -5.81878014e-02 8.63615453e-01 -8.34980965e-01 9.60891306e-01 -2.24468756e+00 1.50807157e-01 2.71192566e-02 5.34586012e-01 3.67692024e-01 -6.80423379e-02 -1.82572216e-01 -8.33899379e-02 7.18842968e-02 -4.69499797e-01 -3.97999555e-01 -4.44457024e-01 8.07370320e-02 -2.65137970e-01 5.46736062e-01 7.27148771e-01 1.00774121e+00 -7.95321107e-01 -4.83587861e-01 2.47305334e-01 9.42501724e-01 -7.35739112e-01 1.11987427e-01 1.88451000e-02 3.38821739e-01 -1.29175363e-02 3.86800230e-01 7.67472923e-01 -3.60140115e-01 -5.83986863e-02 -4.30730045e-01 -2.82153666e-01 1.27849042e-01 -1.05769420e+00 1.72931993e+00 -4.73118007e-01 9.86156166e-01 5.10449745e-02 -6.54348433e-01 8.22824001e-01 1.03430264e-01 4.18113172e-01 -6.59359276e-01 1.24019876e-01 2.26849422e-01 -2.33005971e-01 -6.88322783e-02 7.30428636e-01 6.45883456e-02 4.43026096e-01 1.81594834e-01 2.56514251e-01 2.82820826e-03 2.03653768e-01 4.78932112e-02 9.90639329e-01 1.97606564e-01 3.25664848e-01 -6.17053390e-01 3.64624530e-01 -3.81805748e-01 5.27984262e-01 5.56809545e-01 -2.23164126e-01 9.87591863e-01 6.57330930e-01 -6.61950707e-01 -1.48921502e+00 -1.10137677e+00 -4.31383789e-01 1.04829907e+00 -1.21517166e-01 -2.19031468e-01 -8.95006776e-01 -4.60771024e-01 -4.09571193e-02 1.47331189e-02 -7.89977670e-01 1.77805096e-01 -8.45271468e-01 -6.79655492e-01 4.62328762e-01 8.76825869e-01 7.79910386e-01 -1.09554815e+00 -1.00228333e+00 2.15829805e-01 -5.55099957e-02 -1.41396570e+00 -6.18472278e-01 6.36627898e-02 -9.13955390e-01 -9.33742225e-01 -1.05675912e+00 -7.76110590e-01 5.66429317e-01 2.64865994e-01 1.26088655e+00 5.25433160e-02 -4.07910019e-01 -2.73759244e-03 -1.58883348e-01 -1.56408951e-01 -5.74158467e-02 2.04547703e-01 -1.01293512e-01 2.15842098e-01 1.23810828e-01 -7.30325520e-01 -9.69266117e-01 2.48043641e-01 -1.20854330e+00 1.33831441e-01 5.61145186e-01 7.50707746e-01 7.53471971e-01 -2.38448575e-01 2.95238078e-01 -5.24740696e-01 5.28145134e-01 -2.79966235e-01 -8.05554211e-01 9.78445541e-03 -1.13864601e-01 1.85081795e-01 7.01050580e-01 -4.84125942e-01 -8.93248081e-01 2.39057809e-01 -5.94418943e-02 -5.58800459e-01 1.66419268e-01 5.13564944e-02 2.23229587e-01 -4.44324344e-01 7.61402786e-01 8.42115507e-02 4.61655036e-02 -2.48502091e-01 2.47279659e-01 3.00276130e-01 7.30586886e-01 -3.88350815e-01 5.41662633e-01 7.11873531e-01 3.03830439e-03 -8.97129536e-01 -6.04445577e-01 -3.07328463e-01 -6.22645855e-01 -2.49477670e-01 9.83483672e-01 -9.32423174e-01 -5.88112354e-01 4.76452917e-01 -1.17073369e+00 -5.59877694e-01 -4.15866911e-01 4.27110404e-01 -4.39877212e-01 8.70222971e-02 -7.43992567e-01 -4.37190920e-01 -4.66364443e-01 -1.19264817e+00 9.14633453e-01 3.73219222e-01 -8.18637013e-02 -8.05218160e-01 -2.38582283e-01 -2.05070078e-01 7.29727864e-01 4.98691946e-01 4.71462697e-01 -2.34646931e-01 -6.53886616e-01 5.36790378e-02 -7.71892726e-01 6.62370026e-01 -7.44026229e-02 -1.18668447e-03 -1.10311365e+00 -3.09141040e-01 -2.85846859e-01 -3.96215677e-01 1.18939137e+00 6.19376421e-01 1.48164129e+00 -2.39542186e-01 1.06609635e-01 9.49530244e-01 1.41074073e+00 -2.30393752e-01 8.95264030e-01 3.62244904e-01 5.37384331e-01 6.66011751e-01 7.73628801e-02 3.78925145e-01 5.80732524e-02 7.24492490e-01 3.12545687e-01 -1.91743866e-01 -4.18338835e-01 4.92768586e-02 1.77937970e-01 3.93276662e-01 -4.01425570e-01 2.58236498e-01 -7.01348543e-01 5.90960383e-01 -1.59041572e+00 -1.00209510e+00 1.27460612e-02 2.15365911e+00 9.71947491e-01 7.80746490e-02 9.74407941e-02 2.48236973e-02 5.38677394e-01 2.77128965e-01 -4.07407463e-01 -3.55567992e-01 -2.38383979e-01 5.25459588e-01 7.74632752e-01 3.38664889e-01 -1.26749671e+00 7.96204805e-01 6.21127558e+00 7.04074979e-01 -1.19762409e+00 1.71201497e-01 9.37515557e-01 -3.31142098e-01 4.09029424e-03 -3.70771915e-01 -7.87956655e-01 4.07673061e-01 7.63257742e-01 1.85549155e-01 5.55877686e-01 7.93380141e-01 7.67502040e-02 -2.46423073e-02 -1.06136394e+00 9.33578312e-01 -8.48708674e-02 -1.89793324e+00 -5.53084211e-03 -5.71265928e-02 9.09635782e-01 2.62524486e-01 1.79249868e-01 3.97592187e-02 9.61534679e-03 -1.19221365e+00 7.94746101e-01 4.23441589e-01 8.96248817e-01 -8.36021066e-01 6.73759818e-01 -2.20116377e-01 -1.39400399e+00 -9.46682021e-02 -4.80301261e-01 1.67810813e-01 6.18396364e-02 5.46215117e-01 -3.21538746e-01 5.59074357e-02 1.26956594e+00 6.02823436e-01 -4.43070143e-01 1.13424194e+00 5.67815639e-02 2.33845189e-01 -2.36151218e-01 1.80745736e-01 3.50226969e-01 6.37800470e-02 2.47534066e-01 1.50518143e+00 2.21441314e-01 1.95995733e-01 -1.61228254e-01 1.02151990e+00 -5.20861804e-01 2.01847591e-02 -3.39480430e-01 2.49620318e-01 4.26857293e-01 1.40476692e+00 -8.26054871e-01 -2.91026592e-01 -4.52013522e-01 9.20801759e-01 4.92841214e-01 3.77771378e-01 -8.14137578e-01 -5.60077488e-01 9.08626437e-01 3.08780938e-01 5.27196467e-01 -3.39955240e-01 -4.38266873e-01 -1.02222824e+00 3.23954165e-01 -6.04465842e-01 1.80940218e-02 -5.83443284e-01 -9.20222759e-01 8.29386950e-01 -1.03824221e-01 -1.05133355e+00 1.30804166e-01 -8.16960216e-01 -4.27054256e-01 1.04191554e+00 -1.87755656e+00 -9.83143270e-01 -6.41737044e-01 6.44514263e-01 6.36887312e-01 1.60975903e-01 4.74490523e-01 3.38456333e-01 -4.18197453e-01 5.21473289e-01 -1.99613020e-01 2.22737342e-01 6.75055683e-01 -1.09472513e+00 6.03940547e-01 8.79236162e-01 -1.31842241e-01 5.59886038e-01 5.65103948e-01 -2.59294987e-01 -9.68314171e-01 -1.34640300e+00 6.23884678e-01 -2.26754442e-01 5.23161054e-01 -4.44721907e-01 -1.26228595e+00 5.84402680e-01 4.19571362e-02 6.69430196e-01 3.06640744e-01 -2.58522719e-01 -5.30410111e-01 -2.31041014e-01 -1.16735435e+00 6.05602086e-01 7.46652365e-01 -3.26144040e-01 -2.79332191e-01 8.46568123e-02 6.65677547e-01 -6.04237556e-01 -1.09224510e+00 2.42196545e-01 8.29787314e-01 -1.29080057e+00 1.24739707e+00 -3.98392797e-01 8.83045018e-01 -2.26954594e-01 -8.87956545e-02 -1.14836061e+00 -4.21176761e-01 -4.35340822e-01 -2.06325829e-01 8.25267315e-01 3.69417757e-01 -5.69549143e-01 8.50266218e-01 7.50203729e-01 -8.53153765e-02 -9.37774956e-01 -9.39734936e-01 -7.00719059e-01 4.19105627e-02 -2.01424181e-01 5.70719540e-01 7.75658548e-01 -3.21402103e-01 -3.03050429e-01 -3.91538262e-01 8.73257145e-02 6.79948092e-01 -4.08563226e-01 5.62573850e-01 -9.96152878e-01 -1.08105294e-01 -6.61424696e-01 -5.64801097e-01 -1.02129054e+00 -2.09395010e-02 -7.15233147e-01 1.67138241e-02 -1.24825239e+00 1.27583966e-01 -2.72647023e-01 -3.49181265e-01 5.28555572e-01 -5.13188504e-02 8.73195052e-01 1.32049963e-01 2.90690809e-01 -2.96096206e-01 3.80947381e-01 1.15302050e+00 1.73067063e-01 -4.11361545e-01 -3.29970509e-01 -4.03210789e-01 6.82705939e-01 9.65525448e-01 -1.57494068e-01 -1.64583474e-01 -6.66416585e-01 1.73290465e-02 -3.43172073e-01 6.51063144e-01 -1.10943043e+00 1.01885930e-01 -4.05921750e-02 7.48012543e-01 -3.33742619e-01 3.58985364e-01 -4.76998866e-01 -1.56085789e-01 3.65865111e-01 -4.30672854e-01 2.10598916e-01 4.35710609e-01 3.42270225e-01 -1.93089113e-01 1.25760466e-01 1.25967491e+00 -8.51205736e-02 -8.08967471e-01 5.33860087e-01 -1.32492796e-01 -3.23158726e-02 9.60205078e-01 -2.80493319e-01 -2.01334134e-01 -4.12115566e-02 -2.86671430e-01 -6.42264709e-02 4.31194067e-01 4.01202232e-01 7.75593758e-01 -1.33247471e+00 -8.57719243e-01 3.25536937e-01 -6.80879653e-02 8.69697854e-02 3.15696776e-01 9.23759878e-01 -8.96051943e-01 2.97086000e-01 -5.47208667e-01 -5.49836993e-01 -1.13323915e+00 2.70833045e-01 5.13770282e-01 -2.74020433e-01 -8.90031099e-01 1.08429027e+00 2.80191958e-01 -8.14832151e-02 2.80576229e-01 -6.94776714e-01 -1.78995147e-01 -9.21203941e-02 1.11845386e+00 2.28601351e-01 2.28968784e-01 -4.99123991e-01 -2.88418621e-01 3.21093529e-01 -1.82506233e-01 4.72545810e-02 1.46668482e+00 1.47386536e-01 -1.57780096e-01 6.28177375e-02 1.28007793e+00 -2.72296190e-01 -1.93941259e+00 -3.05669755e-01 -1.22990765e-01 -6.89702511e-01 3.37959081e-01 -2.39176512e-01 -1.33542800e+00 7.42806852e-01 5.52988112e-01 1.44236207e-01 1.21025121e+00 -1.25285089e-01 8.57513011e-01 5.28528681e-03 7.31478631e-02 -9.23921406e-01 2.48422623e-02 2.27786720e-01 1.02534473e+00 -1.11022246e+00 3.38527635e-02 -2.66574085e-01 -4.82393712e-01 1.24291730e+00 6.24191105e-01 -5.20310283e-01 4.21639144e-01 5.22726715e-01 -2.75425732e-01 1.91067547e-01 -6.03643835e-01 6.52726740e-02 4.20684814e-01 4.99578655e-01 6.27712786e-01 -2.45350063e-01 -2.15074480e-01 2.72827715e-01 -4.63993065e-02 1.55649677e-01 3.52627993e-01 8.43614936e-01 -4.55509573e-01 -6.73245490e-01 -2.76385933e-01 4.38253909e-01 -6.53933287e-01 -2.54838645e-01 9.09483209e-02 6.53283238e-01 5.69446385e-02 4.11790609e-01 4.41454262e-01 -8.40997472e-02 2.90325493e-01 -2.47897863e-01 4.60226715e-01 -1.81280851e-01 -6.17308557e-01 -2.17928752e-01 -3.08642119e-01 -8.43608797e-01 -4.59315509e-01 -4.20194387e-01 -1.04439974e+00 -4.51601833e-01 1.30053991e-02 -2.21402645e-01 7.01079071e-01 6.90994143e-01 2.56933451e-01 6.34889424e-01 3.34540576e-01 -1.29163229e+00 -3.59588861e-01 -6.91633046e-01 -3.48558992e-01 3.10508639e-01 4.68066573e-01 -5.62452495e-01 -1.63255036e-01 2.74679661e-01]
[10.882368087768555, -1.2733569145202637]
e76dffd9-8a0d-4a27-a3c7-ae2eee5b6eb2
abstract-visual-reasoning-an-algebraic
2303.1173
null
https://arxiv.org/abs/2303.11730v1
https://arxiv.org/pdf/2303.11730v1.pdf
Abstract Visual Reasoning: An Algebraic Approach for Solving Raven's Progressive Matrices
We introduce algebraic machine reasoning, a new reasoning framework that is well-suited for abstract reasoning. Effectively, algebraic machine reasoning reduces the difficult process of novel problem-solving to routine algebraic computation. The fundamental algebraic objects of interest are the ideals of some suitably initialized polynomial ring. We shall explain how solving Raven's Progressive Matrices (RPMs) can be realized as computational problems in algebra, which combine various well-known algebraic subroutines that include: Computing the Gr\"obner basis of an ideal, checking for ideal containment, etc. Crucially, the additional algebraic structure satisfied by ideals allows for more operations on ideals beyond set-theoretic operations. Our algebraic machine reasoning framework is not only able to select the correct answer from a given answer set, but also able to generate the correct answer with only the question matrix given. Experiments on the I-RAVEN dataset yield an overall $93.2\%$ accuracy, which significantly outperforms the current state-of-the-art accuracy of $77.0\%$ and exceeds human performance at $84.4\%$ accuracy.
['Kai Fong Ernest Chong', 'Zhangsheng Lai', 'Saket Chandra', 'Yufei Wu', 'Tushar Vaidya', 'Jingyi Xu']
2023-03-21
null
http://openaccess.thecvf.com//content/CVPR2023/html/Xu_Abstract_Visual_Reasoning_An_Algebraic_Approach_for_Solving_Ravens_Progressive_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Xu_Abstract_Visual_Reasoning_An_Algebraic_Approach_for_Solving_Ravens_Progressive_CVPR_2023_paper.pdf
cvpr-2023-1
['visual-reasoning', 'visual-reasoning']
['computer-vision', 'reasoning']
[ 5.44424318e-02 4.23009455e-01 6.58904091e-02 -1.62314877e-01 -5.00393271e-01 -7.55855381e-01 2.84981430e-01 4.72896397e-01 -1.27008945e-01 4.45304096e-01 -3.61784369e-01 -9.79569554e-01 -5.94498217e-01 -1.31696403e+00 -5.65050840e-01 -2.48576537e-01 -4.90994036e-01 8.61849010e-01 2.95583683e-04 -7.59104133e-01 3.28282684e-01 5.83408892e-01 -1.81223977e+00 3.41470420e-01 9.32205737e-01 1.19344485e+00 -5.26369810e-01 9.56009865e-01 -3.21084648e-01 1.31914413e+00 -3.72000962e-01 -9.82033074e-01 6.37057662e-01 4.78076860e-02 -1.35770929e+00 -7.90830702e-02 7.95428813e-01 -1.34414271e-01 -2.14012742e-01 1.33783650e+00 -1.46812573e-01 4.84092757e-02 5.90545774e-01 -1.45503509e+00 -5.05768836e-01 7.72264123e-01 -2.70222604e-01 1.51120469e-01 9.33896124e-01 1.49433583e-03 1.72695351e+00 -1.00459647e+00 5.33224940e-01 1.27572751e+00 5.72121680e-01 1.95674762e-01 -1.39891136e+00 -4.97177929e-01 -3.20294768e-01 6.65342987e-01 -1.85372901e+00 -9.04824585e-02 4.57627356e-01 -3.77144605e-01 7.71248162e-01 9.71997678e-01 6.00788176e-01 -4.22233522e-01 -2.96163820e-02 6.25033975e-01 9.46877956e-01 -6.37291491e-01 6.35601999e-03 5.89039065e-02 6.87100291e-01 1.31733739e+00 1.51459381e-01 -3.81830484e-01 -5.05320616e-02 -3.08159649e-01 7.83403218e-01 1.97653063e-02 -1.56761527e-01 -3.58413190e-01 -1.33066201e+00 9.87850487e-01 5.99092364e-01 1.25122502e-01 -7.46449381e-02 5.61785372e-03 2.87837982e-01 9.73009646e-01 -3.53548378e-01 8.47651362e-01 -5.04489601e-01 4.20472383e-01 -5.66833019e-01 7.65195847e-01 1.37960124e+00 8.35724056e-01 8.33060622e-01 -1.55556917e-01 5.61599657e-02 4.56576765e-01 5.06672598e-02 5.04445970e-01 -2.86908329e-01 -1.25935638e+00 3.95053566e-01 1.01653767e+00 1.45218140e-02 -1.26782978e+00 -3.81647795e-01 -4.00224000e-01 -9.49388444e-01 1.75935060e-01 8.46645355e-01 3.19137543e-01 -3.42455387e-01 1.61237872e+00 3.14814478e-01 -2.60515511e-01 2.38035917e-01 7.05889523e-01 8.51910472e-01 5.86066365e-01 -1.81989446e-01 -3.21285516e-01 1.63909340e+00 -8.01633716e-01 -3.56211722e-01 2.96841562e-01 9.42002296e-01 -6.81015015e-01 8.10190320e-01 7.12476730e-01 -1.43787634e+00 -4.83965546e-01 -1.16663909e+00 -1.64357424e-01 -3.02452058e-01 -9.82143953e-02 1.30247307e+00 4.96794641e-01 -1.06052887e+00 4.24221337e-01 6.66441992e-02 2.40200251e-01 4.79558948e-03 6.57727361e-01 -4.75264579e-01 -3.07952821e-01 -1.32721937e+00 1.00921953e+00 3.98272008e-01 -1.33544892e-01 -4.39223200e-01 -1.02300704e+00 -9.65961874e-01 1.74795389e-01 8.10354769e-01 -7.48766363e-01 1.35646355e+00 -4.26676750e-01 -9.10516798e-01 1.19221628e+00 -4.36209925e-02 -8.48554790e-01 2.14529991e-01 3.38418745e-02 -5.78891098e-01 3.26264441e-01 -2.45246943e-02 3.26680809e-01 6.07373238e-01 -7.33423352e-01 -3.34935099e-01 -3.73442113e-01 7.20610738e-01 5.39261438e-02 8.49244222e-02 1.24884807e-01 -4.61013243e-02 -2.81560391e-01 6.27157211e-01 -9.15289998e-01 -5.68555713e-01 1.01759948e-01 -2.31125802e-01 -7.44111657e-01 4.05379981e-01 -3.57380778e-01 1.42303300e+00 -1.88492930e+00 2.73737341e-01 6.97301507e-01 7.57758260e-01 2.81120121e-01 2.31414527e-01 3.56201857e-01 -5.00672042e-01 1.16342187e-01 -5.63415103e-02 5.95975041e-01 1.10513642e-01 2.64230035e-02 -7.20753729e-01 3.39889675e-01 1.94276914e-01 7.22866178e-01 -7.66596973e-01 -6.30216062e-01 1.68387786e-01 -3.31449300e-01 -1.06644034e+00 2.23111454e-02 -6.36720955e-01 -2.41040915e-01 -1.02751292e-01 5.93552709e-01 8.02274227e-01 -4.33126152e-01 3.92189264e-01 3.46879545e-03 2.81850725e-01 3.40572834e-01 -1.89469528e+00 1.24970329e+00 -2.46751651e-01 2.10325271e-01 1.84348822e-01 -1.09797955e+00 9.67834890e-01 8.24418217e-02 1.30666271e-01 -2.99546927e-01 2.18992028e-02 3.06901127e-01 4.55968857e-01 -2.77188689e-01 5.82585692e-01 -1.73359558e-01 -3.39298934e-01 4.46129262e-01 -1.10253207e-01 -7.04090893e-01 6.06289744e-01 7.20674813e-01 1.27677643e+00 -6.32557273e-01 4.78208065e-01 -5.93632698e-01 1.33395588e+00 2.75209755e-01 4.02591437e-01 7.67710567e-01 2.32116342e-01 -1.08581288e-02 6.08797729e-01 -8.88374627e-01 -8.87754560e-01 -1.40741861e+00 -8.73735100e-02 1.03934968e+00 5.84282465e-02 -1.02083361e+00 -5.15729606e-01 -1.59815654e-01 4.51171324e-02 6.02797568e-01 -2.07286149e-01 9.68037918e-02 -6.07816398e-01 -1.27755731e-01 4.77462709e-01 3.57078016e-01 5.98622620e-01 -6.48990810e-01 -1.57416001e-01 -6.79469183e-02 -2.37940684e-01 -9.81165051e-01 3.93038578e-02 -8.36193189e-03 -9.54781651e-01 -1.51015830e+00 5.85089847e-02 -7.89244771e-01 8.35694134e-01 2.15218827e-01 1.45597589e+00 7.69899428e-01 -5.22194028e-01 3.07832807e-01 1.49656266e-01 -4.45784599e-01 -4.74681109e-01 -2.15161800e-01 1.74377516e-01 -2.49631330e-01 4.35989678e-01 -3.35871965e-01 -1.99228138e-01 3.55085790e-01 -7.70092487e-01 4.71697748e-02 2.54713923e-01 7.96461642e-01 4.69395101e-01 6.69352233e-01 1.10173106e-01 -8.87670100e-01 4.54235792e-01 -1.80816665e-01 -6.80246890e-01 4.06419694e-01 -3.04869324e-01 3.66264731e-01 1.02140260e+00 -2.32381359e-01 -5.18781781e-01 -1.64938480e-01 4.78059649e-02 -2.04277113e-01 3.00244063e-01 5.99493325e-01 -2.37487350e-02 -9.26306844e-02 1.02660310e+00 -2.93289125e-02 3.58597711e-02 8.74007866e-02 5.49410343e-01 3.99964571e-01 8.40522110e-01 -1.10447705e+00 1.22423172e+00 3.35116833e-01 6.68990791e-01 -6.87610090e-01 -9.56759453e-01 -5.28610528e-01 -5.05318582e-01 -3.30280662e-02 2.21745700e-01 -8.42168689e-01 -1.34575629e+00 5.69368936e-02 -1.17304838e+00 7.78659582e-02 -3.91307205e-01 2.26498291e-01 -7.49956071e-01 3.46902162e-01 -7.75512397e-01 -9.25343990e-01 -5.34915030e-01 -1.04835451e+00 6.13575101e-01 -1.38130933e-01 -6.17434204e-01 -6.36770666e-01 -3.23138654e-01 7.65493929e-01 7.68236965e-02 -2.76339743e-02 1.53881264e+00 -7.11496234e-01 -9.84973609e-01 -5.10350883e-01 -3.62092972e-01 3.21031272e-01 -7.37223446e-01 -3.28489505e-02 -5.54745674e-01 3.26039866e-02 5.19241281e-02 -2.41261452e-01 4.41945642e-01 -3.72678965e-01 1.19284475e+00 -4.54247028e-01 7.26633742e-02 2.14321032e-01 1.27369916e+00 -2.54756004e-01 8.56233716e-01 1.12143688e-01 2.16011837e-01 4.61807966e-01 6.94357216e-01 2.19838902e-01 4.35311377e-01 4.04064655e-01 3.88123274e-01 3.81694227e-01 3.42542171e-01 -1.17868781e-01 -4.82317470e-02 8.26168835e-01 -3.71078491e-01 8.51588845e-01 -9.97416079e-01 2.83521026e-01 -1.74181676e+00 -1.20435035e+00 -5.79868674e-01 2.26261640e+00 1.08846247e+00 2.52242148e-01 5.32709397e-02 8.67560744e-01 4.54175413e-01 -2.78451830e-01 9.77467187e-03 -9.48789895e-01 4.55259420e-02 8.82398069e-01 2.53951222e-01 6.49377167e-01 -9.31222379e-01 7.71184862e-01 6.64382029e+00 8.50673020e-01 -4.68030870e-01 -3.02542508e-01 1.82856828e-01 2.98398256e-01 -2.88415700e-01 6.38684109e-02 -5.90254664e-01 -2.88218200e-01 8.49184453e-01 -4.44272369e-01 8.88531685e-01 1.24238563e+00 -6.88313067e-01 -9.97319743e-02 -1.55879390e+00 1.11614907e+00 -6.41735792e-02 -1.43612778e+00 9.35103185e-03 -1.80297509e-01 4.37911838e-01 -7.30703652e-01 7.25644678e-02 6.98942125e-01 4.90830362e-01 -1.37914467e+00 5.02043605e-01 4.35905308e-01 6.31683409e-01 -8.57769847e-01 5.14144480e-01 5.23653984e-01 -1.23016083e+00 -3.16356272e-01 -4.41042811e-01 -7.73746192e-01 -4.69788074e-01 5.49263000e-01 -9.43869472e-01 5.41310668e-01 4.50580657e-01 1.07690431e-01 -5.63326299e-01 8.56988490e-01 -3.44402879e-01 6.87325895e-02 -4.65135992e-01 -9.74557176e-02 -1.24708407e-01 -3.46018702e-01 3.65359038e-01 8.73177826e-01 -3.87630351e-02 6.48218274e-01 2.53658354e-01 8.19332421e-01 -4.99967597e-02 3.48570108e-01 -7.37121165e-01 2.23183006e-01 2.92068422e-01 1.23643327e+00 -3.86027724e-01 -6.13268316e-01 -1.83516949e-01 4.40576613e-01 5.50738037e-01 -1.64300233e-01 -6.51398182e-01 -6.38060510e-01 7.51053691e-01 1.48022205e-01 1.53930053e-01 -3.60568345e-01 -4.52865541e-01 -1.20516980e+00 6.35721534e-03 -1.46565413e+00 8.04185688e-01 -8.74552369e-01 -9.81543422e-01 2.09062606e-01 2.63813902e-02 -1.00265813e+00 -4.12701100e-01 -1.02928030e+00 -3.48747104e-01 8.85367393e-01 -8.63346159e-01 -7.42765367e-01 -6.85454682e-02 8.37962568e-01 6.53012246e-02 9.63803381e-03 1.22204804e+00 1.32456154e-01 -1.36476740e-01 5.01793504e-01 -5.83841324e-01 3.26300591e-01 1.36943027e-01 -1.44513166e+00 -4.59114425e-02 7.65778065e-01 2.13504210e-01 1.10126448e+00 8.63103449e-01 -4.03879695e-02 -2.14443159e+00 -6.23673260e-01 1.08897007e+00 -6.25252187e-01 1.08261704e+00 -2.30124760e-02 -8.69002521e-01 7.55297661e-01 -2.62963861e-01 7.77528137e-02 6.87135816e-01 4.45870012e-01 -1.06537354e+00 -3.35458189e-01 -1.29622293e+00 1.08435190e+00 1.02953255e+00 -7.35639155e-01 -9.19190824e-01 4.99604702e-01 6.93187535e-01 -7.50155747e-01 -1.19782293e+00 5.65175712e-01 4.18267518e-01 -8.93148422e-01 1.43351805e+00 -9.53858674e-01 4.63891715e-01 -6.69211030e-01 -4.37526494e-01 -6.26530528e-01 -3.91091019e-01 -7.57284284e-01 -3.51190299e-01 5.13847947e-01 4.39126760e-01 -5.64148903e-01 5.12511253e-01 8.98349881e-01 3.03332031e-01 -7.63148010e-01 -5.05169094e-01 -4.93224651e-01 8.97884592e-02 -9.32422101e-01 7.64378011e-01 8.78540695e-01 6.17628932e-01 6.26062930e-01 8.80547911e-02 2.76530296e-01 6.10795856e-01 6.00992620e-01 1.11864316e+00 -1.60020733e+00 -5.07643878e-01 -5.32404244e-01 -8.50629926e-01 -8.97099674e-01 1.61759451e-01 -1.39472198e+00 -4.91626322e-01 -1.10065734e+00 9.64501034e-03 -7.23094106e-01 7.25692958e-02 2.89438784e-01 1.69239696e-02 2.42780045e-01 3.94541055e-01 4.21995856e-02 -8.69074166e-01 -2.04268232e-01 1.30563176e+00 -1.83738336e-01 1.64979085e-01 3.53755429e-02 -9.80205655e-01 1.02865434e+00 7.28636205e-01 1.44961432e-01 -3.07404518e-01 3.88736464e-02 8.52933049e-01 3.95010054e-01 4.01508778e-01 -1.13367355e+00 5.01244247e-01 -2.34616488e-01 6.22256435e-02 -5.33541441e-01 2.89364398e-01 -7.75527775e-01 2.38723025e-01 7.69422591e-01 -2.92208374e-01 2.90199906e-01 8.25219154e-02 1.95194960e-01 -7.90359303e-02 -5.25652230e-01 6.63056195e-01 -2.81125009e-01 -8.03530991e-01 5.10909595e-02 -3.76795195e-02 1.87636346e-01 9.99724686e-01 1.88545331e-01 -5.07255614e-01 -2.91394919e-01 -8.00159514e-01 2.35538244e-01 2.26386666e-01 -3.89841013e-02 5.70154309e-01 -1.20663607e+00 -4.73417312e-01 1.88701421e-01 1.09952860e-01 1.51900664e-01 6.64262250e-02 8.59412611e-01 -9.55122054e-01 6.13354146e-01 6.54977411e-02 -4.41961914e-01 -1.63543618e+00 9.69289422e-01 2.80850261e-01 -5.62471330e-01 -3.80901784e-01 6.60274386e-01 -1.60032436e-01 -7.53657877e-01 2.62421340e-01 -4.16010886e-01 8.46145153e-02 -3.50452840e-01 9.16927934e-01 5.57838321e-01 1.20887518e-01 -4.13695008e-01 -2.89177835e-01 3.28283966e-01 -2.99376752e-02 1.68762133e-01 8.88806462e-01 4.28958088e-01 -9.87923920e-01 8.72273296e-02 8.78783762e-01 5.33653796e-02 2.01021507e-01 -6.19059503e-01 -5.69205955e-02 -3.16085637e-01 -5.19966662e-01 -5.23326814e-01 -5.43261230e-01 7.27778316e-01 -1.07777022e-01 7.45296478e-01 1.02317107e+00 1.17450200e-01 3.83026123e-01 1.16095459e+00 7.50485539e-01 -7.68293262e-01 -1.14946738e-01 8.07512581e-01 8.96584511e-01 -9.77527916e-01 3.29561174e-01 -1.12613714e+00 -3.01570207e-01 1.32071197e+00 5.36220610e-01 -3.07275295e-01 6.83779359e-01 1.61224812e-01 -2.66883045e-01 -3.63602459e-01 -9.90251243e-01 -2.17622682e-01 3.83227021e-01 1.94443583e-01 2.50916988e-01 3.43804985e-01 -2.83145607e-01 4.95626599e-01 -9.58373606e-01 -9.35565755e-02 4.90144819e-01 6.48058534e-01 -6.10823989e-01 -9.19340968e-01 -8.67268384e-01 6.90648496e-01 -4.38303322e-01 -2.86843985e-01 -1.37974486e-01 7.27579534e-01 1.14821911e-01 8.53192031e-01 6.88523054e-02 -4.97786582e-01 1.47324577e-01 2.67604977e-01 8.69213521e-01 -6.73924863e-01 -4.73014027e-01 -8.17741394e-01 1.44542292e-01 -2.47101381e-01 -4.60913368e-02 -4.82446402e-01 -1.52881896e+00 -1.17052293e+00 -1.82589024e-01 3.70228767e-01 3.05482060e-01 8.89883041e-01 5.00888266e-02 1.04539938e-01 4.73855406e-01 -4.44483645e-02 -9.74734783e-01 -5.18004239e-01 -5.69234848e-01 3.44177693e-01 7.78669789e-02 -3.35063368e-01 -4.05558735e-01 -9.42859799e-02]
[8.977677345275879, 7.134103298187256]
4dd89e32-9671-4947-80f3-d2eede7d63ed
segmentation-of-skeletal-muscle-in-thigh
1904.04747
null
http://arxiv.org/abs/1904.04747v1
http://arxiv.org/pdf/1904.04747v1.pdf
Segmentation of Skeletal Muscle in Thigh Dixon MRI Based on Texture Analysis
Segmentation of skeletal muscles in Magnetic Resonance Images (MRI) is essential for the study of muscle physiology and diagnosis of muscular pathologies. However, manual segmentation of large MRI volumes is a time-consuming task. The state-of-the-art on algorithms for muscle segmentation in MRI is still not very extensive and is somewhat database-dependent. In this paper, an automated segmentation method based on AdaBoost classification of local texture features is presented. The texture descriptor consists of the Histogram of Oriented Gradients (HOG), Wavelet-based features, and a set of statistical measures computed from both the original and the Laplacian of Gaussian filtering of the grayscale MRI. The classifier performance suggests that texture analysis may be a helpful tool for designing a generalized and automated MRI muscle segmentation framework. Furthermore, an atlas-based approach to individual muscle segmentation is also described in this paper. The atlas is obtained by overlaying the muscle segmentation ground truth, provided by a radiologist, after image alignment using an appropriate affine transformation. Then, it is used to define the muscle labels upon the AdaBoost binary segmentation. The developed atlas method provides reasonable results when an accurate muscle tissue segmentation was obtained.
['Antonio M. G. Pinheiro', 'Rafael Rodrigues']
2019-04-09
null
null
null
null
['texture-classification']
['computer-vision']
[ 2.79809535e-01 -2.07128674e-01 -1.47141322e-01 -4.00263995e-01 -8.43402386e-01 -3.92003536e-01 8.54959339e-02 2.73450226e-01 -6.50449991e-01 3.54406774e-01 -2.41625637e-01 1.04274444e-01 -1.59132197e-01 -6.43399894e-01 -1.94289416e-01 -1.04992867e+00 -2.00462475e-01 7.32838988e-01 6.06178761e-01 5.67140132e-02 4.54744756e-01 7.43268371e-01 -1.20524931e+00 2.62644365e-02 5.49082220e-01 9.91665065e-01 4.63106096e-01 6.54310405e-01 -3.36879343e-01 4.84715790e-01 -3.92288148e-01 2.43049592e-01 5.06208122e-01 -7.18015492e-01 -1.02224481e+00 7.50234902e-01 1.66965455e-01 2.86552384e-02 1.33268550e-01 1.20768559e+00 3.75267297e-01 1.29785195e-01 8.34298849e-01 -5.14838815e-01 -1.75457060e-01 3.72469485e-01 -6.29134178e-01 5.02609849e-01 -3.48523036e-02 3.25449228e-01 3.68539691e-01 -6.42593265e-01 8.56023908e-01 7.73485482e-01 6.48341179e-01 3.25942278e-01 -1.37547398e+00 -7.36392885e-02 -5.50164998e-01 1.68875486e-01 -1.36613500e+00 6.68667629e-02 8.02435100e-01 -9.83633578e-01 4.80663031e-01 4.79202151e-01 1.05606639e+00 8.86093080e-02 6.26740158e-01 4.44678485e-01 1.82606649e+00 -6.02378607e-01 4.54182118e-01 1.62973683e-02 3.14596921e-01 7.24146843e-01 -2.25013681e-02 -2.46348351e-01 -8.28806162e-02 -4.83224839e-02 9.43015039e-01 -2.57214189e-01 8.58514830e-02 -5.92598617e-01 -1.03817940e+00 6.07758045e-01 2.60679632e-01 1.00248003e+00 -7.44547844e-01 -1.24879129e-01 6.10913813e-01 1.96273208e-01 6.37488008e-01 2.92790495e-02 3.20126377e-02 1.85798466e-01 -1.46919358e+00 4.87845391e-02 4.26251560e-01 2.20109001e-01 5.25333107e-01 -3.49718891e-02 -3.73377316e-02 7.39453435e-01 3.11812103e-01 3.93485188e-01 7.54711926e-01 -8.02784264e-01 -2.50074714e-01 7.03938901e-01 -5.72171748e-01 -1.00774980e+00 -4.93116349e-01 -2.48630028e-02 -6.51419342e-01 6.47196174e-01 7.64722586e-01 4.99734394e-02 -9.14139032e-01 8.89069319e-01 6.41600072e-01 -5.91068983e-01 -3.00671697e-01 1.28958178e+00 4.28675503e-01 6.38376847e-02 2.69089460e-01 -2.87322700e-01 1.45348060e+00 -7.16208637e-01 -6.96584344e-01 8.98853913e-02 5.94570100e-01 -8.77461016e-01 8.17828417e-01 2.72537351e-01 -1.10330200e+00 -5.16615331e-01 -7.84923494e-01 2.82864004e-01 -2.28704423e-01 2.97977448e-01 3.13907117e-01 5.95042706e-01 -9.68483984e-01 7.50588655e-01 -1.11601567e+00 -6.72176540e-01 2.20059544e-01 5.30979633e-01 -6.69559300e-01 3.25878650e-01 -5.92310011e-01 1.19705021e+00 2.83534080e-01 2.39249974e-01 -2.72500753e-01 -1.97612777e-01 -6.04491770e-01 -5.50717473e-01 4.88345586e-02 -2.66079783e-01 7.81086624e-01 -1.10894823e+00 -1.47241664e+00 1.57186496e+00 5.48509248e-02 -3.59024763e-01 6.15317225e-01 4.02165055e-01 -1.71252996e-01 7.55606174e-01 1.97050393e-01 3.78565282e-01 8.45432758e-01 -1.09015739e+00 -3.61709863e-01 -8.37787151e-01 -5.37671983e-01 2.34217912e-01 1.84604093e-01 3.93711358e-01 -1.14832431e-01 -6.32819474e-01 5.76443195e-01 -9.40752685e-01 -3.24144602e-01 2.47895159e-02 -2.48622835e-01 -1.48680061e-01 6.57407105e-01 -1.36179852e+00 9.94352579e-01 -2.23984742e+00 2.22811133e-01 6.96838379e-01 1.34635240e-01 -7.63393715e-02 4.53529894e-01 -6.40159324e-02 -1.08056225e-01 -1.95458487e-01 -5.76191902e-01 3.49126428e-01 -2.60989785e-01 9.98905450e-02 6.26729310e-01 1.00978065e+00 -8.13678876e-02 4.68435764e-01 -5.71103990e-01 -1.01035225e+00 4.18448567e-01 9.67892408e-02 -2.49012318e-02 8.94171298e-02 2.08331808e-01 9.18203175e-01 -4.64751035e-01 7.54080713e-01 4.88727540e-01 2.60830760e-01 5.54789901e-01 -3.66543561e-01 -3.14830661e-01 -4.60236758e-01 -1.08998573e+00 1.80201280e+00 6.06692433e-02 3.93406391e-01 3.85132134e-01 -1.34569871e+00 1.24987614e+00 5.09765625e-01 1.22813272e+00 -2.91057467e-01 5.42085946e-01 4.72334266e-01 1.54182732e-01 -8.45815063e-01 -2.01802384e-02 -4.60378647e-01 1.49276316e-01 3.75484914e-01 1.19863205e-01 -4.23974842e-01 5.11398911e-01 -5.09472191e-01 7.76185930e-01 4.25274402e-01 4.31513369e-01 -8.56978357e-01 8.01118672e-01 6.28544092e-01 2.68252462e-01 9.11491811e-02 -5.12207866e-01 4.11292195e-01 1.08381256e-01 -5.54289579e-01 -1.07110941e+00 -9.52485204e-01 -4.90726233e-01 6.28966689e-01 -8.97470042e-02 2.04622880e-01 -1.51195931e+00 -4.15352881e-01 4.52994229e-03 1.42624872e-02 -4.63266820e-01 2.57389545e-01 -6.99194252e-01 -8.27187300e-01 1.57731071e-01 3.01837146e-01 3.47230285e-01 -9.84959185e-01 -8.27141523e-01 2.89510429e-01 -4.58462611e-02 -5.47827601e-01 -4.35055852e-01 2.94836741e-02 -1.41235936e+00 -1.17975271e+00 -9.48849440e-01 -1.05347180e+00 9.55990374e-01 6.97674155e-02 5.84174514e-01 2.79141128e-01 -7.13318408e-01 3.39659810e-01 -1.62105262e-01 -9.78066958e-03 -6.51647389e-01 -2.37356603e-01 3.12449224e-02 -4.20161039e-02 6.27963185e-01 -4.61524755e-01 -5.75316548e-01 2.91438013e-01 -9.36571479e-01 -3.35161626e-01 6.23588800e-01 6.91825449e-01 1.06693709e+00 2.52294362e-01 -4.68474589e-02 -7.77222097e-01 3.75550151e-01 -2.03528583e-01 -3.46838027e-01 1.04884110e-01 -4.43410248e-01 -2.24860117e-01 2.76638418e-01 -3.76491547e-01 -8.27839971e-01 4.10159707e-01 -1.45474508e-01 -1.77476138e-01 -4.72864151e-01 3.53076279e-01 3.03495049e-01 -6.44644439e-01 8.24310958e-01 1.96009949e-01 6.69555664e-01 -5.65913379e-01 5.28285578e-02 7.78289378e-01 6.87292993e-01 -3.94529611e-01 3.87928635e-01 6.01344347e-01 2.01242417e-01 -9.87538815e-01 -3.33577693e-02 -8.51350069e-01 -1.34061706e+00 -7.99665153e-01 1.29043150e+00 -2.81134874e-01 -2.36922428e-01 5.22084951e-01 -6.18786573e-01 -3.53538185e-01 -4.42051440e-01 7.25878477e-01 -8.25225055e-01 7.70748198e-01 -6.22332633e-01 -7.74833262e-01 -4.70865339e-01 -1.41887689e+00 7.47262597e-01 4.61038426e-02 -3.69196594e-01 -1.13774812e+00 8.25217888e-02 7.02614486e-01 3.18120718e-01 4.12001550e-01 1.02326119e+00 -5.39537549e-01 -1.42025769e-01 -4.14379388e-01 2.21437603e-01 5.18973053e-01 2.32361168e-01 7.03111589e-02 -5.75064957e-01 -5.73098958e-02 4.41927403e-01 -9.18674693e-02 5.61375141e-01 8.67060125e-01 9.96541440e-01 1.37055472e-01 -1.89074889e-01 2.54360825e-01 1.61382842e+00 3.37334692e-01 3.98224026e-01 4.54936892e-01 4.15417790e-01 9.89841163e-01 8.53093445e-01 2.15801626e-01 4.43182476e-02 6.48297846e-01 -6.14101849e-02 -5.16364217e-01 -3.19683135e-01 4.25041586e-01 -7.65081402e-03 1.18113112e+00 -5.19717395e-01 7.80301094e-01 -9.68619287e-01 5.28017402e-01 -1.35590506e+00 -7.55501091e-01 -4.54706877e-01 2.20603085e+00 1.04295719e+00 8.98562446e-02 6.47870660e-01 4.11572129e-01 1.00326371e+00 -3.11653823e-01 -1.48771510e-01 -3.37384939e-01 8.54492113e-02 4.09669220e-01 7.38410175e-01 4.83943164e-01 -1.29446137e+00 6.49487197e-01 6.75794888e+00 9.14129853e-01 -1.31169629e+00 3.26561034e-01 4.76245522e-01 3.83410901e-01 3.44858974e-01 -8.29968601e-02 -1.41343385e-01 4.89944667e-01 4.26434845e-01 6.69325963e-02 2.93825060e-01 7.26297617e-01 3.64353538e-01 -7.61546910e-01 -4.42712486e-01 6.74239874e-01 3.87330391e-02 -9.08997118e-01 -3.98129582e-01 8.20473358e-02 3.21152836e-01 -2.87296087e-01 -3.16237301e-01 -3.20078969e-01 -3.29569340e-01 -6.18820965e-01 5.66289842e-01 7.81697750e-01 7.12026179e-01 -4.04327780e-01 8.43312740e-01 1.91765521e-02 -1.12143052e+00 2.21849993e-01 -1.62272438e-01 6.90691620e-02 1.30952790e-01 4.54225868e-01 -5.92755258e-01 3.49451929e-01 6.04707778e-01 -1.87137742e-02 -4.05945659e-01 1.11503720e+00 3.17241132e-01 5.07455349e-01 -2.85570234e-01 7.21977428e-02 1.82844505e-01 -6.85760081e-01 6.35102391e-01 1.24439919e+00 6.34906888e-02 -1.64246798e-01 4.74605590e-01 7.34990776e-01 6.90131903e-01 9.01864767e-01 -4.78386909e-01 3.10331155e-02 -3.49181369e-02 1.62121189e+00 -1.58579779e+00 -3.75998795e-01 -2.64195621e-01 8.17215562e-01 -1.03730738e-01 -1.00703023e-01 -1.96046904e-01 -7.58443698e-02 1.13543142e-02 4.26450312e-01 -1.68986395e-01 -2.18971476e-01 -4.95572656e-01 -5.11627793e-01 -8.51721466e-02 -8.68023336e-01 4.25121099e-01 -4.01631653e-01 -1.21032715e+00 4.64878529e-01 1.84735954e-01 -1.05944979e+00 -2.31436074e-01 -7.34653056e-01 -5.45136750e-01 9.08363760e-01 -4.85258847e-01 -9.48283851e-01 -9.69019625e-03 2.38303363e-01 4.71445054e-01 -1.89717725e-01 8.77978325e-01 1.82509094e-01 -2.86208063e-01 -6.47761747e-02 1.36500821e-01 2.53025621e-01 3.66695940e-01 -1.60746467e+00 -3.81057233e-01 6.85886502e-01 -3.20986181e-01 2.69420356e-01 8.38470221e-01 -1.00003731e+00 -1.08471143e+00 -5.61619759e-01 7.08479404e-01 5.81941120e-02 7.73801863e-01 2.43184283e-01 -8.60057235e-01 2.85866618e-01 5.80717400e-02 1.23270839e-01 8.20703566e-01 -6.10131621e-01 5.65664411e-01 3.42678353e-02 -1.45048964e+00 4.15790766e-01 3.33257377e-01 -2.99122781e-01 -7.26375282e-01 6.01932704e-01 -4.23382968e-01 -4.18351024e-01 -1.57263970e+00 2.87175059e-01 6.00887239e-01 -7.89494395e-01 7.95389354e-01 -3.36743116e-01 1.72535494e-01 -2.46608764e-01 1.30115435e-01 -9.15087700e-01 -3.62694502e-01 -1.33035257e-01 6.08986735e-01 7.81984746e-01 -9.44448709e-02 -1.11445539e-01 9.09374237e-01 5.36640763e-01 -1.25390030e-02 -7.71047890e-01 -1.08653557e+00 -6.66391432e-01 7.09970295e-02 1.25772431e-01 -2.03695491e-01 9.07966793e-01 2.16977552e-01 -3.19314212e-01 1.16088696e-01 -3.39163691e-01 9.67008531e-01 2.04718560e-02 5.18800437e-01 -1.06565189e+00 -2.12118402e-01 -6.39019728e-01 -9.93721485e-01 -1.73440412e-01 -1.14451814e-02 -1.13507938e+00 2.68322438e-01 -1.45641875e+00 1.77608594e-01 -2.79383451e-01 -8.22654888e-02 1.26705408e-01 7.44819269e-02 5.94210327e-01 -4.76085320e-02 5.06909072e-01 2.16666162e-02 -2.48611972e-01 1.48790216e+00 -7.24582970e-02 -8.44030976e-02 9.98486280e-02 5.13949692e-02 9.94205177e-01 9.45333600e-01 -6.12186551e-01 1.25526056e-01 1.45869762e-01 -8.12295973e-01 7.31630176e-02 2.69689083e-01 -1.10147905e+00 1.89744383e-01 -7.64298365e-02 2.59219855e-01 -6.08318865e-01 4.07403782e-02 -8.87340367e-01 3.23675513e-01 8.28312218e-01 -1.67993322e-01 9.01133493e-02 -2.31930912e-01 1.01136953e-01 -3.42159122e-01 -6.99848592e-01 1.07372916e+00 -5.76739550e-01 -5.76356292e-01 -1.42067090e-01 -9.23744977e-01 -2.94307202e-01 1.15920246e+00 -6.56533957e-01 4.33669150e-01 1.94123611e-01 -1.13712871e+00 -3.15016657e-01 5.72725832e-01 -4.04448628e-01 2.85016239e-01 -1.26684380e+00 -5.72614491e-01 6.47670552e-02 -1.82840705e-01 -3.60270232e-01 5.26041627e-01 1.67657387e+00 -1.29069602e+00 1.50013134e-01 -8.31606507e-01 -8.77799690e-01 -1.58849573e+00 4.75921273e-01 4.60062653e-01 -2.50824153e-01 -6.74931109e-01 4.36984479e-01 -3.59902978e-01 -2.22678229e-01 -6.51994869e-02 -3.16410124e-01 -4.41321969e-01 -8.75922367e-02 1.19051509e-01 6.67823911e-01 1.01654358e-01 -1.19997191e+00 -2.50430256e-01 1.08365619e+00 4.58193958e-01 -2.10336626e-01 9.69674468e-01 -1.96953699e-01 -6.63497865e-01 6.27993166e-01 8.71514380e-01 -1.63908452e-02 -8.66636217e-01 -1.43531084e-01 3.89626354e-01 -3.81547302e-01 3.54718029e-01 -5.82172990e-01 -1.20006132e+00 7.23372996e-01 1.11843204e+00 3.24666172e-01 1.19461656e+00 -1.32068366e-01 6.24579847e-01 -1.48318335e-01 3.53366375e-01 -1.63785052e+00 -4.44505602e-01 -2.26088479e-01 6.83880627e-01 -1.04192901e+00 1.39407784e-01 -6.56986833e-01 -6.81382120e-01 1.39002669e+00 1.55364662e-01 -4.33928221e-01 7.49847889e-01 3.89636010e-01 5.29785693e-01 -3.87008399e-01 2.15874121e-01 -3.58156621e-01 5.63710690e-01 5.68792820e-01 5.89746177e-01 2.47469559e-01 -1.24395049e+00 4.11723316e-01 7.24465773e-02 2.24456981e-01 -1.38795912e-01 1.07295346e+00 -6.65761650e-01 -1.07720995e+00 -7.70337880e-01 5.27176857e-01 -8.73802602e-01 3.16383570e-01 -2.55212843e-01 8.17462504e-01 3.09765309e-01 5.05525470e-01 -2.25048382e-02 -2.15471208e-01 1.16625249e-01 2.28097379e-01 8.14096212e-01 -3.97297353e-01 -7.84733593e-01 5.75978577e-01 -1.34461164e-01 -2.38546804e-01 -7.79740334e-01 -8.14584613e-01 -1.51146817e+00 -4.04493473e-02 -1.63064405e-01 4.73088920e-02 1.07980490e+00 1.11396325e+00 -4.45289820e-01 2.86890119e-01 4.88984555e-01 -1.01612306e+00 -5.39637506e-01 -1.03167045e+00 -1.17806959e+00 7.26311803e-01 -4.75951940e-01 -7.70333588e-01 -1.12558790e-01 4.05090541e-01]
[14.151461601257324, -2.5697243213653564]
c7a3f3cc-f7f0-48b1-ad80-36d682ea9f4e
proactive-multi-camera-collaboration-for-3d
2303.03767
null
https://arxiv.org/abs/2303.03767v1
https://arxiv.org/pdf/2303.03767v1.pdf
Proactive Multi-Camera Collaboration For 3D Human Pose Estimation
This paper presents a multi-agent reinforcement learning (MARL) scheme for proactive Multi-Camera Collaboration in 3D Human Pose Estimation in dynamic human crowds. Traditional fixed-viewpoint multi-camera solutions for human motion capture (MoCap) are limited in capture space and susceptible to dynamic occlusions. Active camera approaches proactively control camera poses to find optimal viewpoints for 3D reconstruction. However, current methods still face challenges with credit assignment and environment dynamics. To address these issues, our proposed method introduces a novel Collaborative Triangulation Contribution Reward (CTCR) that improves convergence and alleviates multi-agent credit assignment issues resulting from using 3D reconstruction accuracy as the shared reward. Additionally, we jointly train our model with multiple world dynamics learning tasks to better capture environment dynamics and encourage anticipatory behaviors for occlusion avoidance. We evaluate our proposed method in four photo-realistic UE4 environments to ensure validity and generalizability. Empirical results show that our method outperforms fixed and active baselines in various scenarios with different numbers of cameras and humans.
['Yizhou Wang', 'Fangwei Zhong', 'Xuehai Pan', 'Mickel Liu', 'Hai Ci']
2023-03-07
null
null
null
null
['3d-human-pose-estimation']
['computer-vision']
[-3.2424489e-01 -1.6554534e-01 3.0368559e-02 -3.5488039e-02 -9.4114351e-01 -6.3305795e-01 3.3654845e-01 -1.0940472e-01 -7.5027549e-01 7.7696735e-01 3.6973462e-01 3.2619712e-01 -2.0328736e-02 -3.1315190e-01 -6.1435401e-01 -6.5459371e-01 -1.4662501e-01 7.3818612e-01 5.8328205e-01 -2.0672891e-01 1.0184950e-01 5.0898904e-01 -1.2443513e+00 -7.7734038e-02 6.6267240e-01 3.3318880e-01 2.4271137e-01 1.2879533e+00 7.2381711e-01 1.2901385e+00 -7.9208112e-01 -2.6162016e-01 6.3980323e-01 4.8068982e-02 -4.8746920e-01 3.5453039e-01 3.0224892e-01 -8.7487519e-01 -3.4679112e-01 3.2452103e-01 1.1400139e+00 6.0937715e-01 3.3218324e-01 -1.8101946e+00 -2.7107847e-01 -2.0080955e-01 -9.6612847e-01 2.9256302e-01 1.0154034e+00 6.9408214e-01 5.6843716e-01 -7.4330360e-01 6.0336322e-01 1.3643737e+00 7.2730988e-01 8.0692255e-01 -6.9961256e-01 -4.0381864e-01 4.3123251e-01 1.4727709e-01 -1.2698249e+00 -4.0957108e-01 6.4347833e-01 -3.3057663e-01 1.1119572e+00 1.9393173e-01 9.5542139e-01 1.4008585e+00 1.9488384e-01 8.2673818e-01 8.9501685e-01 -2.5939515e-01 2.7891269e-01 -1.0998640e-01 -5.7561737e-01 5.9755778e-01 3.4360954e-01 9.9390477e-02 -9.2401111e-01 -5.9610111e-01 1.0743443e+00 -5.3142983e-02 9.3078882e-02 -6.2896371e-01 -1.4461662e+00 7.1378797e-01 2.3339543e-01 -4.8405063e-01 -5.5724204e-01 6.2209785e-01 1.3690652e-01 -5.3181402e-02 1.2684366e-01 2.2391190e-01 -1.6159110e-01 -3.8792565e-01 -4.1338828e-01 7.3768938e-01 5.3567892e-01 1.3477596e+00 3.6139804e-01 1.0271368e-01 -2.4332438e-01 4.9329785e-01 4.7330409e-01 8.8304037e-01 -5.9723627e-02 -1.8257215e+00 5.7020319e-01 3.9030367e-01 9.4468063e-01 -1.0843930e+00 -5.8535045e-01 1.8013667e-01 -3.4193453e-01 4.5949641e-01 5.4451334e-01 -6.7216730e-01 -2.1360098e-01 1.4182891e+00 8.4946346e-01 4.1360623e-01 4.0532473e-02 1.5308567e+00 4.9122062e-01 4.8986924e-01 2.1453619e-01 -2.6472998e-01 1.0206854e+00 -1.0999203e+00 -5.0046176e-01 -4.7678903e-01 3.4195012e-01 -6.8228370e-01 7.5416094e-01 3.1008363e-01 -1.4521309e+00 -4.5005435e-01 -6.1419886e-01 9.0273805e-02 4.2917174e-01 3.4322772e-02 4.3213254e-01 5.7177436e-01 -1.1771116e+00 -4.0589981e-02 -9.6062446e-01 -5.0425792e-01 2.7499042e-02 6.6649652e-01 4.3333277e-02 -9.9641688e-02 -8.6636382e-01 9.7901410e-01 -3.1968462e-01 -1.4696832e-01 -1.2494599e+00 -2.4745286e-01 -5.1398027e-01 -4.1590330e-01 7.3072976e-01 -1.0430154e+00 1.4204116e+00 -8.9916569e-01 -1.7065970e+00 5.7248044e-01 -4.0688500e-02 -3.5133472e-01 9.4431019e-01 -5.6827432e-01 3.6069319e-02 3.8246748e-01 1.9599698e-01 1.0027897e+00 6.3444912e-01 -1.5473388e+00 -7.5336713e-01 -1.3177916e-01 4.9210107e-01 1.1099573e+00 -6.1388016e-02 2.4275956e-01 -5.9823513e-01 -4.5328543e-01 -3.5022327e-01 -1.4979841e+00 -7.0491934e-01 4.0714290e-02 1.5623574e-01 -3.4729183e-02 5.5479664e-01 -1.2678686e-01 7.6587278e-01 -1.5398482e+00 1.8568023e-01 -8.5260585e-02 2.6127127e-01 1.3825479e-02 -3.4806851e-02 3.8005498e-01 8.4515840e-01 -3.8542524e-01 4.6399671e-01 -7.2236478e-01 -8.9854151e-02 2.9393497e-01 2.8865167e-01 6.5104789e-01 -4.3214556e-02 7.6072341e-01 -1.1362299e+00 -7.9427862e-01 3.5193872e-01 4.4087330e-01 -7.4957961e-01 5.5220294e-01 -1.8249050e-01 9.3156904e-01 -5.5172521e-01 7.2774094e-01 6.4807558e-01 -3.6527726e-01 2.7002671e-01 4.3438587e-01 -7.9344958e-02 -3.4895051e-01 -1.5716356e+00 1.6462498e+00 -2.5895178e-01 4.1672510e-01 1.2852058e-01 5.9904005e-02 5.0288635e-01 4.5181215e-01 9.4231629e-01 -4.2360169e-01 1.7103559e-01 2.2364652e-02 -3.0227336e-01 -7.1671617e-01 1.0047070e+00 3.7602890e-01 -1.0812461e-01 4.4253492e-01 -5.4938948e-01 1.3638850e-01 -3.2618329e-01 3.5642007e-01 1.2036031e+00 4.3565336e-01 2.9477441e-01 2.6910388e-01 2.8175402e-01 2.8492388e-01 8.6852992e-01 9.7308987e-01 -1.0569637e+00 6.6213977e-01 -1.5740460e-02 -6.5631223e-01 -1.0471247e+00 -1.0316244e+00 7.1147227e-01 1.2420747e+00 8.1590158e-01 3.3730589e-02 -5.6090134e-01 -5.0544858e-01 -4.8258873e-03 2.1570697e-01 -3.1125876e-01 2.8052828e-01 -1.0295534e+00 -7.9472107e-01 4.3078285e-01 5.9569663e-01 4.9204916e-01 -8.7955457e-01 -1.3879321e+00 2.2077131e-01 -6.7821240e-01 -1.3940543e+00 -7.9450244e-01 -5.4376185e-01 -4.6094561e-01 -1.1757252e+00 -8.7844586e-01 -3.9543679e-01 6.0229439e-01 1.0235360e+00 1.0584524e+00 1.5110213e-01 -6.9635403e-03 1.1825136e+00 -4.7605637e-01 -3.6939421e-01 -2.2700831e-01 -1.7027678e-01 3.8114104e-01 -2.6599202e-02 3.1092757e-01 -1.6024242e-01 -1.0752547e+00 7.2516268e-01 -2.2749801e-01 -6.2214758e-02 2.7203283e-01 3.8114911e-01 4.1031599e-01 -6.1063081e-01 4.2817643e-01 -3.8262087e-01 6.8433911e-01 -5.2888107e-01 -7.9194546e-01 1.5589955e-01 -3.3230495e-01 -6.0505229e-01 1.1081851e-01 -7.3491442e-01 -1.2957340e+00 3.9908972e-01 5.2868468e-01 -5.2265877e-01 -9.2127413e-02 -3.5045120e-01 1.5416972e-01 -3.6436316e-01 7.4362040e-01 -3.0168718e-01 -2.0256552e-01 2.8687268e-01 1.2296738e-01 5.0405526e-01 3.4213987e-01 -5.5080104e-01 5.9112543e-01 7.8080297e-01 5.8844078e-02 -5.5719233e-01 -4.7179544e-01 -7.5291479e-01 -5.7908273e-01 -9.0917647e-01 1.1058995e+00 -1.6477759e+00 -1.4270844e+00 4.5659769e-01 -1.3930259e+00 -5.2063757e-01 1.4605901e-01 6.5886515e-01 -7.5957865e-01 5.1663518e-01 -5.8771622e-01 -1.4726244e+00 -4.2072427e-02 -1.1736020e+00 1.3675249e+00 3.3436769e-01 -1.7224196e-01 -9.6734673e-01 3.6303747e-01 6.5973651e-01 2.0073320e-01 5.8847433e-01 -2.3287201e-01 5.7922095e-02 -1.2085907e+00 5.8760639e-02 2.5838828e-01 -5.7542217e-01 -3.1363532e-01 -2.4696650e-01 -8.3120698e-01 -7.8775561e-01 -1.9780318e-01 -4.9275631e-01 1.1258491e-01 7.1637630e-01 4.2330119e-01 -2.2761197e-01 -3.2638645e-01 2.4263518e-01 1.2430221e+00 1.1545546e-02 4.0346485e-01 7.5436699e-01 8.5086548e-01 3.8162670e-01 1.1797181e+00 1.1505494e+00 1.1952606e+00 9.6546960e-01 7.7075326e-01 -9.7504139e-02 1.4953473e-01 2.2013832e-02 6.4526546e-01 4.1556549e-01 -4.9691454e-01 -7.3647445e-01 -9.1410166e-01 5.5437112e-01 -2.2810922e+00 -1.0883510e+00 -3.5803479e-01 2.1856513e+00 2.3632057e-01 -3.5612691e-02 7.4414694e-01 -3.8593408e-01 7.9245752e-01 2.2746645e-02 -7.1377498e-01 6.6202551e-02 -1.0207634e-01 -6.8794918e-01 8.5163754e-01 7.3336911e-01 -9.3170536e-01 9.2748982e-01 6.1100240e+00 -8.5399840e-03 -4.1928285e-01 5.0002927e-01 2.6102984e-01 -8.1272727e-01 1.7440885e-01 1.1301367e-02 -8.0347520e-01 2.5957125e-01 5.9329039e-01 2.2952732e-01 6.0018897e-01 7.3465443e-01 7.6077878e-01 -5.7999998e-01 -8.0459684e-01 1.1795865e+00 1.5564060e-01 -1.0902992e+00 -3.8120377e-01 1.5174766e-01 1.0216179e+00 2.2088300e-01 -8.4228657e-02 4.4877868e-02 8.5783082e-01 -5.7212812e-01 8.2328612e-01 5.1541829e-01 2.6177889e-01 -7.5314879e-01 4.8021314e-01 5.4333216e-01 -1.2017760e+00 -5.0278997e-01 -4.2519948e-01 -2.9326126e-01 6.6067415e-01 -2.4427317e-01 -8.8674736e-01 2.7370727e-01 7.1265990e-01 4.1365182e-01 -5.5531734e-01 1.0638647e+00 1.1674046e-01 -1.1208509e-01 -1.4236784e-01 -1.8257487e-01 1.7494877e-01 1.5897849e-01 1.0138574e+00 8.3539450e-01 4.1229185e-02 3.4119293e-01 8.3362216e-01 1.6594920e-01 1.9122188e-01 -3.2339039e-01 -5.7645017e-01 8.6817491e-01 8.8293588e-01 1.0840335e+00 -5.6024230e-01 -1.5099029e-01 -5.1009732e-01 1.1132250e+00 3.9008510e-01 4.9140379e-01 -1.2130857e+00 3.4632781e-01 6.2312770e-01 2.8116089e-01 -9.6568950e-02 -7.1632135e-01 1.9855559e-01 -1.1674281e+00 4.1968124e-03 -8.5132468e-01 4.7094861e-01 -1.0198059e+00 -9.9427497e-01 4.6864036e-01 2.1632646e-01 -1.6254352e+00 -3.1391609e-01 -1.3468036e-01 -3.2316464e-01 3.3942291e-01 -1.5608878e+00 -1.3955415e+00 -7.6575643e-01 9.5557988e-01 8.2822555e-01 -2.5203711e-01 3.9881018e-01 2.0263948e-01 -3.4671834e-01 4.2346102e-01 -1.7074794e-01 -1.3199572e-01 1.1268793e+00 -1.2722323e+00 3.9206699e-01 8.9945412e-01 -4.7952431e-01 2.1111581e-01 6.8019426e-01 -9.0385234e-01 -1.6905851e+00 -9.4015491e-01 4.4810650e-01 -1.1605595e+00 1.6077818e-01 -2.3922506e-01 -3.0453429e-01 7.0643836e-01 2.7823815e-01 1.7289306e-01 4.6146542e-01 -3.9823079e-01 4.0433729e-01 9.4561696e-02 -1.1967497e+00 6.8879634e-01 1.1323166e+00 3.6067482e-02 -2.0702312e-02 5.2059078e-01 6.7632252e-01 -9.0491527e-01 -6.8517667e-01 -1.1014588e-01 6.6073161e-01 -1.0566670e+00 9.3830264e-01 -2.6316091e-01 -2.1594249e-01 -4.5285159e-01 -1.8744448e-01 -1.0572571e+00 -3.3650640e-01 -1.2949817e+00 -3.2035440e-01 7.1682370e-01 -6.9134280e-02 -2.2848003e-01 1.0910422e+00 1.0287716e+00 6.5582052e-02 -2.0285633e-01 -1.0413072e+00 -6.7490023e-01 -2.7474245e-01 -1.3139778e-01 4.3680772e-01 6.6348350e-01 -2.1312578e-01 1.5991832e-01 -1.1547730e+00 6.5578794e-01 9.4853276e-01 -4.4289908e-01 1.4635986e+00 -9.2119664e-01 -4.5586768e-01 2.3108003e-01 -3.4325399e-02 -1.1514161e+00 -1.7936982e-01 -8.6888932e-02 1.1681623e-01 -1.4966077e+00 3.3361155e-01 -5.5619591e-01 3.2116523e-01 -5.3991857e-03 -2.6160339e-01 2.9871829e-03 5.7897848e-01 7.3292339e-01 -1.5024408e+00 5.0958872e-01 1.3136052e+00 3.4017277e-01 -5.0156963e-01 1.7896964e-01 -1.8028669e-01 7.9742551e-01 6.3400805e-01 -6.5147978e-01 -4.8945963e-01 -1.0728208e+00 4.1540861e-01 7.2014010e-01 7.4960339e-01 -1.2778487e+00 7.5251555e-01 -6.7410320e-01 3.9839113e-01 -4.8492453e-01 7.0283151e-01 -1.0263138e+00 2.4849355e-01 5.0847483e-01 -1.8078734e-01 8.2902235e-01 -7.4191548e-02 1.0622007e+00 5.2765888e-01 9.6684501e-02 6.7885727e-01 -4.9676424e-01 -5.4011828e-01 3.4644851e-01 -7.6535970e-01 2.0419446e-01 1.4011064e+00 -5.2096760e-01 -3.3290550e-01 -8.1694573e-01 -6.1790657e-01 8.3060563e-01 8.4118932e-01 5.1319069e-01 7.8298444e-01 -1.3780003e+00 -7.2165406e-01 -3.3099917e-01 6.3765501e-03 1.1569647e-01 2.4260415e-01 8.6749399e-01 -7.2333366e-01 3.5224963e-02 -2.3520538e-01 -7.7949440e-01 -1.4464481e+00 3.1921300e-01 5.0797594e-01 -2.7291140e-01 -3.4897295e-01 4.7234061e-01 -2.3830710e-01 -2.5497180e-01 4.2292124e-01 1.6464721e-01 -6.7193471e-02 -4.4975346e-01 3.9163822e-01 1.0257355e+00 -4.5344290e-01 -8.5110432e-01 -4.7641498e-01 5.5957264e-01 2.0078233e-01 -4.4538775e-01 1.1526532e+00 -9.2907745e-01 5.0214440e-01 1.0244952e-01 4.2317379e-01 2.3927315e-01 -2.2125838e+00 -4.5543820e-02 -2.1996948e-01 -8.5635912e-01 -2.9729277e-01 -6.3746786e-01 -7.3066640e-01 2.7571824e-01 8.7066495e-01 -2.4624972e-01 7.3305678e-01 -2.9329261e-01 8.1673259e-01 5.2419633e-01 8.0364060e-01 -1.4635469e+00 9.3614537e-01 3.2371935e-01 6.4088506e-01 -1.6994238e+00 2.7512604e-01 -1.9784725e-01 -1.4347087e+00 6.7674488e-01 1.3124579e+00 -2.3968399e-01 7.9408951e-02 1.0719287e-01 3.0127159e-01 -2.1214466e-01 -9.1359001e-01 -9.6228547e-02 -3.2566389e-01 1.1029971e+00 -5.6465048e-02 -1.0313120e-01 2.0674703e-01 8.3349071e-02 3.4697139e-01 -8.6903155e-02 1.0761989e+00 1.2519411e+00 -4.8901755e-01 -8.4346521e-01 -8.6266017e-01 -2.6687506e-01 -2.7578449e-01 6.1349845e-01 -4.2579219e-01 8.0166286e-01 3.3431999e-02 1.3933382e+00 -2.2609595e-02 -1.1565836e-01 4.5674902e-01 -6.4887708e-01 6.6601634e-01 -4.0310490e-01 -9.2544895e-01 4.1830245e-01 3.5923358e-02 -6.0038805e-01 -9.6844822e-01 -1.0287156e+00 -1.2126395e+00 -5.6063914e-01 -3.7445322e-01 -2.3667930e-01 1.8177623e-01 7.2207224e-01 5.3896970e-01 2.8035876e-01 7.8970307e-01 -1.3369415e+00 -4.0737352e-01 -5.5811369e-01 -9.1698624e-02 4.4254935e-01 4.9355313e-01 -8.6444622e-01 -2.7438376e-02 7.9213902e-02]
[7.129496097564697, -0.960759162902832]
72307e60-4401-42b6-9540-a07fb94ff0d2
fast-quasi-optimal-power-flow-of-flexible-dc
2211.02852
null
https://arxiv.org/abs/2211.02852v1
https://arxiv.org/pdf/2211.02852v1.pdf
Fast Quasi-Optimal Power Flow of Flexible DC Traction Power Systems
This paper proposes a quasi-optimal power flow (OPF) algorithm for flexible DC traction power systems (TPSs). Near-optimal solutions can be solved with high computational efficiency by the proposed quasi-OPF. Unlike conventional OPF utilizing mathematical optimization algorithms, the proposed quasi-OPF adopts analytical mapping from load information to near-optimal solutions, hence considerably accelerating the computation. First, we study the mechanism and physical meaning interpretation of conventional OPF based on a new modeling method and successfully interpret the mechanism of conventional OPF in flexible DC TPSs. Then, the analytical mapping from load information to near-optimal solutions is obtained inspired by the mechanism of conventional OPF, and the quasi-OPF algorithm is designed based on the mapping. Since the mapping is based on simple arithmetic, the quasi-OPF algorithm can solve OPF with much less execution time, achieving subsecond level calculation and a speed-up of 57 times compared to conventional OPF. The effectiveness is verified by mathematical proofs and a case study with Beijing Metro Line 13. It provides an insight into the mechanism and physical meaning of OPF, and is a powerful tool for flexible DC TPSs to analyze the effects of coordinated control, design real-time control strategies, and solve operational problems in planning.
['Xuelian Bai', 'Chao Lu', 'Yingdong Wei', 'Xiaoqian Li', 'Zhanhe Li']
2022-11-05
null
null
null
null
['mathematical-proofs']
['miscellaneous']
[-2.47572273e-01 1.29456908e-01 -5.35269320e-01 8.82935151e-02 2.28819996e-02 -5.37662804e-01 -4.92539853e-02 -1.42139941e-01 1.04183212e-01 1.17217910e+00 -3.48555744e-01 -7.92036295e-01 -1.12783611e+00 -9.57455695e-01 -5.45330942e-01 -9.26350951e-01 -3.55064780e-01 5.83600700e-01 8.27786922e-02 -4.49586481e-01 4.47197139e-01 7.95906186e-01 -1.18667936e+00 -4.57948595e-01 1.47981870e+00 9.94955420e-01 5.16926289e-01 3.52455199e-01 2.32248846e-02 2.54452139e-01 -7.51741529e-01 1.01721443e-01 3.48436803e-01 -8.94698501e-02 -8.19377780e-01 -1.43723086e-01 -4.37876403e-01 -4.51466322e-01 -3.76737177e-01 8.82911146e-01 3.83109063e-01 2.41258278e-01 5.65576315e-01 -2.09120870e+00 -3.87765884e-01 3.57957572e-01 -4.74028856e-01 5.97465597e-02 5.74287951e-01 7.36453831e-02 6.09382629e-01 -1.64662167e-01 2.97570884e-01 8.99370372e-01 6.41168594e-01 2.19247520e-01 -8.73469293e-01 -3.68635029e-01 -1.83180287e-01 3.90667975e-01 -1.33300495e+00 2.57744730e-01 5.44103503e-01 -8.38162750e-03 1.21503687e+00 4.67635930e-01 1.16221309e+00 -2.14115724e-01 9.07432020e-01 5.78904748e-01 7.19834268e-01 -4.09467280e-01 2.40692779e-01 -4.77365524e-01 3.22242044e-02 7.48558998e-01 4.41326797e-01 -2.54263338e-02 5.09619266e-02 3.59986536e-02 4.46291357e-01 -4.89662476e-02 -6.85958147e-01 8.99519250e-02 -8.02802980e-01 5.02934575e-01 4.36795056e-01 3.82149994e-01 -2.53752947e-01 4.68808599e-02 1.79910108e-01 -6.16930872e-02 -2.23582927e-02 4.86366332e-01 -5.58441401e-01 -3.13864648e-01 -8.05768549e-01 4.50897098e-01 8.15418541e-01 1.18654263e+00 2.18668595e-01 2.42443413e-01 -1.99624255e-01 3.79461199e-01 2.29386628e-01 9.15871978e-01 8.58860984e-02 -1.36989748e+00 5.34581006e-01 5.15283346e-01 3.56643021e-01 -6.31457984e-01 -8.20627511e-01 -1.82288721e-01 -5.97295344e-01 3.43970537e-01 4.09161210e-01 -4.16190147e-01 -4.10871893e-01 1.06141961e+00 1.80737063e-01 -3.46663326e-01 -2.94276010e-02 8.28295231e-01 4.04066853e-02 1.30374897e+00 -4.50715452e-01 -9.09905314e-01 1.30345273e+00 -8.93860221e-01 -1.26214433e+00 5.04244149e-01 7.09517002e-01 -5.14600575e-01 7.53896713e-01 3.25170457e-01 -1.21720088e+00 -2.82028709e-02 -1.23625541e+00 1.43954933e-01 -2.80627817e-01 4.93575744e-02 5.66007853e-01 4.71015275e-01 -9.46624637e-01 8.62392485e-01 -6.04341149e-01 -3.30678523e-01 3.91491771e-01 4.81973022e-01 2.76986491e-02 -1.15312263e-02 -1.03030956e+00 1.24515426e+00 5.27037799e-01 5.60203791e-01 -1.12239301e-01 -1.20520234e+00 -5.44923067e-01 4.06038135e-01 5.35342157e-01 -7.96649277e-01 1.17201102e+00 -1.85970530e-01 -1.88445628e+00 -4.13030535e-01 -5.05935550e-01 -1.08057648e-01 2.68266529e-01 3.36730301e-01 -3.92493248e-01 4.32093292e-01 1.38413504e-01 -1.90083813e-02 2.19093844e-01 -8.51169407e-01 -8.85712266e-01 1.74459189e-01 2.47412518e-01 9.44352821e-02 -8.50040838e-02 -2.88962185e-01 2.68241882e-01 -1.49275646e-01 6.45122677e-02 -6.19555652e-01 -1.93651959e-01 1.11329339e-01 -2.38564238e-01 -7.14719772e-01 1.18668425e+00 -6.07005894e-01 1.31880999e+00 -1.61666799e+00 8.82959515e-02 6.33211493e-01 -7.88788125e-02 2.15790048e-01 2.90447950e-01 7.08592057e-01 1.93672374e-01 4.61541116e-02 -6.56260774e-02 5.51624119e-01 4.08911765e-01 7.34786034e-01 -3.08168501e-01 5.49454093e-01 -5.31449169e-02 8.18084836e-01 -1.02269006e+00 -3.85670364e-01 4.52495188e-01 -4.17580530e-02 -5.22248924e-01 1.77582111e-02 -4.18559201e-02 -9.29766670e-02 -5.82868516e-01 7.04337180e-01 1.09603894e+00 2.12804094e-01 3.75905663e-01 -6.97607756e-01 -6.65282667e-01 -1.40482590e-01 -9.64505136e-01 1.21663177e+00 -6.15642190e-01 6.53206110e-01 3.40840518e-01 -1.34415805e+00 7.51522601e-01 1.59178779e-01 8.69411230e-01 -1.08698893e+00 2.32368648e-01 3.88205439e-01 -7.75128901e-02 -6.12665713e-01 6.09984159e-01 -1.63304776e-01 7.38606006e-02 4.55639660e-01 -2.28688214e-02 -3.47012669e-01 8.66499782e-01 1.74859852e-01 9.11709011e-01 1.54035185e-02 -7.34086260e-02 -1.13186574e+00 7.71236658e-01 3.10764432e-01 8.55109930e-01 1.79983675e-01 -5.14900722e-02 -3.59245539e-01 7.41514027e-01 -4.71591592e-01 -7.88570344e-01 -8.97772551e-01 -5.18140733e-01 1.86830804e-01 1.03351998e+00 -5.62764943e-01 -5.01908064e-01 -5.49557269e-01 4.11981612e-01 1.01882696e+00 9.94136110e-02 4.93093915e-02 -6.51788771e-01 -7.46160686e-01 7.55871981e-02 5.74713349e-01 6.64436936e-01 -2.30986059e-01 -4.84429300e-01 2.82655150e-01 -1.33958399e-01 -1.08776271e+00 -3.72580409e-01 -1.44013828e-02 -8.77164066e-01 -1.36383224e+00 -5.70183992e-01 -6.18234515e-01 1.15685678e+00 4.21400696e-01 5.56994617e-01 6.05123281e-01 -8.45812708e-02 1.81770533e-01 -1.32206708e-01 -1.44233912e-01 -5.81481643e-02 -1.12471372e-01 2.61544287e-01 -6.40914202e-01 -3.45717490e-01 -4.17428702e-01 -4.23257321e-01 6.87818170e-01 -4.76867676e-01 1.75796688e-01 3.45986456e-01 7.48358130e-01 2.99495339e-01 1.06105042e+00 7.00263798e-01 -1.38712019e-01 7.44104445e-01 -2.77122319e-01 -1.26456809e+00 4.65944439e-01 -9.41934228e-01 1.27889156e-01 1.39191341e+00 9.31971744e-02 -1.24676979e+00 -5.27873337e-01 -1.04644626e-01 -2.58329242e-01 4.83111739e-01 2.60388315e-01 -1.77339718e-01 -4.77004200e-01 -2.69559234e-01 2.44746301e-02 1.24318168e-01 -3.85541022e-01 5.33467159e-02 4.58099514e-01 3.82405728e-01 -9.40630674e-01 1.15615773e+00 2.86602974e-01 7.77625680e-01 -5.74565291e-01 -4.17866975e-01 2.48722985e-01 -4.07173306e-01 -5.07097125e-01 4.64888155e-01 -2.61213452e-01 -1.56545389e+00 1.77845940e-01 -1.14309764e+00 -2.73429692e-01 -2.47746453e-01 4.69321162e-01 -4.38957602e-01 4.43349957e-01 -4.71177191e-01 -7.53166080e-01 -3.39883566e-01 -1.16152632e+00 7.02199340e-01 5.36979198e-01 7.20652640e-02 -1.28353453e+00 -5.22484004e-01 2.03714371e-01 6.06964767e-01 4.15514857e-01 1.21002126e+00 2.71237284e-01 -6.65216923e-01 1.07965149e-01 -1.33833349e-01 -1.35318458e-01 8.25794041e-02 5.38132191e-01 -9.68024209e-02 -6.51417196e-01 -2.95060337e-01 5.66699505e-01 -3.00442606e-01 5.65842152e-01 1.22104430e+00 -6.29576325e-01 -7.42949843e-01 5.79988778e-01 1.79879642e+00 7.30184257e-01 6.51135683e-01 3.80077004e-01 1.48628978e-02 1.80556253e-01 1.17151165e+00 5.19130647e-01 5.85354149e-01 7.51496553e-01 3.86584044e-01 -1.02678694e-01 3.91822994e-01 -1.61370486e-01 2.52151787e-01 8.23210597e-01 -5.14458418e-01 -2.30462357e-01 -4.85696644e-01 2.81439573e-01 -1.80788803e+00 -9.20414925e-01 -4.40234542e-01 1.90499008e+00 3.17734063e-01 1.15450442e-01 -8.77825767e-02 5.98492920e-01 8.04429412e-01 -4.22534823e-01 -6.41800985e-02 -9.61704314e-01 1.73414826e-01 1.75179735e-01 9.18832421e-01 6.77699029e-01 -2.16709465e-01 4.27089632e-02 6.58512878e+00 1.29443169e+00 -1.11515391e+00 -1.99564975e-02 -1.07764505e-01 -8.15711170e-02 -3.00223410e-01 3.70498627e-01 -5.56126118e-01 8.51668715e-01 7.07656145e-01 -1.25609207e+00 8.98691356e-01 6.59940720e-01 7.52984643e-01 -5.65943301e-01 -7.95724392e-01 6.67310178e-01 -5.67504942e-01 -1.59006464e+00 -1.35686472e-01 1.70381486e-01 6.41903937e-01 -4.86717492e-01 -8.35426033e-01 2.67770320e-01 -3.45162630e-01 -4.98157799e-01 9.03272331e-01 5.00174105e-01 5.35003066e-01 -9.14514720e-01 1.02969408e+00 4.46523517e-01 -1.54827058e+00 -5.81236720e-01 -3.74712616e-01 -3.36369157e-01 8.45475256e-01 6.83571517e-01 -3.86886775e-01 1.43595386e+00 5.23363531e-01 5.20529509e-01 1.12752356e-01 9.51174736e-01 -2.95937181e-01 1.46008745e-01 -6.98115528e-01 -3.32907110e-01 5.36373034e-02 -5.64293504e-01 4.52656835e-01 5.51652908e-01 7.59104550e-01 2.85357863e-01 2.59170420e-02 9.71002758e-01 4.45763439e-01 -2.34902620e-01 -2.41726756e-01 1.41344950e-01 7.76270330e-01 1.38548243e+00 -7.09326923e-01 -2.22409561e-01 -1.66695833e-01 2.37263337e-01 -4.06287611e-01 4.61674362e-01 -1.46675158e+00 -1.16133904e+00 5.10314643e-01 2.67773837e-01 -8.78903642e-02 -3.71717185e-01 -3.33543599e-01 -6.76828146e-01 1.27444237e-01 1.13100231e-01 1.86772034e-01 -1.14285755e+00 -8.12526226e-01 9.22714621e-02 4.85889465e-01 -1.31878471e+00 -2.57742077e-01 -8.50022614e-01 -1.17027140e+00 9.77099597e-01 -1.76455522e+00 -6.65369630e-01 -2.53576130e-01 4.57701236e-01 2.65739828e-01 6.90265596e-02 3.31510097e-01 5.56938887e-01 -7.99179971e-01 3.92730683e-01 4.73865181e-01 -4.78714168e-01 -2.98085064e-01 -1.07421517e+00 -5.99755645e-01 8.22739244e-01 -9.87609386e-01 2.37487316e-01 7.38612354e-01 -4.13883269e-01 -2.30772114e+00 -5.61150193e-01 7.97353029e-01 3.58343363e-01 6.48229837e-01 1.01624958e-01 -5.69348872e-01 2.46701136e-01 4.22711521e-01 -8.47169608e-02 9.63314772e-02 -3.88432205e-01 8.66894424e-01 -6.29015744e-01 -1.63602507e+00 4.06467050e-01 8.53684664e-01 -9.08396319e-02 -5.42800128e-01 4.73353624e-01 5.60691595e-01 -5.02619505e-01 -1.32755244e+00 5.38798988e-01 6.53799355e-01 -4.83681530e-01 5.88556588e-01 -2.78182179e-02 -6.28399849e-03 -9.89484608e-01 2.00238332e-01 -1.47843671e+00 -3.93517226e-01 -1.14152658e+00 -1.18154407e-01 1.07849526e+00 8.30194652e-02 -1.23004508e+00 1.33480132e-01 6.40248060e-01 -7.06106007e-01 -1.14506197e+00 -1.42381608e+00 -1.15390944e+00 -1.66108593e-01 -3.34778279e-01 9.59041297e-01 8.02121103e-01 6.37401164e-01 -2.12952361e-01 3.38940978e-01 4.94190156e-01 5.83799303e-01 4.43666786e-01 4.54293579e-01 -7.02412784e-01 4.74509224e-02 -3.62447739e-01 -1.96616799e-01 -9.99235392e-01 1.38727218e-01 -9.49546516e-01 -2.83767968e-01 -2.10598612e+00 -2.98302650e-01 -5.03247678e-01 1.93241790e-01 6.47241414e-01 4.10014808e-01 -1.51973724e-01 4.95902896e-01 5.54004386e-02 -1.51332065e-01 7.08465576e-01 1.64304912e+00 -1.35549709e-01 7.81850070e-02 -1.42992511e-01 -2.85810053e-01 5.26834846e-01 6.58829331e-01 -1.99732870e-01 -3.50936830e-01 -3.85079354e-01 1.46723062e-01 4.02967244e-01 6.70401528e-02 -7.57133424e-01 3.82862747e-01 -5.60094416e-01 2.93624606e-02 -8.32040548e-01 -1.80686429e-01 -1.42559159e+00 3.79173428e-01 1.06109476e+00 6.33830607e-01 1.80898219e-01 3.89226645e-01 4.94215786e-02 6.78176880e-02 -5.33794701e-01 4.02175665e-01 6.09942079e-01 -5.57254136e-01 -1.18273981e-01 -6.64589882e-01 -4.25879091e-01 1.45566547e+00 -4.02110308e-01 -9.93673921e-01 -9.67742577e-02 -4.48071122e-01 1.02140355e+00 1.47005573e-01 2.73352433e-02 2.37927973e-01 -1.60605752e+00 -5.04215546e-02 2.35885739e-01 -7.91940033e-01 1.44269122e-02 2.47341633e-01 1.28450572e+00 -1.17662299e+00 7.96001911e-01 -2.81891346e-01 -7.17408538e-01 -8.07991922e-01 2.90850192e-01 5.23033619e-01 -2.27881879e-01 -5.06299257e-01 -1.86895460e-01 -4.14043486e-01 -1.35423809e-01 -3.34555030e-01 -8.12385678e-01 2.81377465e-01 -2.18762398e-01 1.12005398e-01 1.15670657e+00 1.48350343e-01 -2.33872414e-01 -3.55751812e-01 1.04075766e+00 8.27675045e-01 4.56514686e-01 1.36472762e+00 -1.99913025e-01 -3.16012174e-01 -3.88996422e-01 9.71553922e-01 1.00960493e-01 -9.40838158e-01 6.56866968e-01 -3.40173692e-01 -7.67984211e-01 2.72229314e-01 -8.38457167e-01 -1.43368089e+00 4.32300329e-01 1.85649514e-01 4.82621163e-01 1.52005172e+00 -4.31500286e-01 9.76320267e-01 4.60444719e-01 8.86287868e-01 -1.44568062e+00 -6.16564035e-01 4.65147585e-01 8.62313807e-01 -2.42623553e-01 5.35880625e-01 -7.45100975e-01 -4.42012027e-02 1.79177046e+00 5.12265623e-01 2.28997748e-02 6.28734410e-01 7.49003053e-01 -6.59428477e-01 7.60439858e-02 -5.75720668e-01 1.99496433e-01 -1.60923600e-01 2.59949386e-01 -2.81345904e-01 -2.93465480e-02 -1.10027552e+00 8.75819087e-01 -1.08209662e-01 5.39131463e-01 7.06969023e-01 1.27527094e+00 -3.50557566e-01 -1.00579476e+00 -5.32896399e-01 3.31529468e-01 -1.13366038e-01 5.51266968e-01 7.24487007e-01 1.17790198e+00 2.77148932e-01 7.60817289e-01 4.21778768e-01 -1.76898956e-01 8.16397130e-01 -2.89514214e-01 7.49889672e-01 3.96092206e-01 -1.51932880e-01 -2.76190072e-01 1.91320658e-01 -4.63457942e-01 -3.49914849e-01 -1.97428092e-01 -2.09238815e+00 -9.24358904e-01 -7.22669423e-01 7.76595294e-01 7.53976405e-01 1.26556277e+00 2.72119343e-01 7.25502610e-01 1.05138600e+00 -9.97192383e-01 -3.89403790e-01 -3.71815085e-01 -8.73939991e-01 -2.08307981e-01 -2.05223694e-01 -1.02071536e+00 -5.66923320e-01 -6.17405713e-01]
[5.5834431648254395, 2.411595344543457]
d06faca4-b05e-4f7f-8139-18d4a67dd656
visual-compositional-learning-for-human
2007.12407
null
https://arxiv.org/abs/2007.12407v2
https://arxiv.org/pdf/2007.12407v2.pdf
Visual Compositional Learning for Human-Object Interaction Detection
Human-Object interaction (HOI) detection aims to localize and infer relationships between human and objects in an image. It is challenging because an enormous number of possible combinations of objects and verbs types forms a long-tail distribution. We devise a deep Visual Compositional Learning (VCL) framework, which is a simple yet efficient framework to effectively address this problem. VCL first decomposes an HOI representation into object and verb specific features, and then composes new interaction samples in the feature space via stitching the decomposed features. The integration of decomposition and composition enables VCL to share object and verb features among different HOI samples and images, and to generate new interaction samples and new types of HOI, and thus largely alleviates the long-tail distribution problem and benefits low-shot or zero-shot HOI detection. Extensive experiments demonstrate that the proposed VCL can effectively improve the generalization of HOI detection on HICO-DET and V-COCO and outperforms the recent state-of-the-art methods on HICO-DET. Code is available at https://github.com/zhihou7/VCL.
['DaCheng Tao', 'Yu Qiao', 'Xiaojiang Peng', 'Zhi Hou']
2020-07-24
null
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2400_ECCV_2020_paper.php
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123600579.pdf
eccv-2020-8
['affordance-recognition']
['computer-vision']
[-1.31076440e-01 -5.45176864e-01 -1.76595390e-01 1.54001107e-02 -4.10732538e-01 -3.32664490e-01 6.79058254e-01 -2.16417372e-01 -9.72452983e-02 2.52663136e-01 2.58316398e-01 1.62100911e-01 1.15963496e-01 -5.91522753e-01 -5.17942727e-01 -6.78514779e-01 -1.04521886e-01 2.97136277e-01 5.43857515e-01 4.36676331e-02 -1.35275260e-01 3.36132258e-01 -2.00559044e+00 6.40810609e-01 6.49975538e-01 1.15818346e+00 4.11555976e-01 4.83696878e-01 -9.23534483e-03 8.89446497e-01 -6.28283978e-01 -1.25318110e-01 2.37489268e-01 -5.38104892e-01 -5.87487042e-01 3.90241474e-01 4.05311376e-01 -3.04051489e-01 -4.30260688e-01 1.04763818e+00 5.55236280e-01 1.31925166e-01 7.89343596e-01 -1.84793866e+00 -9.47244823e-01 3.34701657e-01 -8.46036732e-01 2.16669053e-01 4.98181760e-01 5.59377491e-01 1.02142048e+00 -1.44334590e+00 6.94501162e-01 1.61139166e+00 3.51194352e-01 4.09385085e-01 -9.76368964e-01 -9.65617180e-01 4.76894528e-02 5.05691648e-01 -1.64317763e+00 -1.92004174e-01 7.20122755e-01 -6.48749650e-01 6.99840963e-01 2.56929725e-01 1.01816690e+00 1.16509354e+00 -8.31402913e-02 1.63548827e+00 7.40914464e-01 -3.39074641e-01 -2.47865319e-01 5.66774905e-02 2.36537740e-01 8.54002357e-01 4.41216916e-01 1.38997838e-01 -7.19600558e-01 -4.42096666e-02 6.76853001e-01 4.18268740e-01 -4.10339028e-01 -4.77420360e-01 -1.59219277e+00 7.55775988e-01 6.43417358e-01 2.57015139e-01 -1.87544242e-01 1.36296600e-01 5.56709468e-01 -1.51726082e-01 -1.04826696e-01 3.94548066e-02 5.05547039e-02 3.58552307e-01 -3.86184692e-01 4.88244981e-01 3.84753376e-01 1.35563040e+00 7.42023230e-01 -1.49507791e-01 -6.62582636e-01 8.98739159e-01 3.33454847e-01 6.52249217e-01 4.70979571e-01 -8.76315594e-01 5.12917876e-01 8.11903834e-01 1.09064735e-01 -1.06475425e+00 -2.43676558e-01 -3.36348265e-01 -8.82069230e-01 2.63768137e-01 2.20287949e-01 1.45941556e-01 -8.80218625e-01 1.56072748e+00 5.94154358e-01 -1.08259507e-01 -3.34151715e-01 1.10330546e+00 1.31683648e+00 7.54968762e-01 9.33478922e-02 -8.72222260e-02 1.71892607e+00 -1.38695860e+00 -8.36622775e-01 -4.65243578e-01 4.25630450e-01 -6.35952771e-01 1.37967741e+00 2.04855770e-01 -9.50633705e-01 -8.30410063e-01 -9.16529775e-01 -3.88613433e-01 -3.14065903e-01 4.37658966e-01 7.55947828e-01 8.62424821e-02 -4.38002169e-01 -4.41412069e-02 -4.02403861e-01 -2.50146270e-01 7.61396170e-01 1.12880811e-01 -3.84751529e-01 -2.25718901e-01 -1.14883471e+00 4.24402505e-01 5.86106181e-01 1.89449355e-01 -1.03193879e+00 -6.06960237e-01 -9.98756468e-01 1.57843828e-01 8.48942399e-01 -8.06515217e-01 1.15840328e+00 -8.24388206e-01 -9.02136743e-01 8.19058836e-01 -2.40849644e-01 -8.17425102e-02 5.30458212e-01 -1.94603130e-01 -3.99786919e-01 7.28473142e-02 3.44096154e-01 7.43212044e-01 1.06057239e+00 -1.49093831e+00 -9.74495828e-01 -2.81627864e-01 -1.61668390e-01 3.86303991e-01 -2.27274880e-01 1.26332924e-01 -9.17476892e-01 -5.57824194e-01 -7.80065507e-02 -9.53612268e-01 2.87270486e-01 3.90098631e-01 -4.77222055e-01 -6.48244679e-01 1.04303563e+00 -2.85680324e-01 1.33810735e+00 -2.30884767e+00 1.22978918e-01 -2.07571089e-01 5.91283619e-01 4.78243083e-01 -2.75545210e-01 3.95326853e-01 6.62726834e-02 -1.75459921e-01 2.76250357e-04 -2.00297177e-01 -2.84632500e-02 3.45579088e-02 -1.97701249e-02 4.85767156e-01 -1.53284501e-02 1.21490526e+00 -1.07303619e+00 -7.80283689e-01 4.82079297e-01 3.45086306e-01 -3.35332245e-01 5.03161192e-01 -3.35801505e-02 3.75545055e-01 -1.82922155e-01 9.72219944e-01 6.20451987e-01 -4.82982844e-01 -3.10005783e-03 -4.73190188e-01 -5.29964343e-02 -9.65052247e-02 -1.09316158e+00 1.33879042e+00 -1.86058983e-01 6.45544946e-01 -1.45617247e-01 -8.11348855e-01 6.93242729e-01 8.84465799e-02 3.77597183e-01 -6.21858120e-01 2.99230337e-01 6.72817826e-02 2.41879188e-02 -8.17694545e-01 6.69572651e-02 1.68851078e-01 -1.75404012e-01 2.77494173e-03 3.19555342e-01 1.81459755e-01 2.84990638e-01 1.46334380e-01 7.77140260e-01 -6.06066920e-02 8.58928204e-01 -7.04142302e-02 6.15919352e-01 -3.25069427e-01 7.37012029e-01 8.78553152e-01 -5.90013146e-01 4.97518778e-01 3.86674583e-01 -4.47161019e-01 -7.58661151e-01 -1.21104836e+00 3.58782224e-02 1.24990642e+00 5.90725005e-01 -4.02012080e-01 -4.16092575e-01 -8.49253118e-01 1.66753277e-01 5.50858855e-01 -7.96287537e-01 -1.59565076e-01 -3.98809880e-01 -3.61784697e-01 1.78349484e-02 6.17015302e-01 6.51903868e-01 -1.56463003e+00 -6.69882298e-01 -9.97794420e-02 -5.49947321e-01 -9.96414840e-01 -8.53196502e-01 1.98905561e-02 -2.22662583e-01 -1.09095156e+00 -6.68034315e-01 -1.13753927e+00 4.12038207e-01 9.34177995e-01 1.08528280e+00 4.45428193e-02 -9.05094743e-01 2.10550234e-01 -5.25564611e-01 -4.73719835e-01 -6.80992156e-02 -2.53754765e-01 -1.39955357e-01 2.32313350e-01 5.43539643e-01 -2.75869608e-01 -8.15981984e-01 4.12436813e-01 -7.86394835e-01 2.84005851e-01 5.16010404e-01 1.10343218e+00 5.27974904e-01 7.26475045e-02 1.38257757e-01 -6.87247276e-01 2.22906992e-01 -6.41675353e-01 -3.81161511e-01 4.47681695e-01 -2.17136722e-02 -2.29953811e-01 4.04163420e-01 -6.59798980e-01 -1.06023216e+00 1.99211538e-01 3.78745526e-01 -8.36776674e-01 -2.42599323e-01 2.25111976e-01 -5.06107748e-01 4.52727340e-02 5.47179401e-01 1.57995090e-01 -2.14393973e-01 -4.19096261e-01 4.90379840e-01 7.60820031e-01 7.15823531e-01 -1.96013600e-01 8.27463567e-01 5.87555707e-01 -2.64699578e-01 -9.29508507e-01 -1.07112992e+00 -9.04640317e-01 -6.41418695e-01 -3.81816536e-01 1.05806792e+00 -9.76561427e-01 -8.04723203e-01 6.22169077e-01 -1.24082530e+00 -2.87542254e-01 -2.69961625e-01 3.07512641e-01 -4.14807737e-01 5.29163420e-01 -4.35219646e-01 -8.32610965e-01 -2.83819675e-01 -1.15916955e+00 1.22659600e+00 4.19065148e-01 -8.82736966e-02 -4.66210455e-01 -2.86878318e-01 4.41625923e-01 -1.77832484e-01 1.75161287e-01 6.95163012e-01 -2.05117181e-01 -7.26144433e-01 -1.30744502e-01 -6.00468099e-01 2.32334346e-01 1.08394727e-01 -1.18919544e-01 -7.56029546e-01 -3.63284141e-01 -4.59973142e-02 -4.69914138e-01 8.50941718e-01 1.94235429e-01 1.28747272e+00 -1.88569233e-01 -3.97918433e-01 5.84408283e-01 1.26363266e+00 3.46039176e-01 4.66641873e-01 1.60080001e-01 1.10850060e+00 6.42997503e-01 9.18793976e-01 6.83188736e-01 5.06917119e-01 9.63542223e-01 4.74095374e-01 -1.64015844e-01 -6.11607134e-01 -2.94627488e-01 3.84807974e-01 5.93371093e-01 1.93955284e-03 -2.17350721e-01 -8.93620014e-01 7.48026550e-01 -2.23738480e+00 -1.20438504e+00 -3.20568144e-01 1.96813238e+00 4.82554525e-01 -2.24760354e-01 4.32722241e-01 3.98247279e-02 9.73276973e-01 1.63780242e-01 -4.62256312e-01 3.39427888e-01 -4.57183272e-02 -2.55519986e-01 2.05274642e-01 1.90719530e-01 -1.35431576e+00 9.05988514e-01 5.52638054e+00 1.00500429e+00 -8.06826949e-01 4.00715530e-01 1.24462679e-01 -2.29052752e-01 1.58574983e-01 -3.07507068e-01 -9.32979345e-01 7.16660857e-01 7.24637136e-02 -1.96863279e-01 5.48344612e-01 1.08443677e+00 -1.06723316e-01 -1.34379223e-01 -1.10272932e+00 1.55216372e+00 3.57971311e-01 -1.28048623e+00 1.24826066e-01 -9.54689737e-03 7.25954890e-01 -4.34061624e-02 1.44464150e-01 5.27542830e-01 8.56382996e-02 -8.87186944e-01 1.07203197e+00 2.81356364e-01 6.87792003e-01 -5.68463445e-01 6.72650099e-01 3.22389811e-01 -1.86384904e+00 -5.94317853e-01 -3.00294518e-01 8.65209028e-02 4.92898896e-02 2.02388704e-01 -5.34200728e-01 2.91502059e-01 1.07083774e+00 8.49976003e-01 -7.26173401e-01 1.20678616e+00 -3.86779875e-01 1.73415974e-01 -1.55059174e-01 -1.17532879e-01 8.97547677e-02 4.83426973e-02 5.92275202e-01 1.28787518e+00 1.85367480e-01 -1.05149522e-02 7.30143011e-01 9.63246465e-01 -1.82430018e-02 -8.74224771e-03 -6.98917985e-01 5.00220619e-02 6.26172721e-01 1.28225768e+00 -6.03701711e-01 -5.62185645e-01 -8.17129195e-01 1.24291265e+00 5.00100672e-01 4.29441661e-01 -1.11048996e+00 -5.41790724e-01 5.85278928e-01 9.48926359e-02 6.18243337e-01 -4.48680706e-02 2.13219412e-02 -1.41121769e+00 1.23486973e-01 -9.27102566e-01 6.64813519e-01 -8.80823195e-01 -1.40103257e+00 3.45580012e-01 2.16881633e-01 -1.65631461e+00 4.07404602e-02 -5.41721463e-01 -5.54903507e-01 4.89026815e-01 -1.18319643e+00 -1.51699722e+00 -8.67602050e-01 7.99989522e-01 8.86057436e-01 -1.83993295e-01 3.88878465e-01 3.67340982e-01 -6.20105684e-01 6.18295193e-01 -2.01155290e-01 1.78510517e-01 4.91718024e-01 -1.01262569e+00 9.96999368e-02 7.59234071e-01 2.08132625e-01 5.28117537e-01 4.73584294e-01 -5.29159486e-01 -1.33823645e+00 -1.25720561e+00 6.07433498e-01 -4.44786072e-01 5.21898329e-01 -6.10341966e-01 -8.68482649e-01 6.52085900e-01 5.58485985e-02 5.53475320e-01 5.14334440e-01 -1.78800181e-01 -4.56709117e-01 -9.70156789e-02 -7.67525494e-01 7.37563670e-01 1.50536394e+00 -6.94078088e-01 -5.93235493e-01 4.02033657e-01 6.41274989e-01 -1.47943601e-01 -5.01019835e-01 4.13839817e-01 6.47090614e-01 -1.13986111e+00 1.17351103e+00 -2.91672856e-01 2.06742942e-01 -6.70107305e-01 -1.71917230e-01 -9.20804381e-01 -7.36214280e-01 -4.93095636e-01 -5.72453499e-01 1.13258362e+00 -1.08602196e-01 -2.51125932e-01 3.83349150e-01 9.79804993e-02 2.90360767e-02 -6.94239736e-01 -6.09847903e-01 -9.59329545e-01 -4.10431027e-01 -2.41917089e-01 3.79845440e-01 7.16991484e-01 -4.70196567e-02 5.76194465e-01 -6.73733532e-01 7.06523880e-02 9.60526705e-01 3.83443087e-01 1.09272933e+00 -1.06346726e+00 -4.51748490e-01 -4.08613443e-01 -5.43956697e-01 -1.03744042e+00 1.78086560e-03 -9.95952666e-01 2.00402945e-01 -1.35904074e+00 8.12858999e-01 -1.05606243e-01 -1.89109340e-01 6.71678782e-01 -4.42620188e-01 4.52806860e-01 5.73238611e-01 5.64867496e-01 -8.97154212e-01 7.10306823e-01 1.43469441e+00 -2.45829180e-01 -2.48274758e-01 -2.57039130e-01 -4.28702414e-01 8.10853362e-01 4.80562747e-01 -3.38338763e-01 -4.94871020e-01 -1.89791635e-01 -2.75598615e-01 -6.10876754e-02 7.61285484e-01 -1.14741409e+00 1.08293071e-01 -2.39557073e-01 5.59380472e-01 -9.86255765e-01 8.24538320e-02 -7.39452839e-01 1.95149422e-01 5.72231770e-01 -1.81168929e-01 -4.22068894e-01 -1.54803872e-01 6.65221155e-01 -2.09064707e-01 -8.00010711e-02 8.87420118e-01 -2.85833567e-01 -1.09450829e+00 4.98816997e-01 -3.39873940e-01 9.71807614e-02 1.34919560e+00 -9.32685584e-02 -3.07374209e-01 -2.23304555e-01 -5.02518952e-01 4.28512901e-01 2.39098921e-01 7.13609874e-01 7.01140523e-01 -1.72242033e+00 -4.40978169e-01 3.83386910e-01 6.29167020e-01 1.55754583e-02 5.11407614e-01 8.37217450e-01 -3.06120396e-01 2.79112458e-01 -3.63845587e-01 -7.71807075e-01 -1.56271648e+00 9.91405368e-01 9.52889323e-02 -4.56637815e-02 -9.39140439e-01 9.67404246e-01 9.54314709e-01 -2.50740439e-01 5.17028511e-01 -5.21625839e-02 -1.23993710e-01 1.56877741e-01 6.91145062e-01 3.92800778e-01 -5.69633543e-01 -8.78324926e-01 -4.58646387e-01 4.16539699e-01 -4.17258777e-02 1.14315644e-01 8.76251340e-01 -2.30205700e-01 1.50469467e-02 5.64953446e-01 1.52072036e+00 -2.94546634e-01 -1.25334036e+00 -4.17273194e-01 -3.17042857e-01 -7.64556289e-01 -1.89712748e-01 -5.81896722e-01 -1.01177454e+00 1.07054949e+00 6.49454296e-01 5.94544709e-02 9.96702492e-01 4.68193889e-01 8.72964561e-01 3.05081397e-01 3.02377939e-01 -8.38380516e-01 6.89288914e-01 3.17231655e-01 1.18891931e+00 -1.29514718e+00 -1.47007257e-02 -7.20243931e-01 -9.63148654e-01 7.91092515e-01 9.27308321e-01 -4.79083695e-02 5.43926418e-01 6.16322681e-02 -1.95103660e-01 -4.35421258e-01 -5.34278810e-01 -7.24680007e-01 4.76985812e-01 7.88140059e-01 2.07010403e-01 2.46995911e-01 -2.95021117e-01 6.28142893e-01 2.18515918e-01 -1.37319818e-01 5.54756224e-02 8.07290137e-01 -3.77246618e-01 -7.42595613e-01 -3.64988089e-01 2.82407463e-01 9.42774862e-02 1.08187236e-01 -2.97511935e-01 9.51171160e-01 5.84253728e-01 6.98846579e-01 4.59684841e-02 -5.03134787e-01 2.59605676e-01 -2.00410023e-01 4.17819768e-01 -6.92934990e-01 -3.15935284e-01 2.84612507e-01 -1.08736746e-01 -8.38854432e-01 -2.52267390e-01 -6.24146044e-01 -1.13833189e+00 9.98108238e-02 -4.42341238e-01 -3.48837137e-01 1.37945637e-01 6.65653527e-01 3.24217647e-01 5.57454944e-01 5.65159142e-01 -1.29507935e+00 -3.78194779e-01 -7.45272636e-01 -6.66937113e-01 9.19831157e-01 3.96148801e-01 -1.18860900e+00 -3.86908263e-01 7.11198077e-02]
[9.568751335144043, 1.3989900350570679]
22d9ff68-1e7d-4060-a465-dc41b9aaf9d2
self-adjusting-weighted-expected-improvement
2306.04262
null
https://arxiv.org/abs/2306.04262v3
https://arxiv.org/pdf/2306.04262v3.pdf
Self-Adjusting Weighted Expected Improvement for Bayesian Optimization
Bayesian Optimization (BO) is a class of surrogate-based, sample-efficient algorithms for optimizing black-box problems with small evaluation budgets. The BO pipeline itself is highly configurable with many different design choices regarding the initial design, surrogate model, and acquisition function (AF). Unfortunately, our understanding of how to select suitable components for a problem at hand is very limited. In this work, we focus on the definition of the AF, whose main purpose is to balance the trade-off between exploring regions with high uncertainty and those with high promise for good solutions. We propose Self-Adjusting Weighted Expected Improvement (SAWEI), where we let the exploration-exploitation trade-off self-adjust in a data-driven manner, based on a convergence criterion for BO. On the noise-free black-box BBOB functions of the COCO benchmarking platform, our method exhibits a favorable any-time performance compared to handcrafted baselines and serves as a robust default choice for any problem structure. The suitability of our method also transfers to HPOBench. With SAWEI, we are a step closer to on-the-fly, data-driven, and robust BO designs that automatically adjust their sampling behavior to the problem at hand.
['Marius Lindauer', 'Carola Doerr', 'Anja Jankovic', 'Elena Raponi', 'Carolin Benjamins']
2023-06-07
null
null
null
null
['bayesian-optimization']
['methodology']
[-1.55081786e-02 -2.68737078e-01 -2.25936696e-01 -2.98880219e-01 -1.14770174e+00 -7.31751204e-01 4.77731228e-01 3.07890214e-02 -4.94201481e-01 7.00443983e-01 1.71509460e-01 -5.31927109e-01 -6.92747235e-01 -5.72627127e-01 -5.33290505e-01 -8.12448919e-01 -9.56269354e-02 6.79216862e-01 5.58533520e-02 -3.81671071e-01 5.32261074e-01 4.46839541e-01 -1.45357037e+00 -1.17364168e-01 9.37809408e-01 1.10653472e+00 4.72379625e-02 7.01819479e-01 4.05024320e-01 5.23451455e-02 -4.19633359e-01 -1.08413912e-01 5.73645771e-01 -3.83279473e-01 -7.68749595e-01 -2.82227367e-01 8.78904462e-02 4.33620438e-02 3.95523787e-01 9.01490569e-01 9.46836174e-01 3.00101161e-01 4.41412777e-01 -1.08215547e+00 1.07980192e-01 6.17110789e-01 -3.99544716e-01 2.87273258e-01 1.25758767e-01 8.06806445e-01 1.18143463e+00 -7.80385137e-01 5.41891992e-01 1.14754605e+00 5.11117339e-01 3.29498708e-01 -1.70220554e+00 -4.43514019e-01 2.44823009e-01 -1.27472639e-01 -1.47486126e+00 -7.08884597e-01 5.18513262e-01 -4.54228491e-01 7.85697222e-01 4.46728766e-01 6.22535467e-01 8.11343610e-01 2.98142016e-01 5.54040194e-01 1.02874660e+00 -4.23093528e-01 8.41630459e-01 2.54742324e-01 -4.80755083e-02 3.93458664e-01 2.16818452e-01 3.56522262e-01 -6.23140931e-01 -3.28582048e-01 3.77402693e-01 -6.69588566e-01 -2.92128682e-01 -9.09454823e-01 -1.08838844e+00 8.09173167e-01 3.80789101e-01 -1.08687200e-01 -2.60940492e-01 2.79968709e-01 1.65760905e-01 2.52306849e-01 1.18161023e-01 1.24552894e+00 -6.82339728e-01 -4.85962600e-01 -1.15378439e+00 7.89144099e-01 7.39434242e-01 6.97462499e-01 6.51898563e-01 -1.11555986e-01 -5.45247436e-01 7.89107800e-01 4.29796278e-01 1.55664366e-02 4.46355134e-01 -1.04036176e+00 4.65421349e-01 4.69875246e-01 5.43783724e-01 -6.97422981e-01 -5.21730363e-01 -9.02702034e-01 -1.55199915e-01 6.41374350e-01 6.80019438e-01 -4.13557231e-01 -7.49176979e-01 1.66913366e+00 5.99967062e-01 -5.52856922e-01 -1.69379056e-01 1.10765803e+00 2.52629995e-01 5.58851063e-01 -2.60492444e-01 -1.64113835e-01 1.19387972e+00 -9.61776257e-01 -3.04805815e-01 -5.81969023e-01 5.13117790e-01 -5.61984658e-01 1.42875588e+00 8.09640288e-01 -1.19262624e+00 -2.57489562e-01 -1.32056665e+00 4.06423718e-01 -1.19157314e-01 -5.89850098e-02 5.92969537e-01 9.91817832e-01 -7.86882818e-01 8.15923393e-01 -9.08286631e-01 -4.20095511e-02 2.77524263e-01 5.50560653e-01 1.84368119e-01 2.27185741e-01 -8.52436721e-01 9.94008839e-01 5.19353330e-01 1.85161427e-01 -1.07743657e+00 -9.03420210e-01 -5.73599637e-01 4.24452238e-02 8.46605182e-01 -7.76044130e-01 1.28671718e+00 -8.33143353e-01 -1.79781556e+00 4.58425343e-01 2.45910347e-01 -5.24612606e-01 7.56008923e-01 -1.46140233e-01 -6.50061145e-02 -4.21539396e-01 -1.43016532e-01 7.47459829e-01 8.93258989e-01 -1.13085485e+00 -5.29546142e-01 -1.49712920e-01 -1.62816942e-02 3.21571797e-01 -1.15866132e-01 1.03776954e-01 -5.43815196e-01 -5.30274630e-01 -1.24314204e-01 -9.70950782e-01 -7.06180215e-01 -1.69012085e-01 -2.60950238e-01 1.86010018e-01 2.85627782e-01 -2.78479904e-01 1.73662388e+00 -1.88309705e+00 2.03121811e-01 4.64783937e-01 -1.54505938e-01 -6.55779168e-02 -4.84705605e-02 3.76644433e-01 7.75490925e-02 1.34518772e-01 -3.80858034e-01 -6.02454469e-02 1.69214129e-01 -1.37239531e-01 -1.29442424e-01 6.43681765e-01 3.59495014e-01 5.63988805e-01 -9.99065280e-01 -3.47763181e-01 1.14623681e-01 6.20721318e-02 -1.18835318e+00 3.77049536e-01 -5.27489185e-01 6.29058659e-01 -7.04013884e-01 7.14918673e-01 3.77760738e-01 -1.49166644e-01 1.12529077e-01 -2.75380332e-02 -4.12858039e-01 2.79780209e-01 -1.78400028e+00 1.64896297e+00 -6.68709099e-01 2.52593488e-01 3.41800630e-01 -6.35411620e-01 1.07036483e+00 -2.41846487e-01 3.46274018e-01 -4.11444545e-01 1.07279740e-01 3.08680713e-01 3.52586508e-01 -2.23510802e-01 5.07463396e-01 1.32904813e-01 -5.64904548e-02 4.02453572e-01 -1.88229412e-01 -4.32249576e-01 5.16913891e-01 -3.18220764e-01 1.20998311e+00 6.37503922e-01 4.12260324e-01 -7.92132616e-01 2.05514804e-01 1.73041329e-01 7.23046601e-01 7.70591319e-01 8.38720705e-03 9.12130773e-01 6.88571632e-01 -3.80437285e-01 -9.33191299e-01 -7.55823612e-01 -5.16591191e-01 1.06930292e+00 7.79247805e-02 -4.32065427e-01 -7.08771110e-01 -3.42470378e-01 3.18651870e-02 1.06430531e+00 -4.78093743e-01 -9.70575884e-02 -5.41628838e-01 -1.14536321e+00 9.12366286e-02 2.12071240e-01 1.03546344e-01 -9.79988754e-01 -1.19375205e+00 5.16108096e-01 2.90126592e-01 -4.63480324e-01 -5.34149885e-01 6.33300304e-01 -8.14107716e-01 -8.65779698e-01 -5.47618985e-01 -5.09259477e-02 6.42622113e-01 -4.06987876e-01 1.36554193e+00 -2.75417984e-01 -2.69470423e-01 2.21154764e-02 -1.63002044e-01 -4.16452646e-01 -3.20332080e-01 1.43211961e-01 -1.57979265e-01 -7.99678564e-02 -1.97373196e-01 -3.92137319e-01 -8.82592678e-01 8.05201411e-01 -6.80581152e-01 -1.32919461e-01 5.03846169e-01 1.02187371e+00 7.12886751e-01 6.28172532e-02 2.15650246e-01 -6.33041680e-01 9.11226809e-01 -5.10568857e-01 -1.18781948e+00 3.06655526e-01 -1.15462744e+00 5.67090929e-01 4.78084296e-01 -4.26340789e-01 -7.96120524e-01 4.59549874e-02 5.00917062e-03 -2.81864107e-01 1.60816520e-01 5.32031775e-01 -3.96181315e-01 -4.48612384e-02 9.29353118e-01 -4.41932589e-01 -1.14317231e-01 -5.64535201e-01 3.80201161e-01 4.20578480e-01 2.09437743e-01 -1.19734836e+00 4.05213565e-01 6.04636408e-02 -6.77542239e-02 -2.64078736e-01 -6.75342023e-01 -2.24666834e-01 -2.74599582e-01 -1.68595418e-01 4.40178603e-01 -5.48780680e-01 -7.14276671e-01 1.19050331e-01 -6.45331919e-01 -7.71118820e-01 -4.82101440e-01 1.23212405e-01 -6.17641270e-01 -1.30484626e-01 1.92323714e-01 -9.14248645e-01 -3.92359078e-01 -1.69176984e+00 1.06130123e+00 1.52352318e-01 -5.00824153e-01 -7.57382452e-01 2.79306948e-01 3.37791473e-01 5.81370115e-01 5.46232700e-01 6.44899905e-01 -4.45912868e-01 -5.97516418e-01 -1.34871319e-01 1.96655244e-01 3.18312138e-01 -2.80108422e-01 3.11375111e-01 -7.04907954e-01 -5.84833324e-01 -6.13294654e-02 -2.42751926e-01 6.05145931e-01 5.91995597e-01 1.22431660e+00 -3.83057386e-01 -2.12475106e-01 1.01097202e+00 1.43955433e+00 5.25364317e-02 3.06309849e-01 9.34250176e-01 1.42076343e-01 6.18312716e-01 9.50490177e-01 7.63300896e-01 6.04953505e-02 1.17583191e+00 6.75831378e-01 1.67887628e-01 2.39069581e-01 -5.30957095e-02 3.75825971e-01 1.41879469e-01 2.35161945e-01 -3.26050103e-01 -1.17765474e+00 5.39347649e-01 -1.98083603e+00 -4.67679262e-01 1.48249224e-01 2.57409906e+00 1.06973422e+00 6.25314415e-01 4.14232194e-01 -1.08215876e-01 5.44093668e-01 1.18407771e-01 -7.59354770e-01 -5.44983268e-01 1.89984351e-01 1.36420876e-01 7.25617647e-01 3.54046822e-01 -8.48923564e-01 4.61673707e-01 6.43508196e+00 9.88937020e-01 -1.00978673e+00 -1.13912433e-01 8.44778121e-01 -5.89923084e-01 -3.15113276e-01 3.74494433e-01 -1.00103164e+00 6.67679369e-01 1.18633664e+00 -2.22833738e-01 8.95040035e-01 9.26561356e-01 6.07876718e-01 -2.50939280e-01 -1.30387211e+00 6.43208742e-01 -4.16700989e-01 -1.36998415e+00 -5.86046815e-01 6.67251050e-02 7.26154566e-01 -5.43519203e-03 1.98744927e-02 1.22404777e-01 5.42884886e-01 -1.15714216e+00 1.28830349e+00 3.45327675e-01 4.94392067e-01 -8.04926932e-01 5.55147469e-01 2.00275123e-01 -7.06004918e-01 -3.44657511e-01 5.81080168e-02 1.31161913e-01 2.76180625e-01 7.22033441e-01 -7.51231670e-01 2.79108346e-01 9.32604313e-01 1.19590104e-01 -5.28790236e-01 1.34293807e+00 -2.21762910e-01 6.78278446e-01 -5.57626188e-01 -2.10446239e-01 2.91738868e-01 -4.08213228e-01 8.18032384e-01 1.05257964e+00 2.36124307e-01 -3.70816588e-01 4.51686829e-02 1.19850433e+00 3.80279541e-01 1.10072561e-01 -4.59611192e-02 1.20844863e-01 6.13153100e-01 1.18313897e+00 -6.98986769e-01 1.96788371e-01 2.19227463e-01 3.88616621e-02 1.42934501e-01 2.12193057e-01 -8.19163680e-01 -4.11725104e-01 6.31596029e-01 2.17845544e-01 5.36138773e-01 -1.68624625e-01 -7.67383754e-01 -7.09071398e-01 -3.29359598e-03 -1.33997238e+00 5.36350071e-01 -5.85914731e-01 -1.00012088e+00 3.74833196e-01 3.46993446e-01 -1.23810065e+00 -2.53899753e-01 -5.65245390e-01 -6.76198781e-01 9.03694689e-01 -1.04955006e+00 -5.78782916e-01 -1.19244441e-01 -6.64666714e-03 4.61886227e-01 -5.78967333e-02 4.43836153e-01 2.90757064e-02 -7.89151490e-01 5.19289076e-01 3.90157759e-01 -5.54775774e-01 6.25774503e-01 -1.28478682e+00 3.61221075e-01 8.23087275e-01 -1.12642810e-01 7.68670917e-01 1.32517421e+00 -4.71208692e-01 -1.72653794e+00 -6.13133550e-01 7.10630938e-02 -5.92362404e-01 8.38120401e-01 -3.90621483e-01 -5.37260056e-01 8.57643485e-02 -4.92462739e-02 -6.09059371e-02 3.26893926e-01 4.20995355e-01 1.65607750e-01 -4.17373240e-01 -1.17943287e+00 7.60083199e-01 7.52209067e-01 2.23942146e-01 -2.88457304e-01 3.01729172e-01 5.07107913e-01 -6.99383497e-01 -9.07443643e-01 4.78277832e-01 5.43370962e-01 -1.04866350e+00 9.38714087e-01 -4.15195614e-01 2.62795419e-01 -5.67820609e-01 -1.55350706e-02 -1.54440010e+00 -2.20845088e-01 -1.25882697e+00 2.63000582e-03 1.25066388e+00 6.86872244e-01 -6.36248946e-01 8.32104683e-01 8.52272570e-01 -9.58635435e-02 -1.20728803e+00 -8.73408854e-01 -7.94586658e-01 -2.10324749e-02 -4.13340896e-01 9.44405973e-01 3.12695801e-01 -2.58931369e-01 1.78470612e-01 -2.10201919e-01 3.26287635e-02 5.60002327e-01 1.76262960e-01 1.05820394e+00 -7.45881081e-01 -8.71504486e-01 -7.75690079e-01 9.65348482e-02 -9.41491723e-01 -5.42676866e-01 -4.15414691e-01 2.98369050e-01 -1.00551033e+00 -1.36624202e-01 -8.01753581e-01 -3.65905583e-01 3.22625816e-01 -2.95592904e-01 -2.95662671e-01 1.34162813e-01 5.49268164e-02 -4.79500353e-01 6.62466586e-01 9.78376269e-01 -4.68613505e-02 -6.68778360e-01 2.65777886e-01 -8.79614353e-01 5.01495600e-01 6.62847757e-01 -5.94654322e-01 -3.31479698e-01 -3.58213156e-01 4.97855633e-01 -4.04310189e-02 6.11279868e-02 -8.18860114e-01 8.91903043e-02 -5.91293633e-01 -1.05069317e-01 -3.71947020e-01 1.00698680e-01 -6.67157054e-01 5.34762800e-01 3.57267350e-01 -5.29057980e-01 2.36077964e-01 2.21508011e-01 4.09089655e-01 9.68345776e-02 -6.83443248e-01 8.90076458e-01 -1.54004982e-02 -4.49154973e-01 2.86213681e-02 -1.52484059e-01 4.46357280e-01 8.69825840e-01 -3.48876774e-01 -7.09188208e-02 -1.95558250e-01 -3.89052361e-01 7.23149180e-01 6.69545591e-01 2.77348757e-01 5.61813898e-02 -1.00001097e+00 -6.63049102e-01 1.18693456e-01 2.07975164e-01 3.12746972e-01 -1.52582273e-01 7.28894949e-01 -5.55145442e-01 2.08772704e-01 -2.92575341e-02 -5.57343721e-01 -8.40306997e-01 3.67323279e-01 4.97157246e-01 -5.24667323e-01 -2.90761888e-01 9.64611709e-01 -3.27411532e-01 -2.93503225e-01 4.59706783e-01 -3.32755983e-01 1.25447139e-01 2.80199125e-02 2.39631847e-01 6.43645585e-01 4.49918032e-01 4.31178808e-02 -5.26265085e-01 2.17838719e-01 2.62760043e-01 -3.40359628e-01 1.58501732e+00 1.72071025e-01 9.69882533e-02 2.96503127e-01 7.87628651e-01 -4.53695282e-02 -1.74718237e+00 2.69609600e-01 2.41974398e-01 -7.81198561e-01 5.86442351e-01 -9.47529435e-01 -8.29095840e-01 3.77035439e-01 6.54184759e-01 1.25891626e-01 1.01494181e+00 -3.75292301e-01 2.43371904e-01 2.24301264e-01 3.90086472e-01 -1.46366727e+00 -3.99805717e-02 8.07601362e-02 1.11777258e+00 -1.12607419e+00 5.05053759e-01 1.25797823e-01 -6.60902560e-01 1.00272655e+00 6.12439096e-01 -2.85894424e-02 4.35544997e-01 3.75467479e-01 -1.25176251e-01 -1.69660017e-01 -1.10003901e+00 1.00657314e-01 4.27254796e-01 2.65372276e-01 1.44396394e-01 -1.34869546e-01 -2.87463605e-01 3.72823954e-01 -3.73492420e-01 -2.25331560e-01 1.98729470e-01 1.05024993e+00 -4.41341102e-01 -1.25509655e+00 -7.86311567e-01 4.96820807e-01 -2.48355672e-01 -3.67182866e-02 -2.63183676e-02 7.04811454e-01 -2.90186908e-02 9.32040215e-01 -2.18002677e-01 -6.13641478e-02 5.21762311e-01 -2.39921302e-01 3.52261186e-01 -5.59328854e-01 -9.22846079e-01 1.70937106e-01 4.96004969e-01 -7.70034850e-01 6.14916570e-02 -9.21862483e-01 -7.02368855e-01 -8.71001035e-02 -6.17881477e-01 1.78950325e-01 8.81836593e-01 8.16027045e-01 4.77161825e-01 5.36074638e-01 6.56916320e-01 -1.01887059e+00 -1.15445101e+00 -5.31182885e-01 -1.55512810e-01 -8.91300812e-02 1.58052698e-01 -8.35093856e-01 -3.49304706e-01 -4.26864028e-01]
[6.198384761810303, 3.8136327266693115]
f034de99-5f8f-4e8c-af54-e003a0a8f227
recognizing-emotion-cause-in-conversations-1
2012.1182
null
https://arxiv.org/abs/2012.11820v4
https://arxiv.org/pdf/2012.11820v4.pdf
Recognizing Emotion Cause in Conversations
We address the problem of recognizing emotion cause in conversations, define two novel sub-tasks of this problem, and provide a corresponding dialogue-level dataset, along with strong Transformer-based baselines. The dataset is available at https://github.com/declare-lab/RECCON. Introduction: Recognizing the cause behind emotions in text is a fundamental yet under-explored area of research in NLP. Advances in this area hold the potential to improve interpretability and performance in affect-based models. Identifying emotion causes at the utterance level in conversations is particularly challenging due to the intermingling dynamics among the interlocutors. Method: We introduce the task of Recognizing Emotion Cause in CONversations with an accompanying dataset named RECCON, containing over 1,000 dialogues and 10,000 utterance cause-effect pairs. Furthermore, we define different cause types based on the source of the causes, and establish strong Transformer-based baselines to address two different sub-tasks on this dataset: causal span extraction and causal emotion entailment. Result: Our Transformer-based baselines, which leverage contextual pre-trained embeddings, such as RoBERTa, outperform the state-of-the-art emotion cause extraction approaches Conclusion: We introduce a new task highly relevant for (explainable) emotion-aware artificial intelligence: recognizing emotion cause in conversations, provide a new highly challenging publicly available dialogue-level dataset for this task, and give strong baseline results on this dataset.
['Rada Mihalcea', 'Alexander Gelbukh', 'Niyati Chhaya', 'Abhinaba Roy', 'Romila Ghosh', 'Pengfei Hong', 'Samson Yu Bai Jian', 'Rishabh Bhardwaj', 'Deepanway Ghosal', 'Devamanyu Hazarika', 'Navonil Majumder', 'Soujanya Poria']
2020-12-22
recognizing-emotion-cause-in-conversations
https://arxiv.org/abs/2012.11820
https://arxiv.org/pdf/2012.11820.pdf
null
['causal-emotion-entailment', 'emotion-cause-extraction', 'recognizing-emotion-cause-in-conversations']
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
[ 2.28475481e-01 4.03064042e-01 -2.01584384e-01 -6.77204847e-01 -8.53437662e-01 -9.23517168e-01 9.92259681e-01 7.87689239e-02 2.01256797e-01 7.97213376e-01 1.25126100e+00 -9.07086805e-02 -4.61610453e-03 -3.34997296e-01 -4.94685918e-01 -3.56626928e-01 -2.03731164e-01 5.14439881e-01 -7.95026898e-01 -6.57760799e-01 4.79975753e-02 -9.76725714e-04 -1.31790113e+00 9.29271340e-01 5.80325663e-01 9.25344050e-01 -4.90199924e-01 9.31487203e-01 -4.07126307e-01 1.34889162e+00 -7.05994248e-01 -6.44908607e-01 -3.62602055e-01 -6.87444746e-01 -1.34511256e+00 3.35238650e-02 6.56981468e-02 -4.04823944e-02 -1.88675269e-01 2.39243358e-01 6.71771109e-01 1.82170257e-01 7.48421669e-01 -1.62972331e+00 -5.10402799e-01 1.07736087e+00 -1.12930633e-01 1.99393511e-01 1.04278922e+00 -3.49937454e-02 1.70203769e+00 -7.56662786e-01 6.08721018e-01 1.86738491e+00 6.68523192e-01 9.00713205e-01 -1.10218418e+00 -4.21578169e-01 2.84363538e-01 4.67306733e-01 -5.93252480e-01 -5.94202220e-01 9.49569225e-01 -3.50797892e-01 1.44132447e+00 5.33840179e-01 5.24329603e-01 1.81853008e+00 5.38961068e-02 1.28764677e+00 1.08185911e+00 -2.30513781e-01 4.19582147e-03 -1.16699815e-01 4.47552085e-01 5.10146677e-01 -8.84364426e-01 -1.85718372e-01 -9.69106257e-01 -3.62309575e-01 2.40401193e-01 -4.99136716e-01 -5.22071958e-01 3.38646621e-01 -1.29211700e+00 1.05250514e+00 1.57905817e-01 3.84165615e-01 -5.58063149e-01 2.31458291e-01 8.65202546e-01 4.27999973e-01 9.22431171e-01 7.48566628e-01 -1.00346947e+00 -9.85346794e-01 -1.75937220e-01 5.54877818e-01 1.35986269e+00 6.25115991e-01 3.69675905e-01 -1.10971406e-01 -2.62313694e-01 1.06758165e+00 -5.10018691e-02 4.58851643e-02 3.25058967e-01 -8.59957397e-01 5.05337238e-01 6.60313010e-01 -9.41436216e-02 -9.29771423e-01 -6.58363283e-01 -9.42910183e-03 -6.11325741e-01 -6.04683280e-01 5.00884056e-01 -7.32947648e-01 -3.45377654e-01 2.14307928e+00 4.51606214e-01 1.50111601e-01 4.04696554e-01 7.77380586e-01 1.32494998e+00 8.89903843e-01 -6.04438782e-02 -4.64056641e-01 1.59147000e+00 -9.10977006e-01 -1.42348444e+00 -1.98939592e-01 7.29613423e-01 -8.66584182e-01 1.27466464e+00 5.11644483e-01 -8.86533380e-01 -8.90004784e-02 -5.75002909e-01 -3.91916543e-01 -4.44681913e-01 -1.25167474e-01 1.09181821e+00 2.67839223e-01 -7.91724145e-01 2.61195630e-01 2.80749402e-03 -3.88764858e-01 -2.78236773e-02 1.16857290e-01 -2.52047926e-01 1.20624274e-01 -1.89264309e+00 1.03179967e+00 -1.46757931e-01 1.15672506e-01 -9.02325571e-01 -1.20473838e+00 -1.03038573e+00 -1.78084806e-01 2.39765748e-01 -4.08180654e-01 1.57582641e+00 -1.06144500e+00 -1.60495067e+00 8.85654271e-01 -6.26008868e-01 -2.97743469e-01 5.79327457e-02 -6.05630457e-01 -3.71828496e-01 -2.06141010e-01 3.61807235e-02 5.29518723e-01 6.16281152e-01 -1.21375501e+00 -4.26459879e-01 -8.91173705e-02 2.33899236e-01 4.08181429e-01 -3.82462412e-01 4.91641700e-01 -6.15097694e-02 -4.27097529e-01 -6.30089939e-01 -7.44705975e-01 3.88528928e-02 -7.80780017e-01 -7.88269460e-01 -9.85773206e-01 9.10033584e-01 -6.19005322e-01 1.27008176e+00 -2.02111578e+00 4.53472495e-01 -4.77445126e-01 3.87375057e-01 -3.29020500e-01 -3.62879127e-01 8.01433384e-01 -5.52557409e-01 3.57658863e-01 -1.32474110e-01 -6.35041654e-01 3.71891618e-01 4.00273532e-01 -8.76169622e-01 8.07350799e-02 5.74113131e-01 9.76398230e-01 -1.01317739e+00 -2.18664467e-01 3.58229131e-01 3.83804232e-01 -4.56230700e-01 7.68419385e-01 -4.46030557e-01 3.74336064e-01 -2.60722011e-01 1.04034498e-01 1.15687430e-01 2.07823813e-01 1.93237558e-01 -4.00456011e-01 2.99811997e-02 1.04390192e+00 -6.94423735e-01 1.46174622e+00 -9.34890389e-01 9.82313693e-01 1.57251552e-01 -7.93778777e-01 8.56373608e-01 1.04015994e+00 5.61319947e-01 -1.63348183e-01 3.22456926e-01 -1.54198185e-01 1.44541487e-01 -8.72155249e-01 4.56749827e-01 -5.55307269e-01 -7.05943704e-01 7.14699626e-01 1.65664956e-01 -5.93597651e-01 -7.22472556e-03 5.46076775e-01 1.51203322e+00 -2.07954332e-01 2.97919542e-01 1.79291815e-02 3.50344241e-01 -2.78843213e-02 6.81280315e-01 2.18711108e-01 -5.52548468e-01 3.56746674e-01 1.26144445e+00 -7.27197051e-01 -4.36949015e-01 -7.25343585e-01 -1.29122481e-01 1.33810580e+00 -2.55934298e-01 -6.52374029e-01 -5.03205776e-01 -8.63890767e-01 5.16202711e-02 1.04390514e+00 -1.17864788e+00 1.41057730e-01 -5.87677240e-01 -7.83911049e-01 8.56897473e-01 2.20806554e-01 2.48237327e-01 -1.27031493e+00 1.66440997e-02 3.96183610e-01 -1.07851398e+00 -1.40887356e+00 -2.59292513e-01 3.44862819e-01 -1.85478151e-01 -1.20485675e+00 -1.11131288e-01 -2.43172675e-01 -1.41741678e-01 -2.53157377e-01 1.68656397e+00 -2.74472445e-01 -2.52486736e-01 7.62162209e-01 -5.22574186e-01 -7.40863204e-01 -5.42290509e-01 -7.53644854e-02 1.24891400e-01 2.58046150e-01 5.69629252e-01 -4.51384157e-01 -9.58755240e-02 7.96051174e-02 -4.14590538e-01 -1.75658595e-02 -2.89071947e-02 7.99892902e-01 -5.38479760e-02 -2.56709814e-01 9.98867631e-01 -7.38710105e-01 1.18438554e+00 -8.95477295e-01 4.31823105e-01 -6.37910441e-02 -1.48543036e-02 -8.98063034e-02 6.85409248e-01 -1.93731144e-01 -1.40747714e+00 -2.66569346e-01 -2.67335653e-01 -9.91635472e-02 -4.73523051e-01 3.94000024e-01 -4.57468778e-01 8.72108161e-01 4.80439901e-01 -3.02495927e-01 1.48085030e-02 -2.06449091e-01 9.50060904e-01 8.11995924e-01 3.55018944e-01 -7.41954744e-01 2.22972572e-01 1.51451349e-01 -3.09513688e-01 -7.50078559e-01 -1.33301175e+00 -5.77728152e-01 -3.82509083e-01 -5.23078024e-01 1.00356781e+00 -9.67073917e-01 -8.65936697e-01 1.51461244e-01 -1.84376490e+00 -5.96285582e-01 -1.97114378e-01 2.48200029e-01 -6.61440074e-01 -1.69913083e-01 -9.50680912e-01 -8.68927717e-01 -3.57826918e-01 -7.26455152e-01 1.13079691e+00 -9.87930298e-02 -1.19446158e+00 -1.43282759e+00 1.56383604e-01 6.75358474e-01 1.42072290e-01 5.56061327e-01 1.02758825e+00 -9.40623164e-01 1.31059811e-01 1.30622506e-01 -7.00119659e-02 4.06962819e-02 5.37513137e-01 2.50569344e-01 -1.42131710e+00 5.39798677e-01 3.86509784e-02 -9.21323001e-01 6.91873670e-01 2.77509302e-01 8.09137464e-01 -5.78469098e-01 -8.65309611e-02 -1.25944272e-01 5.68310618e-01 -1.37138888e-01 3.94541442e-01 -1.39127076e-01 7.44661272e-01 1.32995760e+00 6.43354118e-01 7.06257880e-01 8.48812938e-01 6.72261834e-01 4.79624093e-01 -3.35130513e-01 -2.01305225e-01 -4.96833920e-02 5.82841456e-01 1.08023882e+00 1.23131871e-01 -4.19298232e-01 -7.03750968e-01 9.95019317e-01 -2.03925300e+00 -1.22612572e+00 -7.48334229e-01 1.17647648e+00 1.46843290e+00 -4.38216895e-01 1.96752489e-01 1.78593010e-01 5.10259092e-01 5.99527538e-01 -2.09378362e-01 -1.27426147e+00 -1.50122702e-01 2.19164938e-02 -5.75703323e-01 8.90698433e-01 -1.21966696e+00 1.07469726e+00 5.65670395e+00 8.29005063e-01 -7.47870088e-01 2.08419621e-01 8.25493574e-01 -2.46619046e-01 -6.69185340e-01 -2.91621208e-01 -5.81562400e-01 1.87869892e-01 1.02606261e+00 -1.88625604e-01 4.01413769e-01 6.19141221e-01 7.11947083e-01 2.39751220e-01 -1.62667382e+00 9.68445241e-01 2.09352195e-01 -1.11085510e+00 -2.47042716e-01 -2.76206225e-01 7.06516027e-01 -1.75170854e-01 -2.25707173e-01 6.44518197e-01 5.50284684e-01 -1.23189986e+00 2.83268601e-01 1.69908196e-01 5.45342743e-01 -7.20425367e-01 7.90027440e-01 -7.06840903e-02 -9.50172007e-01 4.70343381e-02 2.79998556e-02 -6.96021557e-01 4.65739936e-01 9.96334374e-01 -1.01894283e+00 3.02874863e-01 5.71829379e-01 1.16386306e+00 5.07094897e-02 1.59140509e-02 -7.45231688e-01 1.06721950e+00 -1.50926620e-01 -3.00625980e-01 1.98051974e-01 2.62793124e-01 7.09089816e-01 1.69673824e+00 -2.89698452e-01 4.24717903e-01 -6.22052327e-02 8.90492141e-01 -5.40946186e-01 7.27708265e-02 -8.50192904e-01 -3.64528187e-02 4.35038209e-01 1.60526037e+00 -2.20860019e-02 -2.84269005e-01 -4.05178480e-02 9.69349384e-01 4.08289641e-01 1.87986001e-01 -9.61387515e-01 -1.49148300e-01 1.37772751e+00 -6.54273570e-01 -2.54104882e-01 1.38180673e-01 -4.03828084e-01 -1.04914093e+00 -2.49230281e-01 -1.27907801e+00 4.14695352e-01 -6.27528250e-01 -1.80334020e+00 5.05757868e-01 -1.20764934e-01 -6.18921340e-01 -7.31441677e-01 -3.28530073e-01 -1.06421149e+00 7.60431528e-01 -1.46997559e+00 -1.06159735e+00 -1.81291357e-01 6.04713678e-01 1.00323308e+00 1.45534351e-01 1.22716022e+00 -4.27807905e-02 -8.12903166e-01 4.44009662e-01 -6.01860702e-01 2.43679941e-01 1.08595431e+00 -1.62380493e+00 1.26383170e-01 4.39993024e-01 5.18454164e-02 4.09778088e-01 9.99280691e-01 -4.54623967e-01 -1.62502217e+00 -9.37606752e-01 1.56859779e+00 -1.08380318e+00 9.99889314e-01 -7.25150704e-01 -7.05832064e-01 7.00358808e-01 9.29066360e-01 -3.89800102e-01 1.15163469e+00 1.11788356e+00 -5.87578833e-01 1.45569772e-01 -1.01489174e+00 5.53740084e-01 9.77561593e-01 -6.13972783e-01 -9.98269618e-01 5.75201392e-01 1.13601303e+00 -3.20365965e-01 -9.31718647e-01 3.15473139e-01 4.74953502e-01 -9.71516430e-01 5.24872422e-01 -9.39500391e-01 1.33937109e+00 3.48888695e-01 -6.74493387e-02 -1.83168590e+00 5.89917451e-02 -1.15541625e+00 -8.77768397e-02 1.86432159e+00 6.82481050e-01 -3.57625455e-01 2.93126702e-01 1.06955767e+00 -2.58705765e-01 -9.78252649e-01 -9.64387238e-01 -2.91493654e-01 3.16216022e-01 -9.95236456e-01 5.03235459e-01 1.39803612e+00 6.51918590e-01 1.07182431e+00 -4.68757331e-01 -1.37851670e-01 3.00243884e-01 2.30222702e-01 9.16448891e-01 -8.15191627e-01 -2.03385949e-01 -5.29267192e-01 1.19315699e-01 -8.49842429e-01 7.76737869e-01 -6.94240987e-01 3.69664878e-01 -1.54036593e+00 3.81795578e-02 -2.88557142e-01 7.55655020e-02 8.10865998e-01 -5.71733356e-01 -6.95883557e-02 1.46124870e-01 -2.39978805e-01 -4.44443434e-01 1.03389335e+00 9.92549241e-01 -3.77574451e-02 -1.81935817e-01 -3.53700280e-01 -7.10667312e-01 8.19168329e-01 7.74218976e-01 -2.72559464e-01 -3.75211865e-01 -2.26386070e-01 3.70182097e-01 8.29636157e-02 3.16147327e-01 -5.35600483e-02 -3.18260372e-01 -3.13552231e-01 -1.12772226e-01 -2.16711536e-01 6.01659834e-01 -3.23360652e-01 -6.18237913e-01 -1.53073668e-01 -8.20727348e-01 -3.02385986e-02 5.66955388e-01 2.85547405e-01 -4.80606586e-01 1.86140239e-01 1.60764322e-01 7.52081424e-02 -5.49707949e-01 -1.45221889e-01 -7.21005082e-01 5.94370306e-01 6.99448109e-01 4.44823146e-01 -4.68661606e-01 -1.09048033e+00 -4.30200517e-01 3.81596595e-01 -4.86652553e-01 8.47613633e-01 5.23282886e-01 -1.24117434e+00 -1.05600166e+00 -5.29155254e-01 1.52143156e-02 -3.62922072e-01 2.22824723e-01 9.08257961e-01 5.02736688e-01 3.80062133e-01 3.37549925e-01 -2.31565967e-01 -1.51886678e+00 1.98961511e-01 3.55990350e-01 -5.10504067e-01 -2.16432482e-01 1.37660527e+00 4.52567816e-01 -9.75677788e-01 1.07995063e-01 -3.47350597e-01 -3.43751490e-01 5.06667495e-01 5.77134311e-01 1.32519454e-01 -2.35309258e-01 -4.86597389e-01 -5.37603438e-01 2.02956036e-01 1.32631138e-01 -2.53206462e-01 1.29208350e+00 -2.85134822e-01 -5.02416790e-01 9.74535108e-01 1.50645137e+00 9.06985626e-02 -7.31710017e-01 -9.08912495e-02 -1.20025739e-01 1.67528670e-02 -6.36481643e-02 -1.22357845e+00 -6.00620389e-01 9.99569058e-01 -1.66455001e-01 6.21507227e-01 8.90690565e-01 3.36036056e-01 9.43251908e-01 9.17086825e-02 -2.91132331e-01 -9.12304819e-01 3.03248793e-01 1.20819211e+00 1.75152194e+00 -1.33924520e+00 -4.09717798e-01 -6.63157403e-01 -9.67798769e-01 1.08782768e+00 6.68245375e-01 1.79040000e-01 3.49302918e-01 5.91007113e-01 6.06933892e-01 -5.31704605e-01 -1.61317289e+00 -3.43057811e-01 2.32359961e-01 3.11711550e-01 1.15898883e+00 2.81156123e-01 -2.84759820e-01 1.06475794e+00 -5.87074935e-01 -4.77948219e-01 5.99154055e-01 2.28310406e-01 2.48496339e-01 -1.14114273e+00 -2.71410704e-01 3.29771668e-01 -6.15665317e-01 -4.11282778e-01 -1.29374528e+00 6.40062630e-01 9.61352978e-03 1.74119973e+00 5.29889464e-02 -5.77416837e-01 3.34251493e-01 4.81465071e-01 2.21961483e-01 -5.68384230e-01 -1.04755962e+00 -1.28958046e-01 1.20757461e+00 -7.86871612e-01 -5.02295077e-01 -7.73431063e-01 -1.50244713e+00 -5.00153005e-01 -1.87776405e-02 2.76348114e-01 5.54862857e-01 1.24775267e+00 5.03653646e-01 8.18463862e-01 1.07867122e+00 -8.56977284e-01 -1.51436910e-01 -1.28397274e+00 -5.97455353e-02 8.35180104e-01 3.52736771e-01 -5.00156462e-01 -8.61370623e-01 3.58549319e-02]
[12.80037784576416, 6.335583209991455]
980b31d6-2856-486c-8cb0-5c53ff1d5a77
itsa-an-information-theoretic-approach-to
2201.02263
null
https://arxiv.org/abs/2201.02263v2
https://arxiv.org/pdf/2201.02263v2.pdf
ITSA: An Information-Theoretic Approach to Automatic Shortcut Avoidance and Domain Generalization in Stereo Matching Networks
State-of-the-art stereo matching networks trained only on synthetic data often fail to generalize to more challenging real data domains. In this paper, we attempt to unfold an important factor that hinders the networks from generalizing across domains: through the lens of shortcut learning. We demonstrate that the learning of feature representations in stereo matching networks is heavily influenced by synthetic data artefacts (shortcut attributes). To mitigate this issue, we propose an Information-Theoretic Shortcut Avoidance~(ITSA) approach to automatically restrict shortcut-related information from being encoded into the feature representations. As a result, our proposed method learns robust and shortcut-invariant features by minimizing the sensitivity of latent features to input variations. To avoid the prohibitive computational cost of direct input sensitivity optimization, we propose an effective yet feasible algorithm to achieve robustness. We show that using this method, state-of-the-art stereo matching networks that are trained purely on synthetic data can effectively generalize to challenging and previously unseen real data scenarios. Importantly, the proposed method enhances the robustness of the synthetic trained networks to the point that they outperform their fine-tuned counterparts (on real data) for challenging out-of-domain stereo datasets.
['David Suter', 'Alireza Bab-Hadiashar', 'Reza Hoseinnezhad', 'Ruwan Tennakoon', 'WeiQin Chuah']
2022-01-06
null
http://openaccess.thecvf.com//content/CVPR2022/html/Chuah_ITSA_An_Information-Theoretic_Approach_to_Automatic_Shortcut_Avoidance_and_Domain_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Chuah_ITSA_An_Information-Theoretic_Approach_to_Automatic_Shortcut_Avoidance_and_Domain_CVPR_2022_paper.pdf
cvpr-2022-1
['stereo-matching-1']
['computer-vision']
[ 5.13079822e-01 3.02807778e-01 7.33088627e-02 -6.04830384e-01 -5.61538875e-01 -6.23547137e-01 7.94197202e-01 -4.40562725e-01 -4.15513933e-01 7.18998373e-01 9.01470259e-02 1.40674517e-01 -4.22256857e-01 -7.96157837e-01 -1.09681737e+00 -5.67392170e-01 2.08774939e-01 3.24538559e-01 3.76088411e-01 -2.53535211e-01 2.94337720e-01 7.23963141e-01 -1.94719541e+00 2.60875523e-01 9.89799321e-01 8.31412137e-01 1.25793844e-01 1.77249283e-01 4.47289683e-02 3.59978557e-01 -2.85411865e-01 -4.44717467e-01 7.60536015e-01 -1.89582139e-01 -7.12541997e-01 1.80927113e-01 9.42614019e-01 -1.25287473e-01 -6.82538092e-01 1.14746618e+00 3.56721193e-01 8.82783681e-02 8.15667331e-01 -1.27872336e+00 -5.86933136e-01 9.88487080e-02 -3.26859087e-01 -3.55009511e-02 1.72161683e-01 4.52799171e-01 8.68545115e-01 -9.33382750e-01 1.03177559e+00 1.38052464e+00 7.88883507e-01 7.73942173e-01 -1.58163917e+00 -6.85515165e-01 1.23976447e-01 -2.00276449e-01 -1.26274538e+00 -5.58632135e-01 9.58210826e-01 -5.32473803e-01 7.05835462e-01 8.90442058e-02 5.00612617e-01 1.46997035e+00 1.06009215e-01 8.17471385e-01 1.14767992e+00 -3.33240986e-01 2.15423331e-01 1.11716725e-01 -1.20908573e-01 5.72462797e-01 3.25653791e-01 7.26890087e-01 -6.89045668e-01 1.01179525e-01 8.29894245e-01 -3.48698907e-02 -3.28263819e-01 -1.20895517e+00 -1.20953715e+00 8.45443904e-01 6.67594731e-01 1.24637432e-01 2.57338546e-02 1.54023031e-02 3.65846783e-01 5.50725639e-01 4.63898569e-01 8.10207188e-01 -4.53754932e-01 3.16668838e-01 -8.48976076e-01 4.15050536e-01 4.31476861e-01 1.17889607e+00 1.04777372e+00 -2.61926614e-02 -1.05325103e-01 9.10843730e-01 2.03018263e-02 4.30075645e-01 3.80892992e-01 -9.92364228e-01 6.15514755e-01 7.76822209e-01 -5.13867140e-02 -8.60383332e-01 -3.01061064e-01 -5.85565627e-01 -8.01257312e-01 4.89227772e-01 5.32893300e-01 -4.49920818e-02 -1.12955844e+00 2.01462746e+00 7.35078901e-02 -1.62029952e-01 2.10225761e-01 8.18294525e-01 6.81959450e-01 2.04601839e-01 -1.74742192e-01 1.71163008e-01 5.74352264e-01 -7.03893304e-01 -1.49045497e-01 -4.28532183e-01 4.81939584e-01 -5.38670659e-01 1.24717927e+00 1.36205584e-01 -9.20477927e-01 -6.27847254e-01 -1.34185922e+00 -4.49102670e-02 -5.28219402e-01 -2.56488651e-01 5.95925689e-01 5.33451498e-01 -9.72092450e-01 8.88181686e-01 -5.66774905e-01 -8.43601882e-01 6.19831562e-01 5.58410048e-01 -7.25836813e-01 -2.92851537e-01 -9.84269559e-01 9.07326937e-01 5.08842587e-01 -1.52771443e-01 -9.21271563e-01 -9.25558925e-01 -9.80051041e-01 -1.34155437e-01 3.37582231e-01 -8.74255478e-01 9.40091670e-01 -1.20517778e+00 -1.36923707e+00 1.03306341e+00 1.12243809e-01 -3.26169521e-01 7.88192391e-01 -2.62531221e-01 -4.86595780e-02 1.03928626e-01 2.43470296e-01 1.17137516e+00 1.08967113e+00 -1.40607953e+00 -4.08568591e-01 -4.68193978e-01 1.31338775e-01 9.90708321e-02 -2.82755911e-01 -4.44327772e-01 -2.18688235e-01 -8.49195838e-01 2.45420560e-01 -1.05649078e+00 -2.07269534e-01 4.64512795e-01 -3.83873105e-01 9.52457339e-02 7.29455829e-01 4.69685607e-02 6.80882931e-01 -2.28289413e+00 2.96407461e-01 2.75483042e-01 4.74075228e-02 3.25720787e-01 -3.37186158e-01 3.97051990e-01 -2.35415295e-01 -7.69326165e-02 -4.77812946e-01 -3.53888050e-02 -9.12255198e-02 3.77550930e-01 -4.27036405e-01 5.46760142e-01 4.15659726e-01 8.55086744e-01 -8.24727058e-01 -3.15576613e-01 3.26239765e-01 3.27483684e-01 -7.66783655e-01 3.25081170e-01 -2.48263076e-01 6.29752874e-01 -4.70741063e-01 3.63426059e-01 8.44871581e-01 9.46091115e-02 -2.26422116e-01 -1.72810748e-01 4.55029355e-03 -2.42681187e-02 -1.15069807e+00 1.99870360e+00 -5.25335312e-01 6.74232781e-01 -1.43633291e-01 -1.15289903e+00 1.07986712e+00 -5.40697835e-02 2.75054604e-01 -7.68109202e-01 -1.38600528e-01 3.58901739e-01 -1.86153919e-01 -4.13281143e-01 1.39425159e-01 -3.50299805e-01 -1.15269594e-01 2.29739323e-01 4.58215684e-01 -4.32793379e-01 4.67000827e-02 -1.60322133e-02 8.63145769e-01 3.28767419e-01 1.07460290e-01 -6.58598244e-01 5.34204423e-01 -1.01162270e-02 7.00615466e-01 9.49237764e-01 -7.36975297e-02 9.63220119e-01 3.13305885e-01 -6.93746209e-01 -1.32118535e+00 -1.31729853e+00 -3.89854699e-01 9.71953154e-01 1.24055155e-01 5.35147488e-02 -7.82093287e-01 -8.48748803e-01 3.61123025e-01 5.65488756e-01 -8.30355465e-01 -4.27259475e-01 -5.70928872e-01 -4.30049568e-01 6.38172269e-01 4.77850467e-01 6.05310977e-01 -1.03559613e+00 -6.60555124e-01 1.44213453e-01 2.06935897e-01 -9.76679564e-01 -2.93211073e-01 2.71262527e-01 -8.73362243e-01 -1.11692476e+00 -8.72586727e-01 -6.60330832e-01 8.35990965e-01 2.82696515e-01 1.04401934e+00 -9.12256986e-02 -3.90511870e-01 4.17445600e-01 -1.15576029e-01 -1.27214685e-01 -3.52998912e-01 3.19771200e-01 9.02429670e-02 7.00718537e-02 3.04606736e-01 -8.74761581e-01 -5.41755140e-01 5.60807049e-01 -1.13959062e+00 1.17429443e-01 5.54492950e-01 1.21761954e+00 3.69087249e-01 -3.78241837e-01 4.28176820e-01 -9.21301067e-01 4.26920265e-01 -1.22896358e-01 -7.64026880e-01 2.73511350e-01 -4.80989724e-01 6.53814793e-01 8.60696197e-01 -3.54674697e-01 -1.22342026e+00 3.51687372e-01 1.52477384e-01 -4.52235639e-01 -3.52821022e-01 -3.14553119e-02 -1.28000170e-01 -5.45204878e-01 1.08597362e+00 2.51080751e-01 3.98912132e-02 -3.99417371e-01 4.88651097e-01 3.87565404e-01 5.78953564e-01 -7.85817564e-01 1.26062560e+00 8.46612275e-01 2.82153875e-01 -6.97042823e-01 -1.02851427e+00 -3.80794287e-01 -9.80552316e-01 -7.14947730e-02 6.83899820e-01 -8.27761292e-01 -2.21583739e-01 3.84777308e-01 -8.97345185e-01 -1.62872851e-01 -5.22475898e-01 1.79504246e-01 -1.07016194e+00 2.99651146e-01 -2.00371593e-01 -3.03382218e-01 4.51988392e-02 -1.11382258e+00 1.04825342e+00 1.59737721e-01 -1.71855643e-01 -8.92329395e-01 5.50861619e-02 8.19386095e-02 4.53864872e-01 3.96985799e-01 8.12123179e-01 -4.79533792e-01 -7.05118477e-01 2.46643485e-03 -4.77364749e-01 3.96279037e-01 6.67936131e-02 -8.46838132e-02 -1.21500969e+00 -5.96797645e-01 -6.19530864e-02 -7.17874765e-01 1.19231713e+00 1.76961169e-01 1.13700116e+00 -1.06072925e-01 -2.75987267e-01 1.19094408e+00 1.48095906e+00 -1.89670235e-01 4.94755864e-01 5.20314634e-01 5.48199594e-01 9.42065537e-01 5.98437846e-01 1.68801472e-01 -1.28493980e-01 7.01567531e-01 5.22373199e-01 -2.29499072e-01 -2.51481414e-01 -5.17205954e-01 1.47855088e-01 2.14480251e-01 1.24054171e-01 8.41624811e-02 -7.47874439e-01 7.19027400e-01 -1.85115993e+00 -8.17390800e-01 2.89980173e-01 2.31664491e+00 7.18313158e-01 4.21829790e-01 -1.54669806e-01 8.64392146e-02 6.17915988e-01 2.92339951e-01 -8.19633842e-01 -2.56125987e-01 -4.05151188e-01 1.89262360e-01 6.36841297e-01 3.50919753e-01 -1.24877858e+00 1.07756293e+00 6.04002190e+00 6.92082644e-01 -1.08215857e+00 -2.86737204e-01 3.90773118e-01 -1.63805194e-03 -4.68410075e-01 -9.71570015e-02 -4.85380292e-01 6.10948130e-02 3.80523741e-01 -1.28195360e-01 4.39977229e-01 8.99499416e-01 -7.79394805e-02 1.54466689e-01 -1.49980342e+00 8.70338380e-01 -5.54191433e-02 -1.44137812e+00 3.96165818e-01 3.50957252e-02 8.93537819e-01 1.41410723e-01 2.75220782e-01 1.49026245e-01 3.17515284e-01 -9.93216753e-01 6.39929593e-01 4.46046263e-01 9.43660557e-01 -6.08809829e-01 4.25815314e-01 2.64296889e-01 -8.52571905e-01 -1.91228747e-01 -6.57168388e-01 -3.45838144e-02 -9.71306488e-02 4.71562982e-01 -7.83415139e-01 4.22091007e-01 5.17330170e-01 8.41831744e-01 -7.35011876e-01 1.10145414e+00 -1.17284715e-01 1.08896876e-02 -2.82351673e-01 2.81103402e-01 3.85751873e-01 -1.29991742e-02 5.90703428e-01 1.16821766e+00 3.53651702e-01 -1.70856103e-01 -1.25396311e-01 1.12207735e+00 6.43112063e-02 -2.56878346e-01 -1.30127406e+00 3.18627685e-01 1.63277045e-01 7.21073210e-01 -4.43083107e-01 -1.38315689e-02 -3.35020721e-01 9.15495276e-01 6.18409097e-01 5.98464310e-01 -3.79611522e-01 -3.23015928e-01 7.90710330e-01 1.40439808e-01 4.23797876e-01 5.02159186e-02 -4.68473226e-01 -1.31055856e+00 2.02521682e-01 -8.75799179e-01 3.19968969e-01 -5.93248665e-01 -1.62596357e+00 4.98832166e-01 7.87713677e-02 -1.53754377e+00 -3.59912664e-01 -8.27408135e-01 -4.02903616e-01 6.47604704e-01 -1.57797170e+00 -1.07950330e+00 -2.98757672e-01 8.91867578e-01 5.78140676e-01 -3.67159188e-01 7.06168056e-01 4.96656783e-02 -2.25959107e-01 8.05242896e-01 2.67742604e-01 -9.02840961e-03 8.07848513e-01 -1.11848128e+00 6.32608175e-01 7.32044101e-01 -2.28787083e-02 6.54195726e-01 7.65163243e-01 -3.14006627e-01 -1.14005423e+00 -1.10091531e+00 6.20067000e-01 -2.93086290e-01 5.22382677e-01 -6.58473253e-01 -8.33147109e-01 5.86208344e-01 -7.41775185e-02 2.54214704e-01 2.55319268e-01 1.01560511e-01 -7.90671170e-01 -3.78080100e-01 -1.36170375e+00 7.04751134e-01 1.58731282e+00 -6.61923051e-01 -9.62918937e-01 1.63587242e-01 6.68270707e-01 -1.55640200e-01 -5.39977252e-01 8.21628630e-01 5.38139880e-01 -1.36379993e+00 1.10370517e+00 -8.14402759e-01 4.91658241e-01 -7.62883574e-02 -2.38185987e-01 -1.44154346e+00 -1.93800449e-01 -6.29810393e-01 4.37102407e-01 1.01734543e+00 3.95611256e-01 -6.64822459e-01 8.63764346e-01 4.65150595e-01 -1.49292335e-01 -5.21986306e-01 -1.16185820e+00 -1.01957595e+00 3.44793588e-01 -2.18632877e-01 3.71437311e-01 9.69790816e-01 -3.62847634e-02 2.99304537e-02 -3.05473357e-01 8.73095263e-03 1.00406742e+00 2.20786795e-01 8.87421131e-01 -1.42373109e+00 -1.67097509e-01 -2.75393486e-01 -6.71923697e-01 -1.09799135e+00 3.84243786e-01 -6.82962477e-01 4.50682119e-02 -9.96600628e-01 2.01506257e-01 -5.22586823e-01 -2.58874476e-01 7.89657906e-02 6.63256869e-02 1.90093279e-01 2.80213207e-01 1.42151117e-01 -3.77107203e-01 7.97187030e-01 1.46737063e+00 -1.77007243e-01 5.53267486e-02 1.27380446e-01 -4.01451230e-01 8.22221875e-01 6.91191673e-01 -5.74583590e-01 -6.52798712e-01 -5.08783817e-01 7.49063343e-02 -1.25163153e-01 4.55605686e-01 -1.12397587e+00 1.19603120e-01 -2.45354474e-01 3.46545964e-01 -2.80969858e-01 3.30288023e-01 -8.93498719e-01 -1.45740900e-02 3.98745179e-01 -5.31698644e-01 -2.06026047e-01 2.35701263e-01 5.25406241e-01 -2.66029805e-01 -2.07721010e-01 1.03262949e+00 -2.94019848e-01 -6.21227801e-01 2.75505126e-01 -7.39194751e-02 4.07828599e-01 9.28981602e-01 -5.44240892e-01 -4.05405164e-01 -3.95027936e-01 -6.07525766e-01 1.29679233e-01 8.28088760e-01 6.03130758e-01 6.18289828e-01 -1.26339436e+00 -5.49700677e-01 6.56561077e-01 5.53790152e-01 1.27694219e-01 3.00756190e-03 1.51156247e-01 -4.70491499e-01 5.77000439e-01 -7.18501508e-01 -7.88090169e-01 -7.90134668e-01 7.28016615e-01 5.70559263e-01 -4.33721952e-03 -7.28390455e-01 8.54267240e-01 7.62103140e-01 -8.88897717e-01 4.07794207e-01 -2.67059028e-01 1.50391504e-01 -2.37056717e-01 7.92158674e-03 1.52393579e-01 -2.77081933e-02 -5.25879681e-01 -1.93078920e-01 9.30090725e-01 -2.05368087e-01 -1.79925904e-01 1.45002770e+00 -2.03145891e-02 2.83578813e-01 1.84704840e-01 1.27415109e+00 -3.68597716e-01 -1.90363789e+00 -4.13503647e-01 1.66328520e-01 -7.62076259e-01 -2.87451446e-01 -5.76331437e-01 -1.09835136e+00 1.01223624e+00 6.61701083e-01 -7.60266036e-02 1.03826499e+00 4.11325023e-02 5.26089132e-01 6.13570631e-01 4.91268039e-01 -1.15452313e+00 1.20795496e-01 4.31338191e-01 1.06522775e+00 -1.38977206e+00 -2.01666191e-01 -5.95326841e-01 -3.89446050e-01 1.17393398e+00 8.03157270e-01 -4.04497266e-01 4.96208191e-01 -5.27724065e-03 8.99980888e-02 -2.53812134e-01 -5.34560740e-01 -1.73720911e-01 3.20883811e-01 9.54301715e-01 1.47128865e-01 -2.93639570e-01 -1.30324066e-01 -9.66401957e-03 -2.32060537e-01 -1.60709303e-02 4.50653315e-01 8.04690003e-01 -3.56666565e-01 -1.09873700e+00 -1.98512182e-01 1.88689619e-01 -1.47723541e-01 6.13863803e-02 -5.39190710e-01 1.08760619e+00 2.28097528e-01 5.60737967e-01 -2.95199528e-02 -2.28875235e-01 6.79965734e-01 -5.16599007e-02 5.34474850e-01 -5.49499273e-01 -3.20705861e-01 -3.62625241e-01 -1.84454825e-02 -7.88749814e-01 -5.36875844e-01 -6.20790720e-01 -6.37059867e-01 3.58101837e-02 4.63544112e-03 -2.18116805e-01 2.44991019e-01 9.04774964e-01 2.96336055e-01 1.93887562e-01 6.92823410e-01 -9.87403393e-01 -8.69881451e-01 -7.14472830e-01 -4.10037905e-01 9.55544770e-01 5.11893511e-01 -1.00736928e+00 -4.88460988e-01 -2.01748341e-01]
[8.71228313446045, -2.3448166847229004]
d0002de2-9b09-4e18-b3d0-3cdf01c96676
weakly-supervised-multi-face-3d
2101.02
null
https://arxiv.org/abs/2101.02000v1
https://arxiv.org/pdf/2101.02000v1.pdf
Weakly-Supervised Multi-Face 3D Reconstruction
3D face reconstruction plays a very important role in many real-world multimedia applications, including digital entertainment, social media, affection analysis, and person identification. The de-facto pipeline for estimating the parametric face model from an image requires to firstly detect the facial regions with landmarks, and then crop each face to feed the deep learning-based regressor. Comparing to the conventional methods performing forward inference for each detected instance independently, we suggest an effective end-to-end framework for multi-face 3D reconstruction, which is able to predict the model parameters of multiple instances simultaneously using single network inference. Our proposed approach not only greatly reduces the computational redundancy in feature extraction but also makes the deployment procedure much easier using the single network model. More importantly, we employ the same global camera model for the reconstructed faces in each image, which makes it possible to recover the relative head positions and orientations in the 3D scene. We have conducted extensive experiments to evaluate our proposed approach on the sparse and dense face alignment tasks. The experimental results indicate that our proposed approach is very promising on face alignment tasks without fully-supervision and pre-processing like detection and crop. Our implementation is publicly available at \url{https://github.com/kalyo-zjl/WM3DR}.
['Steven C. H. Hoi', 'Jianke Zhu', 'Lixiang Lin', 'Jialiang Zhang']
2021-01-06
null
null
null
null
['person-identification', 'face-alignment', 'face-model', 'face-reconstruction']
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
[-2.03775495e-01 3.42919938e-02 9.40930918e-02 -6.94177389e-01 -4.97262597e-01 -2.31527999e-01 4.28711623e-01 -3.58268768e-01 -4.00904983e-01 1.68031529e-01 2.36947555e-02 1.00249484e-01 1.20839439e-01 -4.53462839e-01 -7.61493504e-01 -7.04153597e-01 1.19113490e-01 5.53008378e-01 -2.55457312e-01 1.14615761e-01 2.03310717e-02 8.53416443e-01 -1.50833774e+00 -2.11693242e-01 2.21023649e-01 1.04559624e+00 1.77453697e-01 3.77298057e-01 2.67346472e-01 4.56850857e-01 -2.06605554e-01 -7.59495974e-01 3.87405813e-01 -7.08858520e-02 -2.91622579e-01 5.01171589e-01 7.70434380e-01 -6.58853233e-01 -2.81275928e-01 1.16934800e+00 7.92461336e-01 1.07129570e-02 3.83239120e-01 -1.23075104e+00 -1.38872534e-01 1.75373599e-01 -1.19030201e+00 -7.17856511e-02 5.36918819e-01 -5.30208610e-02 5.68059742e-01 -1.15148890e+00 4.91646647e-01 1.52990770e+00 6.58021152e-01 5.59230804e-01 -8.52966309e-01 -9.74531651e-01 1.64835721e-01 1.52250469e-01 -1.73259842e+00 -1.14776218e+00 9.05945480e-01 -2.60865033e-01 5.01934111e-01 8.65736753e-02 6.16052330e-01 9.82210100e-01 -2.37780780e-01 6.10172153e-01 8.94728899e-01 -3.84835988e-01 -1.04437239e-01 5.90535477e-02 -1.02648005e-01 1.08513522e+00 6.73348382e-02 -2.33248517e-01 -5.03264129e-01 -1.45016924e-01 1.10300016e+00 1.38020039e-01 -9.86695960e-02 -2.71550268e-01 -9.56678271e-01 6.01157427e-01 2.25648835e-01 5.49688051e-03 -4.33706760e-01 8.19747373e-02 1.37931883e-01 -1.22261951e-02 7.29355931e-01 -4.75798279e-01 -2.96987802e-01 1.60707220e-01 -9.54920709e-01 2.85339057e-01 5.90479970e-01 1.00419748e+00 8.99724603e-01 1.60303502e-03 2.46458337e-01 9.76860702e-01 6.22482240e-01 6.75098360e-01 1.12947911e-01 -1.02203298e+00 3.80160540e-01 2.45202243e-01 7.11377338e-02 -1.36724377e+00 -6.04230106e-01 -3.69739950e-01 -9.56321418e-01 -1.43421011e-03 4.57822442e-01 -2.42341951e-01 -6.53551102e-01 1.75441730e+00 8.52311671e-01 6.07789755e-01 -4.29375023e-01 1.03718460e+00 8.04769278e-01 3.66127163e-01 -2.14960009e-01 -2.97214359e-01 1.51016593e+00 -7.77840197e-01 -6.22343004e-01 -4.02932435e-01 2.33077243e-01 -8.90397668e-01 5.60771704e-01 1.76500142e-01 -1.11081433e+00 -5.62283754e-01 -6.93688750e-01 -1.34625748e-01 2.17861071e-01 5.76730728e-01 5.82885265e-01 5.55324793e-01 -1.27935743e+00 9.12958011e-02 -9.51183498e-01 -4.10187870e-01 5.84954500e-01 7.38344371e-01 -9.07433331e-01 -1.97062358e-01 -5.79593360e-01 7.22467005e-01 -2.59439889e-02 6.95036352e-01 -8.00485730e-01 -1.98219672e-01 -1.07298028e+00 -8.52066744e-03 3.51750731e-01 -6.72676086e-01 1.11105502e+00 -9.99471545e-01 -1.54855323e+00 1.09880066e+00 -6.50241971e-01 2.20527779e-02 4.76649821e-01 -2.32671589e-01 -1.23606667e-01 1.14567965e-01 9.04233977e-02 6.44168079e-01 1.20005381e+00 -1.09895778e+00 -2.76670724e-01 -8.84352684e-01 -9.48405117e-02 3.45516384e-01 -3.91007245e-01 4.74029571e-01 -1.06642532e+00 -3.58124763e-01 3.05019826e-01 -9.81420696e-01 -1.29917085e-01 3.55194807e-01 -4.10571128e-01 -8.40260908e-02 7.87074864e-01 -9.78101850e-01 7.53046155e-01 -2.16278362e+00 3.31249535e-01 3.25409055e-01 2.15989962e-01 2.63265874e-02 -8.75638947e-02 -1.44762933e-01 -1.11433350e-01 -3.83300811e-01 -7.38717988e-02 -1.12560546e+00 -1.48608968e-01 -1.35390490e-01 1.57772869e-01 1.00944626e+00 3.36765163e-02 5.90806484e-01 -3.79415005e-01 -5.19094586e-01 3.58868957e-01 9.81316388e-01 -5.96519887e-01 3.34959120e-01 5.82897924e-02 6.27923787e-01 -3.48826408e-01 6.78851902e-01 1.02731586e+00 -1.51602358e-01 2.51223892e-01 -2.12428138e-01 5.49582392e-02 -2.25809976e-01 -1.50711000e+00 1.81283510e+00 -4.61822659e-01 3.38005334e-01 5.88656187e-01 -9.28956807e-01 9.19028759e-01 4.31735456e-01 4.75584865e-01 -5.03196716e-01 4.43649709e-01 6.46944866e-02 -2.80607283e-01 -5.10251343e-01 9.90356132e-02 2.69760452e-02 1.48707435e-01 2.97890216e-01 2.49372855e-01 2.41255999e-01 -1.39981389e-01 -7.12375715e-02 4.53895539e-01 2.11105958e-01 3.16192061e-01 2.27854811e-02 7.87500858e-01 -7.73108423e-01 7.51569569e-01 -4.83448766e-02 -6.71407133e-02 6.31357312e-01 3.35508823e-01 -3.91006172e-01 -8.43287826e-01 -5.64856946e-01 -1.15555357e-02 8.46059740e-01 3.27726938e-02 -3.54947388e-01 -1.01310194e+00 -5.38427711e-01 -2.69153100e-02 2.37204596e-01 -4.76592869e-01 3.34172517e-01 -6.38578415e-01 -6.45246983e-01 3.05581480e-01 1.91483885e-01 4.36551005e-01 -7.71296620e-01 -9.76872444e-02 -1.87035367e-01 -3.07553440e-01 -1.45928395e+00 -6.51516438e-01 -4.29124445e-01 -6.60545111e-01 -1.04575121e+00 -7.06220746e-01 -8.73409033e-01 1.10534453e+00 5.35467446e-01 6.15872204e-01 2.87046850e-01 -1.84096709e-01 3.74986589e-01 1.16286010e-01 -2.15953410e-01 1.11535162e-01 -1.40534744e-01 4.32060570e-01 5.88185608e-01 3.94843310e-01 -7.07658708e-01 -6.53757513e-01 3.74180168e-01 -3.65623146e-01 -1.18947858e-02 3.79140943e-01 4.86561865e-01 6.46809757e-01 -5.69026582e-02 4.21871513e-01 -7.81807542e-01 1.28718078e-01 -4.63872939e-01 -8.32980216e-01 3.77191082e-02 9.35868919e-03 -3.42021376e-01 3.57376635e-01 -1.73177078e-01 -1.09717536e+00 5.59743345e-01 -4.82832670e-01 -7.24161386e-01 -3.46510857e-01 7.84270614e-02 -5.54083765e-01 -2.42888704e-01 1.73991591e-01 4.20527942e-02 3.12187910e-01 -6.63205087e-01 2.35611811e-01 6.82965517e-01 5.30277550e-01 -3.41619909e-01 9.12889183e-01 6.31671011e-01 1.41519874e-01 -1.13574541e+00 -6.30021572e-01 -4.28610146e-01 -9.99566972e-01 -3.41344565e-01 7.78665304e-01 -1.34898329e+00 -1.08153605e+00 7.51155615e-01 -1.33112800e+00 3.63396928e-02 4.40700144e-01 4.02469635e-01 -3.53810430e-01 5.20028293e-01 -4.49688822e-01 -8.81757975e-01 -4.16840941e-01 -1.21653605e+00 1.32210922e+00 3.05663347e-01 1.47711614e-03 -9.26603615e-01 -2.72211879e-01 5.58104873e-01 9.22712758e-02 1.35193750e-01 3.76807272e-01 -1.80860966e-01 -5.35102069e-01 -3.32706809e-01 -1.12152606e-01 1.45911217e-01 1.16297975e-01 -7.48049393e-02 -1.26363373e+00 -5.22832096e-01 2.14648813e-01 -1.66434348e-01 4.11134213e-01 5.24254203e-01 1.24583435e+00 -2.95214683e-01 -2.87474364e-01 9.04775679e-01 1.09380925e+00 -2.13258803e-01 4.23450083e-01 -2.57441085e-02 1.02007389e+00 8.19262743e-01 5.46523631e-01 6.44797146e-01 6.52680159e-01 1.03944552e+00 5.24047077e-01 -2.13189930e-01 -1.20292462e-01 -2.06246972e-01 3.91712189e-01 7.30131447e-01 -2.03965977e-01 1.31430821e-02 -6.35684729e-01 2.67224908e-01 -1.71792388e+00 -9.67154920e-01 1.12618856e-01 2.24437189e+00 4.47734296e-01 -2.78177589e-01 2.51503974e-01 -1.16140984e-01 8.75931978e-01 8.63484219e-02 -3.97484422e-01 8.29463452e-02 7.54598603e-02 1.79233477e-01 1.88402295e-01 7.72699356e-01 -1.15036929e+00 9.95230675e-01 5.13948584e+00 5.99771440e-01 -1.15041411e+00 3.04468244e-01 8.23932052e-01 -3.27081770e-01 8.38769078e-02 -3.14652413e-01 -1.19197583e+00 3.75550717e-01 6.55701458e-01 2.19018176e-01 4.39160049e-01 6.98101461e-01 5.25854588e-01 -8.24634880e-02 -1.07101643e+00 1.52656770e+00 3.97686899e-01 -9.39068198e-01 -1.50108114e-01 2.05723196e-01 2.80562013e-01 -1.82362452e-01 8.93719494e-02 -2.16326177e-01 -2.81412482e-01 -9.50643897e-01 7.71182418e-01 3.84042948e-01 9.21901047e-01 -9.03210878e-01 5.26044488e-01 4.48577762e-01 -1.17472064e+00 4.72091464e-03 -4.59704608e-01 -4.43747342e-02 1.61724254e-01 6.31948709e-01 -9.11277831e-01 3.48804414e-01 6.57854676e-01 7.41096675e-01 -4.60558861e-01 8.20165455e-01 -2.08372891e-01 1.94607198e-01 -4.96856570e-01 4.38375831e-01 -3.14841151e-01 -4.28289205e-01 4.72511888e-01 9.23620999e-01 5.09836614e-01 8.17310214e-02 1.17796652e-01 4.06986654e-01 -2.90961951e-01 1.39202893e-01 -5.88143706e-01 4.02808458e-01 4.41371679e-01 1.56999993e+00 -7.70192027e-01 5.48387952e-02 -6.63008153e-01 1.11429799e+00 4.93455887e-01 1.88265145e-01 -9.12725508e-01 1.43602148e-01 8.09256792e-01 2.44551137e-01 3.50062966e-01 -4.22804564e-01 1.17371231e-01 -1.22927547e+00 3.13234925e-01 -9.44135845e-01 7.44967908e-02 -5.23494840e-01 -9.18881714e-01 6.66459918e-01 -2.93337018e-03 -1.00947440e+00 -4.11484510e-01 -4.96949971e-01 -4.13676500e-01 8.46042633e-01 -1.46403944e+00 -1.46220565e+00 -3.75417322e-01 9.65369761e-01 4.57263201e-01 -1.79031789e-01 8.15650284e-01 6.72359228e-01 -1.06490052e+00 8.16691697e-01 -2.82214075e-01 2.66498566e-01 7.65100300e-01 -6.12627506e-01 4.15975392e-01 8.81353021e-01 2.78315753e-01 5.54255724e-01 4.88273770e-01 -4.05465156e-01 -1.67216742e+00 -9.26136911e-01 7.90653169e-01 -3.11297506e-01 1.46837950e-01 -7.15985119e-01 -5.69504976e-01 9.09301937e-01 -9.42421239e-03 1.51794016e-01 5.99801302e-01 1.45413622e-01 -2.06331953e-01 -3.65957737e-01 -1.18646359e+00 3.71964663e-01 1.08922756e+00 -4.12511319e-01 8.74036271e-03 5.07386863e-01 2.47217476e-01 -6.32561266e-01 -6.48076296e-01 7.23495707e-02 6.85882688e-01 -1.01739216e+00 1.08976829e+00 -4.78855101e-03 5.87935820e-02 -3.58877748e-01 -4.61903922e-02 -9.84052300e-01 -2.37516597e-01 -5.50257921e-01 -2.84158468e-01 1.43483877e+00 6.73783273e-02 -6.13899350e-01 8.82784367e-01 6.77140176e-01 2.56637752e-01 -6.81665480e-01 -1.00398123e+00 -9.05858874e-02 -4.82690215e-01 -2.82137901e-01 6.22542024e-01 8.25362504e-01 -4.81381893e-01 2.88239211e-01 -7.11602688e-01 7.07477868e-01 9.65974391e-01 2.29560174e-02 1.09830546e+00 -1.20329118e+00 -3.12679440e-01 -8.03646669e-02 -6.17911994e-01 -1.10137653e+00 6.00541830e-01 -6.88015878e-01 -2.41825879e-01 -1.04885507e+00 3.48033130e-01 -3.97142768e-01 1.06414422e-01 6.20090961e-01 -5.48279323e-02 5.07331550e-01 1.73624128e-01 1.26094744e-01 -3.43938738e-01 4.58145678e-01 1.01339018e+00 1.66338757e-01 6.16813041e-02 3.41359466e-01 -7.19206512e-01 9.68488693e-01 6.29481435e-01 -3.14181298e-01 -3.10612679e-01 -8.09180260e-01 -4.64595668e-02 1.04540981e-01 4.75324452e-01 -7.25151300e-01 3.72352391e-01 1.26222134e-01 7.80603230e-01 -4.00271267e-01 7.99128294e-01 -1.03293371e+00 2.52771556e-01 1.56363741e-01 2.95358449e-01 1.34309202e-01 2.03453302e-01 2.58135796e-01 -1.48498639e-01 -8.01536366e-02 8.60917211e-01 -6.88710958e-02 -5.79900026e-01 7.15297103e-01 1.95951894e-01 -4.81672913e-01 1.01596332e+00 -8.68165642e-02 1.20323278e-01 -5.80094278e-01 -7.03683019e-01 1.88466758e-01 5.52280903e-01 3.93646866e-01 7.23275483e-01 -1.22189689e+00 -8.59753191e-01 6.54302239e-01 -3.61978024e-01 2.15331391e-01 4.59616601e-01 8.79814327e-01 -4.49448436e-01 1.46482632e-01 -2.29746044e-01 -7.39620090e-01 -1.77107775e+00 3.47004682e-01 4.56455827e-01 2.40956172e-01 -5.16961277e-01 1.03701162e+00 2.35354766e-01 -4.85271603e-01 3.98391426e-01 2.17859417e-01 -2.13974789e-01 1.80113271e-01 6.93562567e-01 2.94053584e-01 8.66221562e-02 -1.27452636e+00 -4.85829383e-01 9.85292375e-01 -1.47178486e-01 -5.25050014e-02 1.51160622e+00 -2.66663939e-01 -4.74584907e-01 1.41447745e-02 1.35880899e+00 1.00675806e-01 -1.25887978e+00 -1.80070072e-01 -4.01587784e-01 -6.66572511e-01 1.61359429e-01 -1.16185144e-01 -1.57617950e+00 9.78619695e-01 5.56198597e-01 -5.07533193e-01 1.16031551e+00 -1.28962234e-01 4.39080387e-01 1.91270664e-01 5.30909181e-01 -5.97746670e-01 -1.56449243e-01 2.09274203e-01 9.14050400e-01 -1.26970088e+00 2.68321574e-01 -7.30152071e-01 -4.25860107e-01 1.08310509e+00 7.52369702e-01 4.21861261e-02 7.52856374e-01 2.59292990e-01 4.29686718e-02 -2.39125147e-01 -3.36771876e-01 1.10817544e-01 8.50223899e-02 3.69974881e-01 4.33510244e-01 -9.85846594e-02 2.55758017e-01 2.94840574e-01 -2.34661266e-01 -1.13414310e-01 1.84196040e-01 5.36259115e-01 -1.18005946e-01 -1.15421963e+00 -6.41134620e-01 1.58934653e-01 -4.39767063e-01 2.41498590e-01 -2.38550931e-01 5.67551613e-01 4.71069477e-02 7.99662948e-01 1.83595896e-01 -1.99718803e-01 2.56535292e-01 2.11916193e-02 8.03688526e-01 -5.45636415e-01 -2.24207699e-01 4.09789741e-01 -4.65821475e-02 -6.32845461e-01 -5.34909010e-01 -8.97184551e-01 -1.04517770e+00 -5.09934485e-01 -2.96895534e-01 -1.21451251e-01 7.15687037e-01 7.67697394e-01 4.55439419e-01 -5.63275442e-02 9.84434962e-01 -1.39824593e+00 -2.94069171e-01 -9.93536770e-01 -4.74690914e-01 1.84207246e-01 3.88952374e-01 -8.72315526e-01 -2.61855304e-01 4.94513996e-02]
[13.406183242797852, 0.3070332109928131]
086e9e5f-843e-44ee-9dca-a6c4e75b2949
190807899
1908.07899
null
https://arxiv.org/abs/1908.07899v1
https://arxiv.org/pdf/1908.07899v1.pdf
Evaluating Defensive Distillation For Defending Text Processing Neural Networks Against Adversarial Examples
Adversarial examples are artificially modified input samples which lead to misclassifications, while not being detectable by humans. These adversarial examples are a challenge for many tasks such as image and text classification, especially as research shows that many adversarial examples are transferable between different classifiers. In this work, we evaluate the performance of a popular defensive strategy for adversarial examples called defensive distillation, which can be successful in hardening neural networks against adversarial examples in the image domain. However, instead of applying defensive distillation to networks for image classification, we examine, for the first time, its performance on text classification tasks and also evaluate its effect on the transferability of adversarial text examples. Our results indicate that defensive distillation only has a minimal impact on text classifying neural networks and does neither help with increasing their robustness against adversarial examples nor prevent the transferability of adversarial examples between neural networks.
['Tobias Hinz', 'Marcus Soll', 'Sven Magg', 'Stefan Wermter']
2019-08-21
null
null
null
null
['adversarial-text']
['adversarial']
[ 7.43964314e-01 2.76297212e-01 3.83798569e-01 -2.98232913e-01 -2.44794011e-01 -1.27297711e+00 1.02232778e+00 1.97346523e-01 -5.63808858e-01 7.09330022e-01 -2.67094433e-01 -5.44098675e-01 1.36909217e-01 -8.64167035e-01 -1.10424840e+00 -5.92951655e-01 -1.10356836e-02 1.48413941e-01 2.02618644e-01 -4.45599675e-01 9.90765616e-02 9.44094121e-01 -1.25182223e+00 6.29328251e-01 7.28905320e-01 6.69346094e-01 -5.74527800e-01 9.65840697e-01 1.24788649e-01 9.12223577e-01 -1.34591866e+00 -9.79064584e-01 5.82262278e-01 -2.35520154e-01 -8.20257366e-01 -2.02080026e-01 9.41635907e-01 -4.64302570e-01 -5.03634989e-01 1.41975439e+00 3.35470080e-01 2.04572566e-02 8.20797145e-01 -1.53364468e+00 -1.00966799e+00 7.69385815e-01 -7.12058768e-02 1.88909382e-01 2.51437664e-01 4.31995213e-01 5.15910268e-01 -4.81948107e-01 4.61015493e-01 1.56655598e+00 6.10018373e-01 9.72552180e-01 -1.27897608e+00 -9.79915619e-01 1.12073906e-01 -1.26445860e-01 -8.85161042e-01 -3.00970852e-01 6.78914487e-01 -4.40325618e-01 5.93220055e-01 7.35225856e-01 1.38164237e-01 1.55557156e+00 4.20108616e-01 6.56504571e-01 1.32653689e+00 -3.58486801e-01 1.82329968e-01 6.10036969e-01 1.10664293e-01 3.68823171e-01 3.26847523e-01 3.99265796e-01 2.13754433e-03 -2.37215877e-01 2.06062898e-01 -2.18045294e-01 -4.38383430e-01 1.51462480e-02 -7.48318553e-01 8.20389211e-01 7.62131572e-01 3.27075124e-01 -5.36403283e-02 1.82824552e-01 5.91651320e-01 8.22067499e-01 5.30350089e-01 1.05511439e+00 -3.98742437e-01 3.07907283e-01 -4.64062810e-01 3.31020176e-01 7.82822967e-01 6.11232936e-01 3.05039197e-01 4.04852092e-01 -2.08047166e-01 6.70777142e-01 -1.91479057e-01 5.71860850e-01 5.86725593e-01 -4.68321353e-01 4.80912089e-01 6.12003922e-01 -2.57244170e-01 -1.08255041e+00 -1.25686258e-01 -2.87538022e-01 -7.87309647e-01 8.15352917e-01 7.33958542e-01 -3.76922935e-01 -1.05442142e+00 1.64492965e+00 -1.62163489e-02 -1.47483304e-01 3.29135954e-01 5.27442932e-01 6.58036053e-01 5.24666011e-01 1.78419754e-01 1.74529776e-01 9.35119987e-01 -7.15703309e-01 -3.38171095e-01 -5.32996714e-01 6.31034017e-01 -8.07021916e-01 1.22759473e+00 4.01048690e-01 -9.37394202e-01 -4.61838186e-01 -1.27796090e+00 3.38382810e-01 -1.12557805e+00 -5.74012280e-01 2.09197193e-01 1.19358909e+00 -7.13771462e-01 8.52627158e-01 -4.13088083e-01 -9.10496488e-02 6.71257377e-01 5.08544803e-01 -6.38692319e-01 -1.29841909e-01 -1.53047085e+00 1.13784826e+00 5.50056756e-01 -2.29159683e-01 -9.21338618e-01 -6.87305391e-01 -7.01271474e-01 2.10177258e-01 1.03736542e-01 -1.90200225e-01 1.00650418e+00 -1.81199443e+00 -1.00849044e+00 9.57588851e-01 6.68142378e-01 -9.02232289e-01 1.12930632e+00 -7.72135332e-02 -4.62127447e-01 6.82746470e-02 -4.05289561e-01 8.17248404e-01 1.34493792e+00 -1.43835247e+00 -2.00859427e-01 -1.98201925e-01 4.13928777e-01 -6.84666559e-02 -8.67913783e-01 -3.43151651e-02 4.67252851e-01 -1.12315786e+00 -6.04514837e-01 -1.11988032e+00 -1.10443652e-01 2.33291671e-01 -5.40454209e-01 1.51950419e-01 1.15128934e+00 -3.53273898e-01 5.83956718e-01 -2.25176597e+00 -1.43696636e-01 1.33644745e-01 8.41360390e-02 8.64830375e-01 -3.08512121e-01 2.03938365e-01 -6.20782852e-01 6.84018612e-01 -2.97461331e-01 1.20083228e-01 2.35402063e-02 3.80711466e-01 -7.66466916e-01 3.91264707e-01 5.75800836e-01 9.52351451e-01 -7.42778838e-01 -4.64075953e-02 1.83158129e-01 3.88456464e-01 -3.39089602e-01 4.46897261e-02 -2.39767820e-01 7.82988667e-02 -1.47814706e-01 3.35368395e-01 7.65218496e-01 2.54778862e-01 -2.49489143e-01 9.56168771e-02 6.06488287e-01 -2.94360608e-01 -7.00675905e-01 4.55923587e-01 -2.49213859e-01 1.07859075e+00 -1.81268424e-01 -1.01267409e+00 7.17330694e-01 2.21226737e-01 -2.89361864e-01 -4.83803362e-01 3.22809875e-01 -3.46063115e-02 4.73172486e-01 -2.37200320e-01 4.70725596e-01 -2.91398793e-01 -2.34924600e-01 5.07046342e-01 -8.36971030e-02 -3.59791845e-01 -1.00894853e-01 2.21931443e-01 1.16623652e+00 -4.29753006e-01 1.23341672e-01 -2.12734014e-01 7.03665197e-01 1.22182392e-01 3.67229991e-02 1.03187084e+00 -2.82330871e-01 3.55898291e-01 5.94849467e-01 -4.09435511e-01 -1.08549869e+00 -1.19491875e+00 -1.09685816e-01 1.15038097e+00 -1.28173366e-01 1.03065334e-02 -1.05512345e+00 -1.40381324e+00 3.22025001e-01 9.30056512e-01 -1.02041781e+00 -9.68568981e-01 -4.12221164e-01 -5.76171637e-01 1.17734826e+00 4.84413862e-01 6.25007749e-01 -1.11483443e+00 -2.90379822e-01 -3.35517794e-01 3.46075565e-01 -1.04839957e+00 -3.49751294e-01 3.78749430e-01 -6.97627842e-01 -1.06540215e+00 -5.78724205e-01 -5.22896767e-01 1.02658582e+00 9.64408368e-02 1.00519502e+00 4.22461480e-01 -3.93263817e-01 3.85735065e-01 -4.02633905e-01 -9.51139390e-01 -1.32766199e+00 -1.02393977e-01 8.33692476e-02 -1.16110690e-01 5.73324934e-02 -2.71999568e-01 -7.50255138e-02 4.50069994e-01 -1.52070320e+00 -4.75151390e-01 3.22728604e-01 9.04638886e-01 -1.40969321e-01 4.46599126e-01 5.84644556e-01 -1.31974661e+00 1.01514089e+00 -3.69030178e-01 -3.57417613e-01 1.62891790e-01 -5.23963928e-01 -3.62286754e-02 1.36718488e+00 -1.12923801e+00 -8.36401761e-01 -1.35098308e-01 -2.28047460e-01 -5.60373902e-01 -4.48657006e-01 1.89377248e-01 -1.18290238e-01 -6.39811695e-01 1.40641737e+00 -5.10530360e-02 1.56884603e-02 1.08360127e-02 2.60542661e-01 5.28511643e-01 4.42111760e-01 -3.12957555e-01 1.44684994e+00 4.23716426e-01 3.82526815e-02 -8.14627111e-01 -6.48648202e-01 3.06810290e-01 -4.97263253e-01 -2.10851252e-01 6.93735659e-01 -4.26250696e-01 -3.31759423e-01 7.68169641e-01 -9.30208266e-01 -4.28940028e-01 -2.48257175e-01 1.69519626e-03 -1.90268055e-01 3.45849305e-01 -5.27119637e-01 -4.53436971e-01 -1.22293152e-01 -1.05088603e+00 3.23569179e-01 6.70081005e-02 -2.78490633e-01 -1.12542379e+00 -3.85565519e-01 4.63285863e-01 4.36122268e-01 5.21730006e-01 1.07242036e+00 -1.36532891e+00 -2.46281534e-01 -5.84941328e-01 6.87977858e-03 9.04617488e-01 1.81921422e-01 1.20946303e-01 -1.32652462e+00 -4.66411084e-01 7.42854318e-04 -6.21268570e-01 7.76551604e-01 -2.82585591e-01 1.02352989e+00 -8.04693162e-01 -5.98025471e-02 4.27354306e-01 1.04714739e+00 2.98574418e-01 7.92550981e-01 5.08525372e-01 6.62042022e-01 8.56311202e-01 5.42522907e-01 -1.74752936e-01 -6.07623160e-01 2.67594963e-01 8.43576491e-01 -3.67681384e-01 9.99382064e-02 9.72023234e-03 5.13947308e-01 -1.24427803e-01 3.42139661e-01 -6.20517790e-01 -8.82200480e-01 1.99923426e-01 -1.37507331e+00 -1.04017437e+00 1.99600365e-02 1.88557339e+00 8.07578802e-01 6.79875851e-01 -8.84222165e-02 4.68164533e-01 8.65151107e-01 6.99953809e-02 -6.76291466e-01 -1.01481152e+00 -2.15043858e-01 2.76013672e-01 7.04068899e-01 3.37292820e-01 -1.22685277e+00 9.53881860e-01 6.68295574e+00 6.42025650e-01 -1.25361729e+00 -1.27583519e-01 8.53701293e-01 -1.65924430e-01 -6.19855151e-02 -4.53664362e-01 -2.34606996e-01 4.48669940e-01 8.42818797e-01 -2.63972044e-01 2.99808234e-01 1.12012005e+00 -2.30511039e-01 3.82776499e-01 -1.25222313e+00 2.76112974e-01 8.77889767e-02 -1.18033743e+00 5.33205628e-01 -4.86967564e-02 7.44103134e-01 -3.03668201e-01 4.47143376e-01 4.43575203e-01 6.21557295e-01 -1.39612675e+00 5.68888545e-01 1.69775560e-02 3.77503842e-01 -1.19091296e+00 8.81036341e-01 4.12109941e-01 -2.93020219e-01 -1.39365420e-01 -2.71283865e-01 -1.16944253e-01 -4.45911497e-01 1.22559093e-01 -1.06674862e+00 -5.15008233e-02 6.32651091e-01 1.32476017e-01 -1.03167212e+00 3.30334753e-01 -3.24030846e-01 6.34778142e-01 7.84837306e-02 -7.49765290e-03 3.44302595e-01 2.97067434e-01 6.30094230e-01 1.22667706e+00 -2.52245277e-01 -2.23399580e-01 -1.52866438e-01 7.48277605e-01 -4.99894351e-01 -1.72832876e-01 -1.27856767e+00 -3.97551626e-01 3.39608520e-01 1.00312877e+00 -7.65587091e-01 -3.70551318e-01 -2.23682579e-02 1.17287076e+00 1.47904992e-01 2.74170667e-01 -7.65533626e-01 -6.60918295e-01 7.89964199e-01 -4.90591638e-02 2.81061996e-02 2.40475863e-01 -3.22777897e-01 -7.58609414e-01 -2.96654906e-02 -1.46386218e+00 3.30771863e-01 -7.65127420e-01 -1.53899205e+00 6.54088736e-01 -1.16597421e-01 -1.01883173e+00 -3.39719445e-01 -9.31177735e-01 -8.41246307e-01 8.39255631e-01 -9.77458060e-01 -1.09933925e+00 -1.09337613e-01 8.11915040e-01 4.03327733e-01 -4.31653470e-01 7.29162276e-01 -1.52657151e-01 -3.83396029e-01 1.15532565e+00 1.56220227e-01 6.33493483e-01 8.76865566e-01 -1.25833225e+00 7.55674779e-01 1.07392013e+00 8.94695148e-02 6.31345570e-01 1.03058171e+00 -5.32064259e-01 -8.98963511e-01 -1.45946324e+00 2.70307511e-01 -7.51082063e-01 7.57019222e-01 -5.52305996e-01 -1.24727833e+00 8.52845788e-01 1.40118822e-01 1.33503512e-01 5.15812695e-01 -2.73086756e-01 -7.34983683e-01 9.98896137e-02 -1.63931739e+00 1.10760546e+00 7.36296475e-01 -7.83320069e-01 -6.32844687e-01 4.62754786e-01 8.25560927e-01 -2.61932760e-01 -7.61085093e-01 3.78079861e-01 4.01543170e-01 -8.62539351e-01 1.32658422e+00 -1.07423174e+00 7.64385462e-01 -1.46170780e-02 7.36930175e-03 -1.53607976e+00 -6.92921355e-02 -3.75269115e-01 2.10942224e-01 1.35342741e+00 5.16204774e-01 -9.80044305e-01 8.42865407e-01 5.70323825e-01 1.73146814e-01 -3.10572833e-01 -7.12662995e-01 -1.10532200e+00 6.90908015e-01 -2.06289917e-01 4.03134972e-01 1.53748786e+00 -2.80382723e-01 5.75137623e-02 -2.54376501e-01 2.62382001e-01 4.70288396e-01 -4.95883435e-01 7.79723942e-01 -9.11465168e-01 -3.59853208e-01 -4.41459000e-01 -7.89698303e-01 -1.21622317e-01 5.05813181e-01 -9.13909078e-01 9.71834734e-02 -7.13600159e-01 -3.35898429e-01 -2.68716156e-01 -1.34336218e-01 6.40011013e-01 -4.89424884e-01 6.71834171e-01 5.57583749e-01 1.52615048e-02 2.43725047e-01 4.56641428e-02 1.10652792e+00 -6.67200267e-01 2.06230640e-01 1.05423868e-01 -8.50030720e-01 9.17538702e-01 9.55239356e-01 -8.70954454e-01 -6.18340850e-01 -2.56725103e-01 8.63730833e-02 -4.34935540e-01 5.50502360e-01 -9.44103718e-01 -1.84195191e-01 -1.95903614e-01 7.18164384e-01 1.32509455e-01 1.52067304e-01 -9.62416291e-01 -1.50607631e-01 9.40599680e-01 -6.31431937e-01 5.69615252e-02 6.49020553e-01 4.49637622e-01 -4.12175804e-02 -6.60011351e-01 1.20465946e+00 -1.77426070e-01 -3.57419401e-01 -3.04097198e-02 -7.52787232e-01 1.08391903e-01 1.37158668e+00 -2.95293301e-01 -7.33677924e-01 -2.76471645e-01 -6.78539813e-01 3.71418744e-02 6.78418219e-01 6.36869371e-01 4.16840434e-01 -1.05654347e+00 -5.96274734e-01 2.23895833e-01 -2.74129715e-02 -4.44244385e-01 3.05817765e-03 6.22047670e-02 -5.13937831e-01 1.60528824e-01 -5.91448247e-01 -3.50126177e-01 -2.00290465e+00 1.04717636e+00 6.19136393e-01 -1.45038486e-01 -1.76933959e-01 9.09474909e-01 4.97594655e-01 -5.69512308e-01 3.15520048e-01 -7.51312030e-03 -1.13238171e-01 -9.90646854e-02 4.17854816e-01 1.08977243e-01 1.12496957e-01 -2.81794101e-01 -1.71344474e-01 5.71498647e-02 -6.91548467e-01 2.16011003e-01 8.88945222e-01 4.60007995e-01 1.40944362e-01 2.55947322e-01 1.13049304e+00 -4.29970324e-02 -1.06759143e+00 3.61043178e-02 -1.92762285e-01 -5.16882539e-01 -2.12491497e-01 -1.04380155e+00 -9.44175899e-01 9.58735764e-01 7.02875912e-01 7.20600784e-01 1.13301075e+00 -3.69938225e-01 3.65834534e-01 7.65991032e-01 -1.30594492e-01 -7.35400677e-01 4.98634547e-01 4.96827751e-01 9.59842920e-01 -1.24554253e+00 -7.54172578e-02 -3.91849399e-01 -6.44129455e-01 1.01174724e+00 9.01083410e-01 -4.20971125e-01 2.94258207e-01 4.23392594e-01 3.29777390e-01 9.66599584e-02 -6.42114282e-01 4.04077947e-01 2.69853204e-01 8.69849801e-01 1.67181745e-01 -7.14948326e-02 8.39291513e-02 1.06511988e-01 -4.47302639e-01 -4.46068048e-01 8.48678768e-01 1.22806597e+00 -2.26253793e-01 -1.11600077e+00 -8.49581420e-01 5.46491683e-01 -7.61061609e-01 -1.70778796e-01 -1.28001225e+00 1.08161497e+00 1.51399255e-01 7.50534475e-01 1.19037798e-03 -6.81477785e-01 4.05293494e-01 1.28207341e-01 4.38596904e-01 -5.74328780e-01 -1.36658561e+00 -7.79253483e-01 2.13534936e-01 -6.45250604e-02 1.72817949e-02 -4.20035034e-01 -8.25085700e-01 -6.02273703e-01 -4.05072421e-01 -2.98759597e-03 6.71208084e-01 9.77035165e-01 7.83724859e-02 7.37830758e-01 8.03406239e-01 -6.02486670e-01 -1.01433718e+00 -8.43289316e-01 -2.03286543e-01 8.34216714e-01 5.10488987e-01 -3.52620900e-01 -8.19378078e-01 1.81346640e-01]
[5.8174052238464355, 8.012333869934082]
070d98c1-3465-4c4e-85e2-4e95f383f963
perusil-a-framework-to-build-a-continuous
null
null
https://aclanthology.org/2022.signlang-1.1
https://aclanthology.org/2022.signlang-1.1.pdf
PeruSIL: A Framework to Build a Continuous Peruvian Sign Language Interpretation Dataset
Video-based datasets for Continuous Sign Language are scarce due to the challenging task of recording videos from native signers and the reduced number of people who can annotate sign language. COVID-19 has evidenced the key role of sign language interpreters in delivering nationwide health messages to deaf communities. In this paper, we present a framework for creating a multi-modal sign language interpretation dataset based on videos and we use it to create the first dataset for Peruvian Sign Language (LSP) interpretation annotated by hearing volunteers who have intermediate knowledge of PSL guided by the video audio. We rely on hearing people to produce a first version of the annotations, which should be reviewed by native signers in the future. Our contributions: i) we design a framework to annotate a sign Language dataset; ii) we release the first annotated LSP multi-modal interpretation dataset (AEC); iii) we evaluate the annotation done by hearing people by training a sign language recognition model. Our model reaches up to 80.3% of accuracy among a minimum of five classes (signs) AEC dataset, and 52.4% in a second dataset. Nevertheless, analysis by subject in the second dataset show variations worth to discuss.
['Pablo Rivas', 'Fernando Alva-Manchego', 'Francisco Cerna-Herrera', 'Joe Huamani-Malca', 'Gissella Bejarano']
null
null
null
null
signlang-lrec-2022-6
['sign-language-recognition']
['computer-vision']
[ 2.98400998e-01 2.57356673e-01 -2.30228975e-02 -4.60984617e-01 -1.24997818e+00 -7.72042811e-01 4.20808494e-01 -5.44720948e-01 -7.75325000e-01 5.18566370e-01 9.90391910e-01 -2.20204398e-01 -7.87630603e-02 -1.53221458e-01 -5.84555686e-01 -5.18940806e-01 3.26562702e-04 5.88677108e-01 5.71853936e-01 -1.33227438e-01 9.48628634e-02 3.46239477e-01 -1.87944901e+00 8.49638641e-01 7.00564623e-01 5.37071943e-01 -4.54819500e-02 1.04892063e+00 -3.87926884e-02 1.11315835e+00 -5.49665391e-01 -2.28349447e-01 3.06427479e-01 -6.84842229e-01 -1.05614424e+00 3.84289585e-02 1.12602293e+00 -1.03236306e+00 8.11043605e-02 6.23225689e-01 1.15976048e+00 -3.12136084e-01 6.50165677e-01 -1.30999231e+00 -6.29693031e-01 7.72944689e-01 6.83223382e-02 -3.31785500e-01 8.40250731e-01 5.53542376e-01 9.71552372e-01 -6.89551651e-01 1.28895712e+00 1.12216830e+00 7.86989510e-01 1.10006630e+00 -5.67665517e-01 -6.56962156e-01 1.46341592e-01 5.55894673e-01 -1.08558142e+00 -5.41908085e-01 4.47624266e-01 -7.33851373e-01 7.88835526e-01 2.77671665e-01 1.18184721e+00 1.26013434e+00 -9.86003458e-01 1.19692743e+00 1.27492118e+00 -8.37414145e-01 5.30042797e-02 -3.28713179e-01 2.99721926e-01 4.74957854e-01 6.24808297e-02 -1.23553194e-01 -7.73392200e-01 -3.55371609e-02 5.14227927e-01 -7.02745497e-01 -5.63374221e-01 -1.01725027e-01 -1.24215949e+00 4.08166468e-01 -2.06583962e-02 5.39111018e-01 -3.93626630e-01 2.04415143e-01 6.47363186e-01 4.77792859e-01 -2.50296950e-01 -9.39661935e-02 -3.31233382e-01 -7.71382749e-01 -9.22234118e-01 9.42916423e-02 9.58851516e-01 8.82040143e-01 -1.40377969e-01 -1.82896107e-01 -2.20627204e-01 9.42159116e-01 6.84678435e-01 7.82582581e-01 4.54095364e-01 -1.02357578e+00 3.49878699e-01 4.46184039e-01 7.40824789e-02 -1.42714694e-01 -3.07646394e-01 5.08741677e-01 -1.52184188e-01 6.19108021e-01 1.07365274e+00 -2.31483020e-02 -1.30577600e+00 1.37208724e+00 -1.93619430e-01 -2.20792159e-01 6.71059638e-03 1.12442839e+00 1.38622177e+00 -4.55071079e-03 2.96498805e-01 1.11202873e-01 1.36907721e+00 -7.12565362e-01 -6.78449333e-01 3.21840405e-01 7.18622088e-01 -6.59247816e-01 1.25730431e+00 7.73363352e-01 -1.08119667e+00 -1.35911003e-01 -4.49813604e-01 -1.14078410e-01 -1.84030175e-01 4.52135265e-01 5.06863713e-01 9.33931172e-01 -1.45494223e+00 -2.11450443e-01 -8.39806020e-01 -8.20826948e-01 5.40918112e-01 3.52904975e-01 -6.08488083e-01 1.37495121e-03 -8.37257922e-01 8.53062868e-01 1.90267280e-01 3.21738183e-01 -7.50635147e-01 -3.92225623e-01 -7.87514806e-01 -6.71156347e-01 -8.47242249e-04 -2.10163251e-01 1.48245263e+00 -1.23587704e+00 -1.48689103e+00 1.56858742e+00 -2.40794882e-01 -1.68781921e-01 1.38145912e+00 -2.93551803e-01 -4.72397089e-01 5.38254440e-01 1.82388853e-02 7.93201745e-01 5.00456810e-01 -1.25721014e+00 -7.66211390e-01 -4.32844386e-02 4.96517308e-02 -6.61430359e-02 6.95521310e-02 6.24781787e-01 -5.89514613e-01 -4.49653924e-01 8.98994952e-02 -7.00472653e-01 3.25354457e-01 5.06238461e-01 -1.56727448e-01 -2.82632470e-01 7.05386996e-01 -1.37123811e+00 1.05484688e+00 -2.08096266e+00 -2.28884933e-03 2.64454544e-01 1.83459923e-01 6.88667357e-01 -4.12097037e-01 3.92308235e-01 2.34667554e-01 6.26385808e-02 -4.25536782e-01 -1.68085888e-01 2.45683581e-01 4.87911403e-01 -1.81299657e-01 2.89456725e-01 -7.07901716e-02 7.78776109e-01 -1.05865955e+00 -6.78868711e-01 2.14133456e-01 6.78005159e-01 -7.13167369e-01 -2.13371739e-02 2.29252372e-02 8.05897355e-01 -1.45025417e-01 1.26690888e+00 5.87449789e-01 1.45546392e-01 1.62116632e-01 -1.33941337e-01 -2.50269532e-01 -1.51495695e-01 -1.27693582e+00 1.44339216e+00 -3.20963860e-01 1.08540297e+00 3.15517306e-01 -6.02327526e-01 5.34998417e-01 7.83755302e-01 4.65236545e-01 -3.93598467e-01 1.24379709e-01 8.92379284e-01 1.38869211e-01 -1.24515629e+00 6.44507483e-02 4.81329672e-02 1.53911874e-01 4.82402474e-01 1.75193295e-01 -2.18371868e-01 5.67891121e-01 -5.10835908e-02 1.00495875e+00 2.54312158e-01 2.04076529e-01 3.53787422e-01 8.21238935e-01 -2.58063432e-02 1.90708622e-01 9.44120586e-01 -7.70614624e-01 8.20965588e-01 4.40801293e-01 -4.06536102e-01 -8.82058144e-01 -1.09816587e+00 -2.28003755e-01 1.10003185e+00 -4.58324164e-01 -7.99840987e-02 -6.49160147e-01 -6.34830654e-01 -1.81159362e-01 2.84852743e-01 -2.85493433e-01 6.81919575e-01 -8.39980185e-01 -9.46288705e-02 1.21640086e+00 8.54475498e-01 7.36904383e-01 -1.61033356e+00 -7.00892925e-01 -5.35936691e-02 -3.46840948e-01 -1.16654754e+00 -2.98551917e-01 -5.78238368e-01 -2.40790024e-01 -1.51698267e+00 -1.34862220e+00 -1.33042872e+00 7.28290141e-01 -5.99298477e-01 5.28988421e-01 2.01413840e-01 -2.15750575e-01 1.11910367e+00 -8.36514056e-01 -4.70577180e-01 -7.79187858e-01 -2.99246132e-01 -1.22235566e-01 -6.63480535e-02 6.74423933e-01 -3.27381104e-01 -2.88047522e-01 3.69730681e-01 -8.46823752e-01 9.70840268e-03 6.09319806e-01 6.72546029e-01 2.69824892e-01 -1.11622047e+00 2.85220295e-01 -3.70266855e-01 4.67259347e-01 2.54500598e-01 -4.95812058e-01 5.58665872e-01 1.93972290e-01 -2.08436549e-01 -6.80484921e-02 -6.18844569e-01 -9.66340184e-01 3.23669136e-01 -5.86692154e-01 2.16267779e-01 -4.38864827e-01 1.91320896e-01 -5.65232970e-02 -3.26493353e-01 5.63943386e-01 1.57835290e-01 -1.28382081e-02 -5.26337504e-01 4.27819848e-01 1.36412215e+00 9.37248647e-01 -3.64247024e-01 3.78699452e-01 6.21171892e-01 -3.37244481e-01 -1.26672804e+00 -4.02837336e-01 -7.79489398e-01 -1.09763920e+00 -8.89152408e-01 9.29631174e-01 -7.30816066e-01 -9.70119774e-01 1.22168016e+00 -1.17794836e+00 -6.75480783e-01 -3.44618320e-01 8.16204131e-01 -8.45308602e-01 6.76670730e-01 -2.93113470e-01 -1.00304389e+00 -1.49136916e-01 -1.00498533e+00 1.34034026e+00 -1.17103182e-01 -4.30209428e-01 -6.28484190e-01 1.56278074e-01 7.29440987e-01 3.70601028e-01 4.15785879e-01 2.91991293e-01 -4.46692973e-01 -4.73413348e-01 -3.67104322e-01 -4.57879215e-01 6.18501008e-01 -1.61353245e-01 -3.46399695e-02 -1.15740037e+00 -1.00676920e-02 -7.86401272e-01 -6.92316830e-01 8.67482007e-01 5.57619154e-01 6.31386220e-01 -5.66705018e-02 1.47852346e-01 4.09261763e-01 8.96951139e-01 8.07708800e-02 7.98120916e-01 2.02021316e-01 6.01671338e-01 5.75516284e-01 1.73494086e-01 1.79418728e-01 4.64566618e-01 6.69751167e-01 -7.36934990e-02 7.17312247e-02 -7.67131984e-01 -4.15901661e-01 7.17506468e-01 8.57852221e-01 -9.40985322e-01 -4.60176617e-02 -1.17709732e+00 8.95335734e-01 -1.75460696e+00 -1.12912667e+00 -4.83995527e-01 1.91484559e+00 6.76260471e-01 -4.76848811e-01 5.83217561e-01 5.67190766e-01 4.45799738e-01 -2.85094261e-01 -1.00238785e-01 -2.84456849e-01 -5.10106504e-01 1.70216233e-01 4.06648278e-01 7.28849351e-01 -1.09973145e+00 9.03863907e-01 6.66238260e+00 2.42712125e-01 -1.29038084e+00 2.76349515e-01 -4.25221264e-01 9.15149003e-02 1.90604497e-02 -1.95041478e-01 -7.47909844e-01 2.96835303e-01 5.85751295e-01 2.96797991e-01 1.54775113e-01 5.80119550e-01 5.89113057e-01 -1.73775688e-01 -9.25610840e-01 1.03082383e+00 5.85841477e-01 -9.47662830e-01 2.69296706e-01 -5.89794926e-02 5.32319009e-01 4.36266005e-01 -6.31238043e-01 2.00188547e-01 1.39198393e-01 -7.86377430e-01 1.05050635e+00 7.90792227e-01 1.33627498e+00 1.89066067e-01 8.56491208e-01 -9.03665274e-03 -1.30508864e+00 -2.12116033e-01 4.38208580e-01 -2.25966647e-02 8.91532958e-01 -2.73680866e-01 -7.48816490e-01 -8.22358578e-03 7.08302081e-01 6.80097520e-01 -6.63444579e-01 1.58302200e+00 -7.94562876e-01 8.81998301e-01 -5.21457076e-01 -2.17781544e-01 1.57368600e-01 2.47736514e-01 6.44551277e-01 1.43819416e+00 2.95882136e-01 -2.86836568e-02 6.23644292e-02 2.79618233e-01 3.10447633e-01 3.56920749e-01 -3.72226924e-01 -9.20943320e-02 5.45171015e-02 4.53517258e-01 -4.11458760e-01 -3.59074295e-01 -6.13816679e-01 1.08540893e+00 -2.31718197e-01 4.44750786e-01 -4.06667531e-01 -3.61993253e-01 2.25053504e-01 1.99971825e-01 9.60729346e-02 -8.73618275e-02 -1.85588319e-02 -1.15020537e+00 5.39916873e-01 -8.42353761e-01 6.11580849e-01 -1.00385821e+00 -1.17760491e+00 3.84667963e-01 1.25425503e-01 -1.57109010e+00 -5.78647673e-01 -1.13780642e+00 -3.22274804e-01 6.47691548e-01 -1.58536386e+00 -1.89898336e+00 -7.31267333e-01 6.44944370e-01 3.02021056e-01 -1.55112907e-01 8.82260084e-01 6.15239203e-01 2.65730619e-01 6.16034925e-01 -3.48599195e-01 7.84096181e-01 7.26145506e-01 -9.99033213e-01 -2.53727436e-01 6.73062682e-01 1.06481807e-02 1.64673477e-01 3.00334096e-01 -5.88867188e-01 -7.37108886e-01 -5.68986475e-01 1.56708205e+00 -7.82720923e-01 6.43915176e-01 5.29494695e-02 -4.07813132e-01 8.78201604e-01 -1.79787770e-01 -1.76486317e-02 5.90806961e-01 -3.98629904e-01 -3.45332056e-01 3.91406208e-01 -1.29567444e+00 4.59778577e-01 1.55052936e+00 -8.57378483e-01 -9.22997952e-01 2.91823894e-01 7.59747326e-02 -4.43727612e-01 -6.63662910e-01 1.39191166e-01 1.45016730e+00 -7.87209451e-01 7.04537928e-01 -5.94447494e-01 1.73641279e-01 -5.21224320e-01 -1.26373738e-01 -6.04176044e-01 6.19251192e-01 -5.59916139e-01 2.05245852e-01 1.10632992e+00 4.25587922e-01 -6.97665215e-01 5.62254667e-01 6.52740836e-01 -1.47352964e-01 4.69072759e-02 -1.26384950e+00 -8.94703805e-01 -2.04892382e-01 -9.90337074e-01 1.92308247e-01 5.84565997e-01 1.16528518e-01 -6.64535999e-01 -3.82631421e-01 -5.49001358e-02 4.92577523e-01 -4.40053016e-01 1.06471467e+00 -1.19263089e+00 -2.72093385e-01 -5.78357100e-01 -9.74932671e-01 -8.19688082e-01 2.15290543e-02 -9.92809534e-01 1.42285272e-01 -1.88481534e+00 2.56167222e-02 1.21197507e-01 3.00039440e-01 9.35980797e-01 4.09341514e-01 5.88977218e-01 3.83309603e-01 2.90777862e-01 -5.31364143e-01 -1.37881383e-01 1.23170435e+00 -6.69154972e-02 -3.12308073e-01 1.13693438e-01 -1.53316438e-01 9.90307212e-01 4.11691487e-01 5.45087904e-02 2.30113849e-01 -6.42376661e-01 4.95249741e-02 -3.43940973e-01 8.62570882e-01 -9.72199619e-01 1.70428202e-01 2.19101280e-01 -2.11384863e-01 -6.62881017e-01 7.87047595e-02 -6.95005119e-01 -1.27294913e-01 7.26070464e-01 -2.94466823e-01 -5.51764369e-01 -5.77075519e-02 -3.30957323e-02 -4.80702192e-01 -3.28897506e-01 6.62995994e-01 -1.63438216e-01 -1.24729013e+00 -1.95135847e-01 -7.22263157e-01 2.72650391e-01 6.38325751e-01 -5.53049088e-01 -2.57704526e-01 -7.83011734e-01 -1.14640093e+00 2.41302714e-01 2.44206071e-01 3.96003038e-01 4.56877619e-01 -1.19977868e+00 -8.86904836e-01 3.88369709e-01 5.29926598e-01 -2.65641183e-01 2.41512671e-01 1.03819418e+00 -1.18085253e+00 3.25997382e-01 -4.84212041e-01 -5.75053096e-01 -1.81620109e+00 -5.37370861e-01 2.67901421e-01 2.12402120e-01 -8.20097327e-01 8.38474154e-01 -7.40332842e-01 -5.79725385e-01 7.31270850e-01 -8.40051889e-01 -3.97828579e-01 3.41041625e-01 9.02220488e-01 4.86306161e-01 -2.32888252e-01 -1.12238026e+00 -4.62274760e-01 9.54792678e-01 4.39312071e-01 -6.55706286e-01 1.37665439e+00 2.39613354e-01 -9.19464161e-04 3.50035429e-01 8.25159371e-01 2.25115240e-01 -9.44305599e-01 6.36983290e-03 1.06760338e-01 -3.96506637e-01 -3.48508269e-01 -1.26710749e+00 -6.30735755e-01 6.75718367e-01 9.10545290e-01 -3.41450900e-01 1.19999433e+00 3.53280783e-01 6.06574714e-01 4.21315461e-01 4.34754908e-01 -1.29293585e+00 -2.68104017e-01 5.45990944e-01 1.32637155e+00 -1.28428268e+00 -3.85765314e-01 -2.01021031e-01 -7.58795142e-01 1.13049662e+00 1.17737636e-01 2.43176490e-01 5.77734351e-01 1.20460279e-01 8.66062522e-01 -2.00396895e-01 1.15407698e-01 -6.74341738e-01 5.66474974e-01 1.19742644e+00 5.59547544e-01 2.23492846e-01 -9.67914104e-01 6.23499155e-01 -3.09561312e-01 7.38056958e-01 5.90434730e-01 1.03507268e+00 -1.36079207e-01 -1.31000793e+00 -4.30439919e-01 2.20169410e-01 -1.01082265e-01 1.25094697e-01 -8.37953746e-01 1.02467382e+00 4.46704894e-01 9.27630424e-01 -2.62177587e-01 2.49905601e-01 7.62800872e-01 3.70014727e-01 6.47291839e-01 -2.98682004e-01 -6.12543344e-01 -2.30511501e-01 5.11402071e-01 -4.34503436e-01 -1.00300419e+00 -1.09369791e+00 -1.24622512e+00 3.43919545e-01 1.60949290e-01 -4.15930063e-01 6.36670649e-01 9.95592356e-01 -2.98239619e-01 -3.29625085e-02 -2.84914851e-01 -1.02386391e+00 -3.16700280e-01 -1.09543264e+00 -5.33941388e-01 7.66869485e-01 4.62513387e-01 -2.94514567e-01 -5.84280789e-01 5.98292172e-01]
[9.134204864501953, -6.443332195281982]
83750966-f6b0-47ad-90f4-7f4219f60ab9
unsupervised-kinematic-motion-detection-for
2206.08497
null
https://arxiv.org/abs/2206.08497v1
https://arxiv.org/pdf/2206.08497v1.pdf
Unsupervised Kinematic Motion Detection for Part-segmented 3D Shape Collections
3D models of manufactured objects are important for populating virtual worlds and for synthetic data generation for vision and robotics. To be most useful, such objects should be articulated: their parts should move when interacted with. While articulated object datasets exist, creating them is labor-intensive. Learning-based prediction of part motions can help, but all existing methods require annotated training data. In this paper, we present an unsupervised approach for discovering articulated motions in a part-segmented 3D shape collection. Our approach is based on a concept we call category closure: any valid articulation of an object's parts should keep the object in the same semantic category (e.g. a chair stays a chair). We operationalize this concept with an algorithm that optimizes a shape's part motion parameters such that it can transform into other shapes in the collection. We evaluate our approach by using it to re-discover part motions from the PartNet-Mobility dataset. For almost all shape categories, our method's predicted motion parameters have low error with respect to ground truth annotations, outperforming two supervised motion prediction methods.
['Daniel Ritchie', 'Srinath Sridhar', 'Yifan Ruan', 'Xianghao Xu']
2022-06-17
null
null
null
null
['motion-detection']
['computer-vision']
[ 1.42426148e-01 6.03634179e-01 -2.16258317e-01 -1.92972973e-01 -2.56601244e-01 -7.40037322e-01 7.25102425e-01 1.50958942e-02 1.41327992e-01 3.01989645e-01 2.44306490e-01 -4.09466960e-02 -1.19846612e-01 -6.93964779e-01 -9.34164166e-01 -4.32371944e-01 3.77989113e-02 1.21931016e+00 8.22003722e-01 -2.17004627e-01 1.64314687e-01 9.28119361e-01 -1.83433962e+00 2.49689952e-01 4.97715503e-01 7.16492534e-01 4.91634279e-01 5.14507592e-01 -4.85003471e-01 4.48099315e-01 -3.80401582e-01 -4.25082594e-01 3.14090133e-01 -2.31611863e-01 -1.33402610e+00 6.77774489e-01 3.08194995e-01 3.89387049e-02 6.02464974e-02 6.73532009e-01 3.45025621e-02 3.12673181e-01 9.17371750e-01 -1.43637526e+00 -1.54508233e-01 6.13807142e-01 -7.18116984e-02 -5.71183383e-01 4.42462802e-01 2.70256586e-02 9.86384928e-01 -7.82814264e-01 1.39211452e+00 1.33982372e+00 6.01033449e-01 8.85646939e-01 -1.36565900e+00 1.18572049e-01 3.02501768e-01 2.69507114e-02 -1.20257163e+00 -3.97589862e-01 1.09549248e+00 -6.91162944e-01 7.39939332e-01 3.26775938e-01 1.05280495e+00 8.68821621e-01 -1.47897989e-01 1.22545218e+00 2.29418799e-01 -3.06243956e-01 4.79516745e-01 5.78349903e-02 -2.91495144e-01 7.46099293e-01 2.52660751e-01 -5.50583541e-01 -2.77173996e-01 -1.27931938e-01 8.48269880e-01 -1.34135187e-01 3.63694038e-03 -1.35953259e+00 -1.38674664e+00 4.75688785e-01 3.84557068e-01 3.96050006e-01 -4.15646732e-01 4.74137336e-01 1.67500243e-01 -1.60955325e-01 3.73622000e-01 4.34267670e-01 -8.29816759e-01 -1.67835087e-01 -8.16867173e-01 5.80492318e-01 9.88710940e-01 1.22616208e+00 5.92605054e-01 -1.54178843e-01 1.44368246e-01 6.67281568e-01 4.61432576e-01 2.22675249e-01 2.76737124e-01 -1.27874589e+00 1.95547611e-01 9.86253321e-01 4.17851090e-01 -6.59347236e-01 -5.48555017e-01 1.04182430e-01 -3.94325405e-01 9.98458043e-02 3.21234018e-01 4.19440895e-01 -1.01740336e+00 1.28610933e+00 7.63979793e-01 1.76102761e-02 -1.00100532e-01 9.14492369e-01 7.57943511e-01 2.97902048e-01 -2.12842803e-02 9.74316448e-02 1.19908488e+00 -9.72808003e-01 -4.03466851e-01 -6.31678104e-02 7.09904015e-01 -6.78978086e-01 8.63783598e-01 3.44017774e-01 -1.12795043e+00 -5.94242096e-01 -6.83550000e-01 6.67318925e-02 -7.53024593e-02 2.70375051e-02 6.37282729e-01 5.20214081e-01 -8.45819831e-01 9.98922825e-01 -1.10249186e+00 -5.21187067e-01 5.49047172e-01 3.84456962e-01 -5.84367990e-01 2.30049506e-01 -1.60922989e-01 8.27048182e-01 4.88054395e-01 -1.68414339e-01 -9.25544202e-01 -3.47188562e-01 -9.11755741e-01 -3.56829584e-01 3.47432017e-01 -9.58436310e-01 1.49295199e+00 -9.68155682e-01 -1.36985719e+00 1.12180912e+00 -1.59657300e-01 -4.72340882e-01 7.58106291e-01 -2.47636050e-01 -9.99250039e-02 9.44835991e-02 1.07189916e-01 1.10463583e+00 8.63374233e-01 -1.83945155e+00 -5.02518058e-01 -3.03416997e-01 -4.87637110e-02 7.71708488e-02 1.28436863e-01 -4.74451691e-01 -7.96089888e-01 -5.08227110e-01 7.42833555e-01 -1.23785067e+00 -4.09717023e-01 4.21769291e-01 -6.63549602e-01 -4.01736766e-01 1.14761519e+00 -4.78502810e-01 6.76350713e-01 -1.75741863e+00 3.46716583e-01 1.83002159e-01 1.13526890e-02 7.42516145e-02 -2.44160555e-02 3.39214981e-01 3.59782130e-01 1.59974799e-01 -4.61623877e-01 -7.09907830e-01 8.54410231e-02 5.38890839e-01 -1.47514552e-01 4.16290939e-01 1.44440413e-01 9.43523884e-01 -8.68842840e-01 -6.19346559e-01 3.38832289e-01 3.34600240e-01 -8.21403205e-01 8.77485424e-02 -8.33130956e-01 6.24576449e-01 -5.46299279e-01 7.36267209e-01 4.87144023e-01 -1.54327497e-01 2.19225928e-01 -2.15635002e-01 -3.80647294e-02 1.53028920e-01 -1.30258489e+00 2.03099608e+00 -2.49099538e-01 3.75434160e-01 -3.66029609e-03 -8.93800318e-01 1.10123253e+00 2.78704137e-01 9.67326462e-01 6.75954595e-02 -3.39604951e-02 3.60293537e-01 -1.19487971e-01 -6.83645606e-01 6.69813395e-01 -7.70741254e-02 3.52430157e-02 2.98732847e-01 -1.85657755e-01 -8.60319257e-01 -5.32438494e-02 -9.39583555e-02 8.50278139e-01 9.15038526e-01 1.85319602e-01 -1.29505545e-01 3.53452295e-01 4.58844572e-01 3.59021306e-01 4.93408948e-01 -5.59900142e-02 7.75185466e-01 1.07203618e-01 -5.62928081e-01 -1.31979454e+00 -1.11023927e+00 2.83508506e-02 6.25244975e-01 4.99455333e-01 -1.96898833e-01 -6.80277467e-01 -7.27987766e-01 1.43311962e-01 6.72471285e-01 -5.28496981e-01 3.98089997e-02 -8.20693195e-01 1.12165056e-01 9.31636021e-02 5.60989201e-01 1.33290276e-01 -1.28150022e+00 -1.12011433e+00 3.53761941e-01 -2.64361352e-01 -1.07012045e+00 -2.99883991e-01 -1.42609745e-01 -1.13075984e+00 -1.27047288e+00 -6.55191302e-01 -9.72945273e-01 8.79284322e-01 2.46364519e-01 1.17951894e+00 1.29641548e-01 -1.98033959e-01 6.94426537e-01 -5.70792139e-01 -5.13041377e-01 -7.22820103e-01 2.11160295e-02 -1.25052808e-02 -1.32820383e-02 -1.44539341e-01 -6.18017435e-01 -4.95789409e-01 5.74129164e-01 -8.26554656e-01 3.60998958e-01 4.60472316e-01 2.71732956e-01 1.03115320e+00 -1.90042347e-01 -1.97342765e-02 -5.58722615e-01 2.50483900e-02 -2.95116812e-01 -1.85672000e-01 1.27471820e-01 4.82478505e-03 3.42270166e-01 2.93973655e-01 -6.54606402e-01 -8.59854817e-01 7.23214924e-01 -2.99422611e-02 -7.88703144e-01 -3.37930650e-01 5.43520376e-02 -4.63299155e-01 3.30937803e-01 4.59369004e-01 1.30130187e-01 -7.62697235e-02 -8.27520549e-01 6.04666114e-01 2.33226329e-01 5.17539084e-01 -6.42841578e-01 7.78369188e-01 9.63550270e-01 1.93999112e-01 -1.01835203e+00 -4.28830624e-01 -7.07291722e-01 -1.17607296e+00 -5.09705901e-01 1.03069246e+00 -4.24157053e-01 -7.16639757e-01 1.66257709e-01 -1.40468037e+00 -4.54570651e-01 -6.08519256e-01 3.43894035e-01 -1.01958084e+00 3.08592796e-01 -7.70305097e-02 -8.52743804e-01 -6.09660037e-02 -1.07040727e+00 1.55845678e+00 -1.59155905e-01 -5.85681617e-01 -7.41045415e-01 1.17078707e-01 3.35058898e-01 -1.90085024e-01 6.15440309e-01 7.92479932e-01 -6.42857850e-01 -6.59175992e-01 -1.58067480e-01 4.72011626e-01 -6.22215169e-03 1.56960294e-01 2.89067417e-01 -6.45429254e-01 5.33425361e-02 -3.76364648e-01 -1.58705171e-02 4.32208419e-01 3.73695701e-01 1.26916420e+00 -4.50222164e-01 -7.61429012e-01 2.23190576e-01 1.09903038e+00 2.18740433e-01 4.25617754e-01 2.76643753e-01 9.09149051e-01 9.15143907e-01 7.15381682e-01 4.45920169e-01 3.98407459e-01 1.14889097e+00 8.26003253e-01 4.11966890e-01 -1.63688734e-01 -4.85380054e-01 6.82698563e-02 6.01393104e-01 -3.34601969e-01 -1.20504126e-01 -1.07442784e+00 7.69979298e-01 -1.93324125e+00 -9.13940549e-01 -6.17759407e-01 2.18646908e+00 4.20144379e-01 2.11558804e-01 3.14224839e-01 2.40830719e-01 4.82671112e-01 -1.54972151e-01 -5.35273075e-01 -2.11895689e-01 1.32902443e-01 6.84967563e-02 1.79395363e-01 3.18340570e-01 -1.14782798e+00 1.02046502e+00 5.42745543e+00 3.56034875e-01 -6.27108872e-01 -4.64525223e-02 5.80372848e-02 1.69148579e-01 -6.23820841e-01 2.60574073e-01 -5.58176875e-01 1.17546506e-01 4.18605298e-01 1.59136325e-01 4.68791649e-03 1.15393555e+00 2.97487259e-01 -1.49957374e-01 -1.34820390e+00 8.74064326e-01 -9.84182283e-02 -1.44412851e+00 3.08648020e-01 1.23885632e-01 7.92871058e-01 -3.60794216e-01 -5.16741395e-01 1.25246197e-02 2.71502137e-01 -8.16645205e-01 1.57494462e+00 7.48041034e-01 2.46505767e-01 -5.89586139e-01 3.30727220e-01 7.68198252e-01 -1.50245714e+00 2.34794453e-01 -1.83976248e-01 2.07570538e-01 4.30016011e-01 3.43254000e-01 -1.14699173e+00 5.98807395e-01 4.22221184e-01 5.71856856e-01 -3.78302783e-01 1.18759596e+00 -1.58235416e-01 2.96131074e-01 -2.66266584e-01 -1.73818499e-01 4.70899567e-02 -2.48727366e-01 8.87708068e-01 7.82570243e-01 2.21571714e-01 -7.74157569e-02 5.05670846e-01 8.03257167e-01 -1.07334428e-01 -4.97931167e-02 -7.90826380e-01 7.67154470e-02 5.87336540e-01 9.45769966e-01 -1.33074927e+00 -3.71047020e-01 7.46938437e-02 1.21829426e+00 -1.47186697e-03 -1.19263262e-01 -8.03030610e-01 2.80091967e-02 6.39221251e-01 4.56931293e-01 6.06790602e-01 -4.67985183e-01 -2.52636373e-01 -7.22229600e-01 2.19148993e-01 -3.83238137e-01 -9.81548876e-02 -9.28941846e-01 -6.96861207e-01 4.17381525e-01 1.73270836e-01 -1.65458536e+00 -2.89390296e-01 -4.87494648e-01 -3.15521836e-01 1.71549916e-01 -6.76268637e-01 -1.46033525e+00 -3.35189611e-01 3.99892002e-01 9.77680981e-01 2.23345160e-01 6.60832584e-01 -1.40281186e-01 5.66323102e-02 -1.98202673e-02 -3.79432023e-01 -2.07182649e-03 1.68425683e-02 -1.12404418e+00 6.54203653e-01 6.20161951e-01 6.63011551e-01 2.48846889e-01 9.15194094e-01 -9.30014133e-01 -1.58276618e+00 -1.21973395e+00 7.03272820e-01 -8.88296306e-01 4.11565989e-01 -3.70429993e-01 -7.42203355e-01 7.91944146e-01 -4.66892749e-01 1.67100150e-02 1.63376018e-01 -2.49748647e-01 -1.23024449e-01 4.81123090e-01 -1.10193133e+00 6.91984594e-01 1.82925820e+00 -1.52442768e-01 -6.82734370e-01 3.90370041e-01 7.80164957e-01 -6.00551724e-01 -8.30208361e-01 5.01463413e-01 6.06165290e-01 -7.92829216e-01 1.10443544e+00 -8.82164717e-01 3.79055679e-01 -5.31017542e-01 -1.92874521e-01 -1.08046234e+00 -7.06041381e-02 -6.20143712e-01 -5.75640559e-01 8.87526691e-01 1.16040975e-01 -3.36659066e-02 1.26893544e+00 5.04422009e-01 -5.17827868e-01 -8.28357697e-01 -9.36264932e-01 -1.16536677e+00 -3.04065138e-01 -6.22338891e-01 6.63998723e-01 7.24089622e-01 -3.47789854e-01 1.74861320e-03 1.41519615e-02 4.49099056e-02 4.72652227e-01 3.84317994e-01 1.25554717e+00 -1.58965194e+00 -1.04101546e-01 -6.32075727e-01 -7.68954158e-01 -1.19982338e+00 3.13930452e-01 -1.01555824e+00 3.41211319e-01 -2.11281371e+00 -1.27629474e-01 -6.97580099e-01 7.22610652e-01 5.91496766e-01 3.39758217e-01 -4.31028381e-02 2.76356488e-01 5.56550801e-01 -5.97449064e-01 6.34013355e-01 1.49207556e+00 -3.09664667e-01 -5.33933640e-01 3.58462691e-01 -5.51822037e-02 1.07722330e+00 6.59864366e-01 -4.53910798e-01 -3.73992085e-01 -3.39368343e-01 -1.10675059e-01 6.33872449e-02 5.23621202e-01 -1.07656467e+00 -6.36504497e-03 -2.80629396e-01 9.84948352e-02 -8.60282838e-01 7.12946177e-01 -1.08292592e+00 7.99800038e-01 8.11711073e-01 -8.04564133e-02 -1.69295873e-02 -1.55650976e-03 6.01739883e-01 1.59439221e-01 -3.42483103e-01 4.90793318e-01 -3.74775797e-01 -7.94730127e-01 3.29039514e-01 -2.74400860e-01 -1.43276826e-01 1.22349477e+00 -7.64857411e-01 2.54022866e-01 -8.81866589e-02 -1.03207290e+00 3.44308726e-02 8.75103712e-01 7.60416508e-01 7.30253756e-01 -1.34476113e+00 -4.92543578e-01 -1.38374880e-01 2.62058407e-01 5.43360233e-01 -2.94085853e-02 6.65568233e-01 -7.07987487e-01 2.96431601e-01 3.44792679e-02 -9.15569782e-01 -1.24362493e+00 4.06880260e-01 3.03292155e-01 1.11212023e-01 -8.88493836e-01 7.80630529e-01 -8.26643333e-02 -7.44429350e-01 7.12562799e-02 -7.00630486e-01 -1.03213049e-01 -1.14272550e-01 -5.76438978e-02 4.67384249e-01 1.95046291e-01 -9.46700633e-01 -4.51084644e-01 5.59284091e-01 4.92890984e-01 -2.48434231e-01 1.41186857e+00 2.42717519e-01 3.94020602e-02 6.33626878e-01 1.00769031e+00 9.54512730e-02 -1.42136288e+00 1.35973379e-01 3.71674240e-01 -4.20599759e-01 -6.45237148e-01 -3.40900570e-01 -8.03140938e-01 4.12278295e-01 2.29660541e-01 2.77919620e-01 5.46340823e-01 7.54281104e-01 5.45564890e-01 5.40549099e-01 1.05231571e+00 -1.03888369e+00 3.43705326e-01 4.20543104e-01 1.13446128e+00 -7.98583984e-01 -5.35034984e-02 -6.50226057e-01 -6.85141921e-01 9.61180151e-01 4.40363795e-01 -5.13244197e-02 5.78428626e-01 -8.38147849e-02 -3.29189360e-01 -5.40881515e-01 -4.07189399e-01 -2.78341055e-01 6.30103946e-01 8.02748740e-01 4.10031714e-02 2.88547724e-01 -1.86459154e-01 3.36043686e-01 -4.98974800e-01 -3.61349195e-01 4.04384822e-01 1.30217612e+00 -6.47916913e-01 -1.14298809e+00 -3.86681139e-01 4.03166622e-01 1.73267692e-01 6.06114626e-01 -7.32401192e-01 9.64682281e-01 3.69993597e-01 6.81204498e-01 3.03010553e-01 -3.44603866e-01 6.32260561e-01 1.42683163e-01 7.77419388e-01 -7.96279430e-01 -1.69704422e-01 -7.51217157e-02 1.53898910e-01 -6.44234300e-01 -8.98026288e-01 -1.03373051e+00 -1.60968876e+00 1.28529668e-01 -2.23471105e-01 -1.64728961e-03 8.56314600e-01 9.28458512e-01 2.13332623e-01 2.29231492e-01 4.63234603e-01 -1.42282748e+00 -2.10165620e-01 -6.70537174e-01 -2.87620068e-01 8.55316639e-01 -3.94205488e-02 -8.89446914e-01 5.23004122e-02 6.26330793e-01]
[7.05206298828125, -1.644737958908081]
85844b81-fcd6-4555-896a-32e54229bda2
slue-new-benchmark-tasks-for-spoken-language
2111.10367
null
https://arxiv.org/abs/2111.10367v3
https://arxiv.org/pdf/2111.10367v3.pdf
SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech
Progress in speech processing has been facilitated by shared datasets and benchmarks. Historically these have focused on automatic speech recognition (ASR), speaker identification, or other lower-level tasks. Interest has been growing in higher-level spoken language understanding tasks, including using end-to-end models, but there are fewer annotated datasets for such tasks. At the same time, recent work shows the possibility of pre-training generic representations and then fine-tuning for several tasks using relatively little labeled data. We propose to create a suite of benchmark tasks for Spoken Language Understanding Evaluation (SLUE) consisting of limited-size labeled training sets and corresponding evaluation sets. This resource would allow the research community to track progress, evaluate pre-trained representations for higher-level tasks, and study open questions such as the utility of pipeline versus end-to-end approaches. We present the first phase of the SLUE benchmark suite, consisting of named entity recognition, sentiment analysis, and ASR on the corresponding datasets. We focus on naturally produced (not read or synthesized) speech, and freely available datasets. We provide new transcriptions and annotations on subsets of the VoxCeleb and VoxPopuli datasets, evaluation metrics and results for baseline models, and an open-source toolkit to reproduce the baselines and evaluate new models.
['Kyu J. Han', 'Karen Livescu', 'Yoav Artzi', 'Pablo Brusco', 'Felix Wu', 'Ankita Pasad', 'Suwon Shon']
2021-11-19
null
null
null
null
['speaker-identification']
['speech']
[ 3.77204061e-01 2.46778414e-01 3.07132676e-02 -9.79731798e-01 -1.54909956e+00 -6.40436411e-01 7.78197050e-01 5.98123334e-02 -6.89065635e-01 4.39496934e-01 8.78614485e-01 -3.53263050e-01 3.74939591e-01 -1.13548942e-01 -3.59286606e-01 -2.81201035e-01 1.05502352e-01 8.05707574e-01 1.65133942e-02 -3.16154897e-01 5.05486093e-02 1.13316312e-01 -1.42347336e+00 6.93605840e-01 4.83496398e-01 7.54475117e-01 -3.73357572e-02 1.13219142e+00 -2.09680766e-01 7.79025078e-01 -9.97164190e-01 -4.90416199e-01 -3.97974670e-01 -3.63048345e-01 -1.14488971e+00 1.41818067e-02 4.54766095e-01 -1.26104102e-01 -2.10815310e-01 4.81994808e-01 1.00494218e+00 4.84268874e-01 4.83849198e-01 -1.12018216e+00 -7.47877777e-01 9.37208176e-01 6.16930313e-02 2.98725277e-01 5.48726320e-01 1.07163846e-01 1.10329628e+00 -1.07390547e+00 5.17971337e-01 1.39860523e+00 3.84294480e-01 1.05000663e+00 -1.09303284e+00 -4.89869922e-01 3.01323924e-02 3.82103100e-02 -1.22856653e+00 -1.56098974e+00 4.57627863e-01 -9.68021080e-02 1.56432259e+00 4.34286892e-01 -1.23786591e-01 1.69277704e+00 -7.30703712e-01 1.25590658e+00 8.87561917e-01 -4.54952031e-01 2.50902683e-01 3.50580066e-01 3.35731566e-01 4.71423715e-01 -4.46407735e-01 -4.96046431e-02 -9.71211791e-01 6.21192493e-02 1.50155693e-01 -7.09910154e-01 -5.81747055e-01 8.57387707e-02 -1.50270426e+00 7.59844303e-01 -1.03778012e-01 3.36535215e-01 -1.02877818e-01 -2.78911501e-01 8.90018761e-01 4.09374505e-01 7.51928091e-01 3.99640024e-01 -9.38409388e-01 -6.90885127e-01 -1.07303691e+00 -2.23585777e-02 1.33429956e+00 8.36246490e-01 3.80772531e-01 3.14086169e-01 -4.20145512e-01 1.52909589e+00 3.23971540e-01 3.16126704e-01 8.15394223e-01 -7.72506297e-01 8.33083510e-01 1.05378360e-01 -2.27262631e-01 -1.40846521e-01 -3.40588033e-01 -2.68460512e-01 -5.70530772e-01 -1.64299056e-01 3.76356602e-01 -2.29871228e-01 -1.13975585e+00 1.84410214e+00 -7.22688735e-02 2.14541346e-01 7.47458160e-01 6.68822229e-01 1.42543852e+00 9.57305610e-01 1.85590610e-01 1.00929305e-01 1.31412768e+00 -1.29080057e+00 -8.32728982e-01 -5.15514255e-01 9.55231488e-01 -6.65316224e-01 1.35497653e+00 2.42280632e-01 -1.13869917e+00 -4.26937789e-01 -7.34096766e-01 -2.93130994e-01 -5.83985209e-01 3.27439874e-01 1.04643881e-01 9.98293102e-01 -1.38935876e+00 1.91964343e-01 -9.33836758e-01 -6.72878563e-01 2.82356381e-01 9.22613889e-02 -4.77776974e-01 4.18386161e-02 -1.20487905e+00 9.32872355e-01 1.46462530e-01 -6.07680008e-02 -1.21672487e+00 -8.47875118e-01 -1.10166430e+00 6.47384450e-02 1.65897518e-01 -2.85545319e-01 1.81533563e+00 -8.82641017e-01 -1.96801436e+00 1.29484940e+00 -3.63318652e-01 -5.70805788e-01 2.48016238e-01 -1.54901892e-01 -7.31577575e-01 -2.02433467e-01 -9.02413055e-02 7.28347242e-01 4.68207031e-01 -9.04486299e-01 -4.58190709e-01 -3.33258867e-01 -2.56227612e-01 3.34579647e-01 -1.93932712e-01 6.38356268e-01 -2.11300105e-01 -5.16382098e-01 -2.37716183e-01 -6.56208932e-01 9.56768326e-06 -6.88904166e-01 -4.15550530e-01 -4.91540104e-01 5.31961322e-01 -1.01943052e+00 1.08480740e+00 -2.37491870e+00 8.81477222e-02 -3.70089859e-01 -3.12808841e-01 4.72109735e-01 -4.76606756e-01 5.61315060e-01 -2.36408144e-01 4.49445933e-01 -4.06626523e-01 -1.03500283e+00 1.27745137e-01 1.36934996e-01 -5.05485415e-01 2.33493056e-02 3.96815240e-01 7.82895744e-01 -7.12069213e-01 9.21207890e-02 2.27640048e-01 6.00637078e-01 -2.82634556e-01 5.09749472e-01 -4.25628126e-02 5.74746609e-01 4.39832173e-02 4.86422837e-01 1.65830657e-01 2.92631239e-03 -2.36649990e-01 7.19679967e-02 -7.62133375e-02 1.22599125e+00 -1.06802297e+00 1.82916713e+00 -9.15913284e-01 7.78043985e-01 4.74864006e-01 -8.67291152e-01 9.76918280e-01 7.59282351e-01 -1.48233339e-01 -5.42145133e-01 -1.22807892e-02 3.71093482e-01 -9.84017625e-02 -3.10434639e-01 4.58182603e-01 -1.81279983e-02 -1.04440406e-01 6.04279518e-01 5.51728725e-01 -2.39676893e-01 -1.38813779e-01 3.40454549e-01 1.17413104e+00 -3.48089010e-01 3.52069110e-01 -2.20783260e-02 6.99011803e-01 -1.53046533e-01 9.18476582e-02 6.14551783e-01 -4.16960388e-01 9.38867152e-01 1.33252144e-01 -7.42728561e-02 -8.47949684e-01 -1.02522802e+00 -2.00720519e-01 1.77990961e+00 -6.19032562e-01 -6.00405991e-01 -9.40169334e-01 -6.23052359e-01 -3.34516168e-01 1.09548640e+00 -4.53404099e-01 -3.39030176e-02 -4.28533822e-01 -4.30408031e-01 1.12338603e+00 6.21505737e-01 3.76800597e-01 -1.39247286e+00 9.20162946e-02 3.08603734e-01 -3.41896832e-01 -1.35402465e+00 -4.74634469e-01 2.67323047e-01 -4.32256550e-01 -5.73730767e-01 -8.17032695e-01 -8.95929396e-01 9.27711874e-02 -3.72200087e-02 1.40263200e+00 -1.91774398e-01 3.69034559e-02 6.71900690e-01 -5.50862312e-01 -5.03334463e-01 -7.82084703e-01 4.86913800e-01 4.63649305e-03 1.00307129e-01 4.08048034e-01 -2.37519462e-02 -1.72551855e-01 4.19498205e-01 -5.62839091e-01 -1.03515901e-01 3.20304900e-01 9.20070827e-01 1.95419297e-01 -8.01844835e-01 1.09236765e+00 -6.66804075e-01 8.98983181e-01 -4.35584992e-01 -5.05868755e-02 4.29495931e-01 8.08765292e-02 2.60874126e-02 2.56166756e-01 -1.52108818e-01 -1.29007733e+00 -5.90430312e-02 -9.08497274e-01 -2.35713929e-01 -6.90879762e-01 6.31058991e-01 -3.91597271e-01 4.36160386e-01 6.64738119e-01 3.97913873e-01 5.45998383e-03 -7.10020423e-01 6.33215606e-01 1.41948998e+00 6.14880741e-01 -4.55982685e-01 1.21207625e-01 -1.27882184e-02 -1.09183884e+00 -1.51822448e+00 -9.01683807e-01 -5.93769193e-01 -4.26267624e-01 2.49664247e-01 9.04464602e-01 -1.23010457e+00 -3.40339512e-01 4.48109180e-01 -1.19810247e+00 -6.09811127e-01 -3.98829341e-01 3.80499870e-01 -5.78157008e-01 9.51584615e-03 -5.93640625e-01 -1.01222289e+00 -4.52403039e-01 -1.28941727e+00 1.41618991e+00 -7.37420171e-02 -7.02499151e-01 -9.93768096e-01 4.09203351e-01 9.10478175e-01 6.79927051e-01 -5.51449955e-01 4.95594412e-01 -1.57275665e+00 -2.08506495e-01 -2.04487033e-02 1.43795907e-01 7.65298247e-01 -5.81242181e-02 2.72857789e-02 -1.59994519e+00 -3.13933700e-01 -2.79695004e-01 -9.09756005e-01 1.01243210e+00 2.02531353e-01 9.70212877e-01 -1.37212634e-01 -6.38832971e-02 5.41809916e-01 3.87188315e-01 1.08089156e-01 4.63735133e-01 5.83309680e-02 4.54478145e-01 1.02579391e+00 1.70013055e-01 6.82015345e-02 6.71810806e-01 6.03985965e-01 -1.84750944e-01 7.62550533e-02 -2.49611884e-01 -2.74877012e-01 7.49902487e-01 1.28004229e+00 3.59987617e-01 -6.93714261e-01 -1.10156786e+00 9.05578613e-01 -1.43550551e+00 -7.21518934e-01 2.78644413e-01 2.11072516e+00 9.15134668e-01 -2.09852681e-01 1.78737253e-01 1.05566800e-01 6.17590547e-01 3.96714300e-01 -3.46163779e-01 -6.68007314e-01 -2.04370156e-01 4.31171477e-01 -5.70809096e-02 7.13832736e-01 -1.16624713e+00 1.28865135e+00 6.55072069e+00 7.63084471e-01 -1.06911933e+00 3.48069787e-01 9.54273045e-01 -8.56238529e-02 -2.96442598e-01 -3.31227303e-01 -1.09235072e+00 1.35745272e-01 1.79334962e+00 -1.40204489e-01 4.13860112e-01 8.76442671e-01 5.73530942e-02 2.84526318e-01 -1.27867484e+00 1.04309058e+00 3.72642577e-01 -1.13980103e+00 -1.10057890e-01 -3.43471944e-01 3.95751595e-01 8.05167496e-01 -7.15277418e-02 8.01289082e-01 4.14680064e-01 -1.21454561e+00 5.02932072e-01 -4.57878336e-02 8.75530064e-01 -5.16114533e-01 7.40402222e-01 1.51505455e-01 -9.76322591e-01 2.43463039e-01 -7.12042227e-02 2.01294035e-01 4.44699347e-01 -6.03499124e-03 -1.22147357e+00 2.55975783e-01 5.86097062e-01 5.52029729e-01 -4.72843021e-01 7.27966487e-01 -3.47599149e-01 1.17470753e+00 -4.15272027e-01 -1.73084065e-01 1.82155833e-01 1.82908237e-01 5.23264289e-01 1.65572953e+00 9.33339074e-02 8.75156547e-04 3.84373702e-02 4.68006402e-01 -4.47936922e-01 4.14643049e-01 -5.47237694e-01 -5.04925728e-01 6.00934684e-01 1.20848238e+00 -3.75809878e-01 -4.77979600e-01 -6.22192621e-01 9.12049711e-01 4.86103058e-01 4.17065322e-01 -3.58797610e-01 -3.28757435e-01 1.08887196e+00 -1.27112180e-01 -3.79410316e-03 -3.02717775e-01 -2.03692049e-01 -1.29109085e+00 -2.63127297e-01 -1.25175333e+00 4.53564703e-01 -6.24116480e-01 -1.37303805e+00 1.02806199e+00 -2.86079466e-01 -4.61532474e-01 -6.46920383e-01 -5.39898217e-01 -6.97958291e-01 9.48126972e-01 -1.47538233e+00 -1.08913791e+00 -1.49470240e-01 5.61843991e-01 1.25489831e+00 -5.84993303e-01 1.14714909e+00 2.79764563e-01 -8.10108483e-01 8.70868385e-01 6.41756281e-02 5.11037469e-01 1.05154967e+00 -1.12057984e+00 8.26815307e-01 5.51541030e-01 6.14752471e-01 4.25533533e-01 4.55841303e-01 -3.44144642e-01 -9.69492257e-01 -9.30446029e-01 1.00439155e+00 -9.80661869e-01 8.15853715e-01 -9.29257631e-01 -1.19556093e+00 9.92659688e-01 5.37367046e-01 -8.62284973e-02 9.73693967e-01 5.88513911e-01 -3.98924977e-01 1.13383085e-01 -9.37145770e-01 2.73177415e-01 9.96525645e-01 -9.78130102e-01 -7.25368381e-01 2.70342410e-01 1.04770827e+00 -2.98891187e-01 -7.74104238e-01 3.17438781e-01 3.55975091e-01 -6.57902956e-01 8.08527052e-01 -9.55761552e-01 1.31782219e-01 1.59038246e-01 -4.13162053e-01 -1.76889443e+00 2.11668178e-01 -5.67592382e-01 3.13220561e-01 1.79264426e+00 1.01083612e+00 -5.67829370e-01 5.77177227e-01 5.67897975e-01 -7.22475946e-01 -5.10005176e-01 -1.07631910e+00 -6.53969407e-01 7.66834021e-02 -7.15269148e-01 4.79386777e-01 9.22299027e-01 -2.40917634e-02 7.70836473e-01 -3.34549025e-02 1.91130996e-01 1.91717744e-01 -4.22784925e-01 8.73185694e-01 -8.31401885e-01 -1.32080346e-01 -3.66223574e-01 -3.24157119e-01 -1.05615354e+00 5.67787051e-01 -9.89534318e-01 3.06520909e-01 -1.53516746e+00 -2.58012384e-01 -2.55907804e-01 -2.00938091e-01 6.31225705e-01 -9.86584648e-02 -2.29890756e-02 -1.28693525e-02 -1.43194482e-01 -6.59601510e-01 8.35925937e-01 6.54909909e-01 -1.23510316e-01 -3.26635480e-01 -7.34159499e-02 -6.44533575e-01 5.10663927e-01 7.85588324e-01 -1.94969490e-01 -3.95565122e-01 -7.27401316e-01 -3.07953537e-01 8.52636469e-04 7.95948729e-02 -9.31621432e-01 2.50339299e-01 3.02749604e-01 -7.59615079e-02 -5.52492321e-01 6.09409571e-01 -8.51718113e-02 -5.35260260e-01 -2.60259956e-01 -7.99411237e-01 -1.53803095e-01 2.99848676e-01 8.43849108e-02 -6.43502295e-01 -1.47701904e-01 8.29380393e-01 -5.71466722e-02 -8.86031866e-01 6.30441383e-02 -5.39940059e-01 5.55994511e-01 7.26870477e-01 1.86408609e-02 -3.84722263e-01 -7.89142966e-01 -1.14065623e+00 3.05602968e-01 -5.31532355e-02 9.86803114e-01 5.36113799e-01 -8.74503732e-01 -1.29572260e+00 4.50768650e-01 4.56649333e-01 -1.14475772e-01 2.65915036e-01 4.51239258e-01 -1.19172983e-01 6.14311099e-01 2.16198176e-01 -5.22676766e-01 -1.42861319e+00 4.48764786e-02 4.15228277e-01 -9.15064141e-02 -1.07556447e-01 1.35300684e+00 1.85700953e-01 -1.22311568e+00 5.12754381e-01 -2.62631953e-01 -3.13218027e-01 2.41540641e-01 8.34485531e-01 3.43628556e-01 4.80551690e-01 -8.81511927e-01 -6.66414917e-01 -1.42499909e-01 -1.51625410e-01 -5.33380389e-01 1.40082061e+00 -2.63573170e-01 3.17281753e-01 7.37356901e-01 1.33895183e+00 2.06324346e-02 -8.13688040e-01 -2.36719236e-01 2.16083840e-01 4.74915244e-02 1.49137303e-01 -1.21630740e+00 -8.30253303e-01 1.25244439e+00 5.25571287e-01 1.20731093e-01 7.57181406e-01 4.69211102e-01 6.60982430e-01 5.25149882e-01 9.95173454e-02 -1.02615178e+00 5.97314909e-03 9.56509292e-01 1.23381841e+00 -1.42521024e+00 -7.07186043e-01 -2.47884125e-01 -1.10689068e+00 7.33999133e-01 5.86751342e-01 3.17801505e-01 4.76221919e-01 4.19514477e-01 5.23735225e-01 1.04617797e-01 -1.00311017e+00 -3.61310840e-01 2.80629098e-01 6.27843022e-01 9.59573388e-01 1.02464326e-01 3.23297083e-01 8.53055000e-01 -5.12798071e-01 -3.47692877e-01 3.71109635e-01 6.18075192e-01 -3.33205968e-01 -1.20649552e+00 -1.41585484e-01 3.70149285e-01 -4.91986811e-01 -4.56265152e-01 -7.23868549e-01 5.41449726e-01 -5.80563664e-01 1.39242792e+00 2.63674378e-01 -2.94380456e-01 6.48180544e-01 6.68263793e-01 5.90382516e-02 -1.05663228e+00 -6.35779798e-01 -3.63815837e-02 8.88936400e-01 -5.67092240e-01 -2.04818115e-01 -7.87586391e-01 -9.93457794e-01 9.03856978e-02 -2.05789834e-01 3.72063220e-01 9.97410655e-01 8.90997946e-01 6.13120735e-01 5.56842446e-01 4.22535717e-01 -7.70395815e-01 -5.84557295e-01 -1.44098389e+00 -2.38170132e-01 2.74786919e-01 4.59036022e-01 -2.82569796e-01 -6.88867688e-01 -2.99022365e-02]
[14.165833473205566, 6.872860908508301]
b6bdf0ac-a549-45d9-8fce-919ec675ee66
dgsac-density-guided-sampling-and-consensus
2006.02413
null
https://arxiv.org/abs/2006.02413v1
https://arxiv.org/pdf/2006.02413v1.pdf
DGSAC: Density Guided Sampling and Consensus
Robust multiple model fitting plays a crucial role in many computer vision applications. Unlike single model fitting problems, the multi-model fitting has additional challenges. The unknown number of models and the inlier noise scale are the two most important of them, which are in general provided by the user using ground-truth or some other auxiliary information. Mode seeking/ clustering-based approaches crucially depend on the quality of model hypotheses generated. While preference analysis based guided sampling approaches have shown remarkable performance, they operate in a time budget framework, and the user provides the time as a reasonable guess. In this paper, we deviate from the mode seeking and time budget framework. We propose a concept called Kernel Residual Density (KRD) and apply it to various components of a multiple-model fitting pipeline. The Kernel Residual Density act as a key differentiator between inliers and outliers. We use KRD to guide and automatically stop the sampling process. The sampling process stops after generating a set of hypotheses that can explain all the data points. An explanation score is maintained for each data point, which is updated on-the-fly. We propose two model selection algorithms, an optimal quadratic program based, and a greedy. Unlike mode seeking approaches, our model selection algorithms seek to find one representative hypothesis for each genuine structure present in the data. We evaluate our method (dubbed as DGSAC) on a wide variety of tasks like planar segmentation, motion segmentation, vanishing point estimation, plane fitting to 3D point cloud, line, and circle fitting, which shows the effectiveness of our method and its unified nature.
['Saket Anand', 'Lokender Tiwari']
2020-06-03
null
null
null
null
['motion-segmentation']
['computer-vision']
[ 4.19512689e-02 -2.30715424e-01 -2.00506449e-01 -1.41524836e-01 -1.16366160e+00 -5.51679075e-01 2.92446524e-01 2.70892203e-01 -2.27921575e-01 2.15407208e-01 -4.79689211e-01 -4.22715768e-02 -3.96798432e-01 -2.81066477e-01 -6.38493419e-01 -8.01535606e-01 4.05771196e-01 9.30974603e-01 5.78274786e-01 8.81925523e-02 6.65205777e-01 7.56222546e-01 -1.37142408e+00 -2.98393667e-01 1.22496831e+00 9.11727309e-01 2.66524255e-01 5.14513135e-01 5.20425327e-02 4.23046164e-02 -2.53300428e-01 -1.72154918e-01 3.71035337e-01 -1.83981612e-01 -3.11418951e-01 4.07178968e-01 3.45877916e-01 -6.00786395e-02 3.81146222e-01 1.07186019e+00 4.58778858e-01 2.08617717e-01 7.94724405e-01 -1.36269569e+00 1.37459368e-01 -4.84343693e-02 -8.62894058e-01 -3.78318816e-01 2.44137928e-01 1.70337800e-02 6.82860672e-01 -1.32871985e+00 5.53662360e-01 1.08446062e+00 8.87333512e-01 3.65366399e-01 -1.28015399e+00 -2.73215950e-01 1.17407804e-02 -3.64481620e-02 -1.45725954e+00 -5.28024852e-01 1.02127421e+00 -6.26342356e-01 3.04867178e-01 3.92529190e-01 5.69066048e-01 4.86169636e-01 -4.39214036e-02 6.72456443e-01 6.71391964e-01 -4.72186416e-01 3.73787373e-01 1.64828494e-01 2.44706869e-01 6.61086440e-01 1.72387123e-01 -3.89963314e-02 -2.67875284e-01 -6.79978490e-01 1.15533793e+00 -1.13790497e-01 -4.61250693e-01 -8.53277028e-01 -1.21979320e+00 8.38620782e-01 3.36210728e-02 -3.06663360e-03 -4.26515907e-01 -2.20392980e-02 6.18996732e-02 1.21393688e-02 3.47439826e-01 4.33013141e-01 -5.53599656e-01 -3.67126614e-02 -1.30094552e+00 4.05915827e-01 6.31773591e-01 9.65417981e-01 9.46722806e-01 -2.66697168e-01 1.56735942e-01 9.19771373e-01 5.42850614e-01 3.88467610e-01 3.46408457e-01 -1.05655074e+00 1.47324994e-01 6.88055038e-01 2.53287554e-01 -1.37250388e+00 -3.28308046e-01 -2.96670437e-01 -8.50087166e-01 2.38846540e-01 5.50668716e-01 1.84824139e-01 -8.65908921e-01 1.26391315e+00 8.55832696e-01 3.48345578e-01 -3.66653442e-01 9.14934337e-01 5.21034539e-01 3.54503721e-01 -4.90610689e-01 -5.65852880e-01 9.03295815e-01 -1.00701177e+00 -5.02604842e-01 -1.39526963e-01 5.18838882e-01 -1.16838372e+00 8.01599026e-01 6.41456485e-01 -1.12599015e+00 -6.14683092e-01 -7.64169395e-01 4.66943905e-02 1.67616427e-01 3.92945766e-01 3.86016965e-01 2.51265854e-01 -7.46016622e-01 8.09840977e-01 -9.64504421e-01 -1.98867291e-01 1.25280231e-01 4.34579223e-01 -3.36814910e-01 1.53343529e-01 -3.28217655e-01 7.70496011e-01 1.54060692e-01 2.51616180e-01 -4.61534679e-01 -6.85181618e-01 -7.43509352e-01 -2.31163412e-01 5.95047712e-01 -8.46452236e-01 1.01313710e+00 -6.80816174e-01 -1.32454121e+00 8.19890797e-01 -5.35114527e-01 -2.47581050e-01 8.81406605e-01 -1.70563966e-01 1.05859302e-01 2.20857099e-01 1.25722006e-01 4.44116443e-01 1.33475244e+00 -1.57356393e+00 -4.77118790e-01 -2.44311482e-01 -5.15979111e-01 5.19983284e-02 2.63528198e-01 7.84771144e-02 -1.02806306e+00 -5.73880851e-01 8.97137165e-01 -1.01329231e+00 -6.67765319e-01 2.14787126e-01 -4.78535175e-01 -1.70695379e-01 8.57225060e-01 -5.63506126e-01 1.34512353e+00 -2.12168741e+00 2.37223789e-01 7.07843304e-01 2.19695851e-01 -5.68334386e-02 1.89742357e-01 4.76479679e-02 2.64873495e-03 4.93482873e-02 -4.37046826e-01 -6.55899048e-01 -4.98990491e-02 -2.87348274e-02 -1.18745826e-01 6.18850410e-01 1.73008323e-01 5.73656797e-01 -6.39051795e-01 -7.21038461e-01 3.89709175e-01 2.89653301e-01 -5.48232734e-01 1.15157440e-01 -2.93974221e-01 6.04916453e-01 -5.30054212e-01 7.37747729e-01 9.36321914e-01 -2.80641109e-01 -4.21497524e-01 -4.58357722e-01 -1.86848789e-01 -3.87423515e-01 -1.89569211e+00 1.69824409e+00 4.25820798e-03 2.15520233e-01 2.44136095e-01 -6.37489855e-01 1.14315343e+00 1.87379330e-01 7.73899674e-01 -9.80676617e-03 8.00434500e-02 6.37591958e-01 -1.36644498e-01 -2.47933149e-01 5.86419463e-01 1.47986963e-01 2.68417329e-01 1.69640660e-01 -2.45881841e-01 -5.26082695e-01 6.38145357e-02 1.05362415e-01 6.43339932e-01 3.75318706e-01 6.03793979e-01 -2.72710472e-01 4.79904830e-01 3.07557404e-01 9.17465329e-01 6.01606309e-01 -2.19990581e-01 1.14051914e+00 4.24744129e-01 -2.56787241e-01 -1.06016946e+00 -5.28261065e-01 -3.60725880e-01 3.46843421e-01 3.60406250e-01 -3.64982933e-01 -6.85431123e-01 -4.05154228e-01 1.18838577e-02 3.49460125e-01 -3.39095473e-01 1.57275409e-01 -6.94770038e-01 -3.93117011e-01 -1.26546159e-01 9.76400524e-02 1.17507502e-01 -6.35402679e-01 -6.22988701e-01 2.76480496e-01 -1.15190484e-01 -9.68701363e-01 -7.12770164e-01 1.34474188e-01 -1.08923602e+00 -1.36532950e+00 -7.76238084e-01 -5.20598710e-01 9.74443793e-01 3.06946188e-01 9.94652927e-01 2.73388773e-01 -1.09574392e-01 4.25169647e-01 -1.55724570e-01 -3.13398302e-01 -2.86988825e-01 -1.08769797e-01 1.08633183e-01 2.38004357e-01 2.77775936e-02 -4.09605801e-01 -5.42648494e-01 8.19455743e-01 -6.40539229e-01 3.59262712e-02 3.11920255e-01 7.14376807e-01 1.17426717e+00 -3.45479092e-03 4.12476063e-02 -6.77091479e-01 3.86165857e-01 -1.48205042e-01 -8.94411206e-01 1.67273358e-01 -3.80538940e-01 -1.12204716e-01 3.46163332e-01 -5.62240124e-01 -4.96870428e-01 5.15870750e-01 1.72241181e-02 -1.06099713e+00 -1.94008633e-01 4.65111911e-01 -2.60893404e-01 -4.20754194e-01 5.87409616e-01 2.77604461e-02 4.62075472e-02 -5.62525213e-01 1.78610444e-01 2.24252179e-01 5.32918334e-01 -5.41834652e-01 1.02250206e+00 4.73374277e-01 2.71129757e-01 -1.02725458e+00 -5.45688212e-01 -1.08012843e+00 -9.93649006e-01 -3.72449636e-01 5.80174923e-01 -5.18563926e-01 -6.04834855e-01 5.19966125e-01 -1.33670521e+00 -2.66947784e-04 -8.45586509e-02 4.50835466e-01 -7.32659578e-01 6.05612576e-01 -6.48844335e-03 -9.80023742e-01 -1.85841441e-01 -1.46812296e+00 1.28454340e+00 1.84820414e-01 -1.57765746e-01 -1.02202058e+00 3.85615155e-02 2.18167827e-01 -4.35340628e-02 5.62783778e-01 7.04539835e-01 -6.29476726e-01 -7.52135694e-01 -4.46836233e-01 1.12975359e-01 1.22698266e-02 -1.56281829e-01 4.86091465e-01 -8.22545052e-01 -3.29680741e-01 2.74162143e-01 6.31151646e-02 6.86384320e-01 8.24685872e-01 1.06224990e+00 1.13263071e-01 -5.80176711e-01 8.28245759e-01 1.52341866e+00 1.03327058e-01 4.26965058e-01 2.32135624e-01 8.44945729e-01 6.04808688e-01 1.01723576e+00 4.67538834e-01 1.49426356e-01 9.18596864e-01 5.11526108e-01 -1.00337029e-01 3.40396345e-01 -9.85379815e-02 -1.83693995e-03 8.63429546e-01 -2.09549423e-02 1.09747909e-01 -9.89571750e-01 4.13549334e-01 -2.26275110e+00 -6.48486674e-01 -6.40897393e-01 2.63185453e+00 4.84423250e-01 1.28165722e-01 2.88865656e-01 1.91885561e-01 7.25742996e-01 -3.39314580e-01 -7.82494009e-01 -6.45630807e-02 -4.63727228e-02 -2.30519459e-01 4.27545220e-01 6.87946677e-01 -1.09187222e+00 7.75072515e-01 5.36730480e+00 1.11642683e+00 -9.07456398e-01 -3.20000917e-01 4.23415273e-01 3.16316724e-01 -2.57236302e-01 3.79078329e-01 -9.14848685e-01 2.79541105e-01 4.06666756e-01 -2.67141080e-03 1.97456628e-01 1.09710431e+00 5.21471202e-01 -3.53230089e-01 -1.14923573e+00 1.40407217e+00 4.42858040e-02 -1.14525700e+00 -1.37835547e-01 1.35033160e-01 7.02350736e-01 -2.63365746e-01 -1.39308900e-01 -2.25502342e-01 -1.65608764e-01 -8.35484624e-01 8.52826238e-01 8.63484204e-01 4.48977292e-01 -7.33912826e-01 3.97035748e-01 6.73667073e-01 -1.22824955e+00 2.62767941e-01 -4.52158064e-01 5.07945180e-01 3.63437325e-01 9.13024068e-01 -6.95657790e-01 8.11838150e-01 3.19373369e-01 6.77224517e-01 -5.45557976e-01 1.59209502e+00 2.61864848e-02 3.18736076e-01 -7.40244508e-01 2.33783931e-01 2.83152098e-03 -7.19576359e-01 8.64083409e-01 7.50344455e-01 3.58026713e-01 4.61482480e-02 4.91777956e-01 8.94598663e-01 4.55457509e-01 2.65510798e-01 -3.92156571e-01 4.78245795e-01 5.27336240e-01 1.43909228e+00 -1.03469312e+00 -7.70393834e-02 -1.25733122e-01 8.86574268e-01 1.68482624e-02 2.48172909e-01 -7.31130421e-01 -1.37334928e-01 2.30451554e-01 8.43170732e-02 2.62795269e-01 -4.47385401e-01 -5.64877927e-01 -1.17057550e+00 2.75272846e-01 -9.35957730e-01 1.52476087e-01 -7.04051793e-01 -1.02571297e+00 3.68635416e-01 1.35870753e-02 -1.62940657e+00 -3.28003943e-01 -4.93304819e-01 -7.37112164e-01 6.59392536e-01 -1.35456634e+00 -1.07011056e+00 -2.88587660e-01 4.56933171e-01 7.30085015e-01 9.31178778e-02 5.49936473e-01 2.12933972e-01 -5.77906013e-01 3.66361618e-01 1.38951734e-01 -2.59054005e-01 7.20951855e-01 -1.19898093e+00 1.57904834e-01 8.71495247e-01 4.73135635e-02 8.23073566e-01 9.01730180e-01 -8.07221889e-01 -1.21228433e+00 -7.54385829e-01 6.45998836e-01 -4.37318712e-01 4.58367258e-01 -9.69821215e-02 -1.19407487e+00 4.19425815e-01 -4.65522826e-01 -1.44917043e-02 3.24424148e-01 -1.15683816e-01 2.06647962e-01 1.72203809e-01 -1.13598573e+00 5.96893191e-01 6.11880898e-01 1.96066014e-02 -2.73870319e-01 1.55351117e-01 4.80466604e-01 -6.43367648e-01 -7.47611821e-01 5.48592269e-01 4.08729494e-01 -1.04385293e+00 9.74323392e-01 -1.31220207e-01 7.06107765e-02 -7.11849630e-01 6.90147057e-02 -1.13565874e+00 -2.39419311e-01 -1.09553397e+00 -1.07707120e-01 1.20984054e+00 4.29411232e-01 -4.31844145e-01 8.73984993e-01 8.38753104e-01 -2.32299715e-01 -9.09920514e-01 -8.78403306e-01 -5.92513442e-01 -1.94495589e-01 -5.94769597e-01 5.11383176e-01 9.97731030e-01 -5.00971198e-01 5.81227653e-02 -2.39178374e-01 3.68217915e-01 1.10911787e+00 2.20125765e-01 1.23514688e+00 -1.60625494e+00 -2.74205953e-01 -4.05497730e-01 -1.90149754e-01 -1.32900369e+00 -1.54949918e-01 -4.99995649e-01 6.53191358e-02 -1.37257755e+00 -1.99911632e-02 -6.72897220e-01 3.26066166e-01 1.06056571e-01 -2.50013739e-01 -2.68144876e-01 1.75569132e-01 6.50401115e-01 -4.67952669e-01 3.93896520e-01 1.13755298e+00 2.99918979e-01 -6.13714516e-01 4.52692449e-01 -2.52433360e-01 1.19872880e+00 5.81423759e-01 -3.80048335e-01 -2.25475937e-01 -2.45569319e-01 1.11958444e-01 2.13342831e-01 5.07314563e-01 -9.19947147e-01 6.42828286e-01 -3.29363883e-01 1.72886699e-01 -1.06265485e+00 4.76057142e-01 -1.16220641e+00 3.81030262e-01 1.56424969e-01 6.86887503e-02 8.17947835e-02 -3.60915437e-02 4.49242115e-01 -6.34074435e-02 -5.33291042e-01 9.98535514e-01 -1.51904255e-01 -5.28718472e-01 5.47603250e-01 2.03828961e-01 -2.31483459e-01 1.03276730e+00 -8.34398091e-01 2.56593108e-01 -7.01799393e-02 -7.86344230e-01 4.83918369e-01 7.55949020e-01 4.01683114e-02 8.90267730e-01 -1.30022681e+00 -5.46820879e-01 3.54777396e-01 1.28833264e-01 5.36939740e-01 2.56798621e-02 1.19225883e+00 -4.24335629e-01 1.55077547e-01 4.36346620e-01 -1.17366099e+00 -1.27604187e+00 4.15984720e-01 3.83944035e-01 -6.86460733e-02 -4.14595276e-01 6.47686005e-01 -1.60158247e-01 -3.85236889e-01 2.99719691e-01 -5.95764160e-01 -1.38606668e-01 1.01234056e-01 2.24340945e-01 6.38202190e-01 1.51591527e-03 -6.43920302e-01 -1.99356467e-01 1.20181942e+00 1.27755925e-01 -5.91042116e-02 1.21046531e+00 -1.16044596e-01 -1.87043801e-01 6.12350523e-01 9.16769683e-01 1.10350713e-01 -1.28320456e+00 -7.75281116e-02 2.39974499e-01 -6.39625251e-01 -2.78053824e-02 -2.84479588e-01 -7.14702785e-01 6.24712825e-01 3.32334965e-01 2.33099982e-01 9.62636769e-01 -8.26647282e-02 5.89401364e-01 -7.43986666e-03 3.86150360e-01 -1.13713288e+00 -9.66660008e-02 3.01783025e-01 9.89435196e-01 -1.40469670e+00 1.25508592e-01 -6.95325553e-01 -4.80901629e-01 1.33227658e+00 5.03933430e-01 -1.19068518e-01 6.90191865e-01 3.78151722e-02 1.01880491e-01 -2.31781289e-01 -3.42086881e-01 -1.74674198e-01 6.02028251e-01 4.12849277e-01 3.29194255e-02 -3.08634520e-01 -2.54141182e-01 4.71233279e-01 -1.37189418e-01 -1.83564946e-01 2.39401415e-01 5.99880934e-01 -6.73345387e-01 -1.14599311e+00 -9.59724605e-01 3.93679559e-01 -1.22241780e-01 2.80143023e-01 -1.84368297e-01 6.88608468e-01 -8.67123995e-03 9.04313684e-01 -2.11246923e-01 -8.73319060e-02 5.29648960e-01 -1.47703812e-01 1.50681704e-01 -4.77013737e-01 -3.95094931e-01 8.13066483e-01 -3.04657876e-01 -5.97306073e-01 -4.24768537e-01 -8.66412520e-01 -1.22881973e+00 -1.11482739e-02 -8.96888554e-01 2.49538153e-01 7.38537192e-01 8.54036510e-01 2.74921834e-01 2.98463367e-02 8.00631344e-01 -1.13036752e+00 -5.74809432e-01 -6.83339298e-01 -4.17676151e-01 3.14143628e-01 4.07176673e-01 -5.92439950e-01 -5.41203558e-01 7.97734633e-02]
[7.845843315124512, -2.6791834831237793]
0a8c0292-95a0-40dd-bf9f-b6020b03bf88
one-shot-learning-from-a-demonstration-with
null
null
https://openreview.net/forum?id=uecnUP0-PjZ
https://openreview.net/pdf?id=uecnUP0-PjZ
One-Shot Learning from a Demonstration with Hierarchical Latent Language
Humans have the capability, aided by the expressive compositionality of their language, to learn quickly by demonstration. They are able to describe unseen task-performing procedures and generalize their execution to other contexts. In this work, we introduce DescribeWorld, an environment designed to test this sort of generalization skill in grounded agents, where tasks are linguistically and procedurally composed of elementary concepts. The agent observes a single task demonstration in a Minecraft-like grid world, and is then asked to carry out the same task in a new map.To enable such a level of generalization, we propose a neural agent infused with hierarchical latent language—both at the level of task inference and subtask planning. Our agent first generates a textual description of the demonstrated unseen task, then leverages this description to replicate it. Through multiple evaluation scenarios and a suite of generalization tests, we find that agents that perform text-based inference are better equipped for the challenge under a random split of tasks.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['one-shot-learning']
['methodology']
[ 2.58084297e-01 4.06700373e-01 2.68730849e-01 -3.72865170e-01 -4.89596725e-01 -1.04835916e+00 1.28715682e+00 8.28318596e-02 -2.64153183e-01 7.07945287e-01 4.06313926e-01 -3.33198071e-01 9.01004225e-02 -7.09089696e-01 -6.76069558e-01 -3.63716513e-01 -2.88757861e-01 9.94467020e-01 3.75707373e-02 -3.64603817e-01 1.84260085e-01 5.34332633e-01 -1.71525395e+00 5.09634674e-01 8.15169573e-01 3.35108846e-01 7.55599916e-01 6.66445911e-01 2.33396426e-01 1.48636651e+00 -5.02027512e-01 2.51209065e-02 2.79060662e-01 -4.12416488e-01 -1.09017062e+00 5.15323207e-02 2.67297685e-01 -5.88227153e-01 -3.96243811e-01 6.94144011e-01 2.94957086e-02 6.94266558e-01 8.12625468e-01 -1.35601032e+00 -5.30540943e-01 9.01078224e-01 3.51623893e-01 -5.87000884e-02 8.12507510e-01 7.89115191e-01 8.91992390e-01 -6.50678992e-01 7.23587751e-01 1.29252863e+00 4.36715513e-01 8.29905570e-01 -1.36190498e+00 -4.88794088e-01 2.46093482e-01 -2.03564018e-01 -1.12244606e+00 -3.79806757e-01 4.13352311e-01 -5.79750240e-01 1.43864846e+00 -1.18419707e-01 6.47199333e-01 1.45296550e+00 1.29797891e-01 6.33300960e-01 1.33213568e+00 -2.05517828e-01 7.06835568e-01 9.35124606e-02 -2.85313278e-01 1.12029767e+00 1.69155836e-01 7.27974236e-01 -9.13171053e-01 -9.35160145e-02 7.75023341e-01 9.74003822e-02 -1.10709198e-01 -4.76422369e-01 -1.50824535e+00 5.15072286e-01 5.13055623e-01 2.82355398e-01 -6.01771176e-01 3.82458687e-01 4.96929616e-01 2.83920944e-01 -1.25895068e-01 1.29081941e+00 -3.15615088e-01 -1.47086620e-01 -8.81625712e-01 8.74866724e-01 1.19476032e+00 1.34115553e+00 6.92401230e-01 1.80537343e-01 -3.86899054e-01 6.36451542e-02 -7.52481446e-02 3.75826806e-01 6.95288539e-01 -1.40978599e+00 4.17841285e-01 5.23124278e-01 3.38199824e-01 -4.54484165e-01 -5.67821920e-01 -2.57755518e-01 -4.09493536e-01 6.53195381e-01 2.16473863e-01 -1.95835516e-01 -8.04669380e-01 2.03695011e+00 7.78474361e-02 5.02602533e-02 5.27518988e-01 8.76924455e-01 3.27076644e-01 6.24296546e-01 4.44801986e-01 5.72435856e-02 1.41050816e+00 -1.14908624e+00 -1.54892206e-01 -7.65340865e-01 8.52727413e-01 3.29481095e-01 1.53985798e+00 3.59201312e-01 -1.24393284e+00 -7.22136497e-01 -1.25536215e+00 -2.61243641e-01 -5.12518406e-01 -4.14533585e-01 8.99673104e-01 5.15012033e-02 -1.21369946e+00 6.42699122e-01 -1.02677417e+00 -5.62553108e-01 3.96686137e-01 -8.79021436e-02 -4.38123971e-01 -1.10844947e-01 -8.26894045e-01 1.33511734e+00 9.71609294e-01 -1.64644152e-01 -2.02562332e+00 -6.01375580e-01 -1.26043177e+00 3.56498569e-01 5.47074676e-01 -1.23850453e+00 1.81547177e+00 -4.68546689e-01 -1.43902338e+00 9.20528114e-01 1.14503905e-01 -5.11389434e-01 4.66179132e-01 3.10350414e-02 1.55980974e-01 -4.96021993e-02 4.09183562e-01 8.85448158e-01 6.55861557e-01 -1.38740599e+00 -7.36215413e-01 -4.24990267e-01 6.72111690e-01 5.05718470e-01 3.66590083e-01 -4.16230559e-01 2.89639711e-01 -3.59302670e-01 6.26554415e-02 -1.02672708e+00 -1.80130675e-01 -4.20214981e-01 -2.90648758e-01 -2.64145523e-01 2.88813323e-01 -4.01840210e-01 4.29177970e-01 -2.09522390e+00 4.78177071e-01 -1.48180351e-01 3.07607859e-01 -3.45458895e-01 -2.32294202e-02 7.33990073e-01 2.72492647e-01 1.00287013e-01 -2.93153405e-01 -5.51309168e-01 7.47605860e-01 1.97315007e-01 -6.21840596e-01 -5.18293725e-03 -3.46808173e-02 1.24823868e+00 -1.35306609e+00 -1.70299813e-01 1.58905387e-02 -6.36481196e-02 -6.50510073e-01 4.98396963e-01 -8.69575858e-01 7.06693411e-01 -5.02690136e-01 3.93453330e-01 -8.32722858e-02 -1.39618129e-01 3.00752163e-01 2.03915462e-01 -1.51935920e-01 4.91419226e-01 -7.31606364e-01 2.39671206e+00 -9.08652365e-01 5.24715543e-01 -6.36172155e-03 -5.39153576e-01 5.29993117e-01 3.15481216e-01 -4.45724398e-01 -4.10611987e-01 -1.60651207e-01 1.15516104e-01 1.82299465e-01 -7.17613399e-01 5.78335285e-01 -4.23757941e-01 -4.86531287e-01 8.61181438e-01 1.74114674e-01 -9.54764724e-01 2.40984365e-01 3.74913394e-01 1.32087207e+00 6.64620817e-01 4.72108006e-01 -4.06509161e-01 2.12748885e-01 5.33978164e-01 -6.43060282e-02 1.33787870e+00 -8.04358944e-02 4.69508581e-02 2.51240134e-01 -7.52564549e-01 -1.24104631e+00 -1.35596478e+00 4.40229595e-01 1.58483696e+00 -1.54194891e-01 -3.42297167e-01 -6.77998424e-01 -5.83920062e-01 2.75452174e-02 1.46392238e+00 -8.06186259e-01 -2.18555510e-01 -5.69270730e-01 1.37725890e-01 7.43945420e-01 6.79895520e-01 7.24890709e-01 -1.82302094e+00 -1.43732202e+00 1.38586506e-01 1.26563907e-01 -1.03661239e+00 -1.87602237e-01 4.21459138e-01 -8.04042578e-01 -7.29684889e-01 -2.25829691e-01 -9.94781673e-01 5.01326740e-01 -2.03958616e-01 1.36302066e+00 2.38954183e-02 -9.54742059e-02 6.61743164e-01 -1.75533429e-01 -3.57836455e-01 -8.19213271e-01 2.06722245e-01 1.96687147e-01 -7.21241534e-01 5.71446791e-02 -8.91205609e-01 -2.41199583e-01 -1.10050939e-01 -9.09588993e-01 4.97876555e-01 6.53845668e-01 7.83978641e-01 6.40917867e-02 7.65280947e-02 3.90961707e-01 -7.74547756e-01 1.18851328e+00 -4.53891695e-01 -7.80864954e-01 3.17503631e-01 -3.50351363e-01 5.41526973e-01 7.81477273e-01 -3.41516495e-01 -1.21503520e+00 -4.65038307e-02 5.64129353e-01 2.04187236e-03 -5.71765304e-01 6.75182998e-01 -2.40341704e-02 4.27961916e-01 1.08519447e+00 6.18061900e-01 -1.97086915e-01 -7.99135044e-02 7.24563062e-01 1.24674395e-01 9.96903479e-01 -1.32982862e+00 8.16664040e-01 3.23457658e-01 -3.32090631e-02 -4.74220306e-01 -8.16757202e-01 1.36043116e-01 -5.27264178e-01 9.11809281e-02 1.03612959e+00 -7.72578657e-01 -1.19703269e+00 1.66254804e-01 -1.09572804e+00 -1.41871393e+00 -5.79530180e-01 3.00307453e-01 -1.25918317e+00 -2.79717058e-01 -5.44391990e-01 -6.20489180e-01 9.98756513e-02 -1.33122706e+00 1.15503621e+00 4.44880947e-02 -5.57814300e-01 -9.67058361e-01 2.24133074e-01 -1.14541221e-02 5.26632786e-01 2.10071906e-01 1.28351235e+00 -9.61474895e-01 -6.96546078e-01 2.57930279e-01 -1.65296141e-02 -1.08925663e-01 -6.10886067e-02 -7.90911198e-01 -1.03522766e+00 -3.89946103e-01 2.79898435e-01 -9.75596011e-01 3.89884979e-01 -1.91398442e-01 9.70722318e-01 -4.23394799e-01 -4.11474228e-01 6.06173992e-01 1.02271771e+00 4.71748784e-03 2.90534347e-01 4.93612111e-01 2.11145476e-01 6.68184102e-01 3.21079552e-01 2.19506100e-01 6.21803164e-01 5.47313690e-01 2.30656117e-01 4.17873412e-01 4.66010869e-02 -7.51790285e-01 4.26561922e-01 1.91805676e-01 -6.39151260e-02 2.28056516e-02 -1.36600709e+00 3.81690830e-01 -1.71591711e+00 -1.14003897e+00 8.81936491e-01 1.70578897e+00 1.03001738e+00 1.93153650e-01 1.26562566e-02 -5.06055593e-01 1.43118173e-01 1.21735394e-01 -8.37968051e-01 -5.02997875e-01 2.89121300e-01 1.35990620e-01 -1.16569713e-01 8.50768685e-01 -6.25072002e-01 1.27887928e+00 6.54728270e+00 1.75013065e-01 -5.59941113e-01 -1.32929206e-01 2.36165464e-01 -2.31002551e-02 -1.81133151e-01 -3.26046837e-03 -4.15649295e-01 2.70596463e-02 9.87776995e-01 -3.74295026e-01 1.07426560e+00 8.31439316e-01 -8.99314880e-02 -2.34985724e-01 -2.10747743e+00 5.46263456e-01 9.79233235e-02 -1.33159578e+00 2.40477234e-01 -2.18881309e-01 6.51915014e-01 4.99990247e-02 2.79183775e-01 1.08636796e+00 1.00065756e+00 -1.60929286e+00 8.93488765e-01 4.16051954e-01 6.39394820e-01 -1.50002077e-01 3.62186968e-01 1.11361229e+00 -7.91904211e-01 -3.45848441e-01 -5.36391996e-02 -6.51082277e-01 1.41988084e-01 -5.45932114e-01 -1.33487356e+00 1.41537800e-01 3.44848514e-01 2.08768576e-01 -4.17943805e-01 3.88077080e-01 -6.95062935e-01 -1.81500521e-02 -6.72760904e-02 -1.65701464e-01 3.87362331e-01 -6.92088157e-02 4.14487720e-01 1.09346128e+00 3.23034704e-01 4.61225241e-01 6.37701333e-01 1.44354367e+00 7.50574097e-02 -4.83847916e-01 -1.04798806e+00 -8.96820873e-02 6.63879037e-01 9.03537393e-01 -3.93811017e-01 -6.73763692e-01 -3.79565246e-02 1.06052101e+00 7.01230228e-01 7.60390878e-01 -5.70285380e-01 1.92113817e-02 3.05312634e-01 -1.31084830e-01 -9.24042091e-02 -6.70052707e-01 -1.73631042e-01 -1.02243936e+00 -7.13514164e-02 -1.22071898e+00 5.24889603e-02 -1.53821695e+00 -1.02241611e+00 7.22513735e-01 4.35575843e-01 -5.45789003e-01 -9.15604413e-01 -6.05793118e-01 -5.20811021e-01 9.75453734e-01 -1.02861094e+00 -1.23459768e+00 -6.90531373e-01 5.35357118e-01 8.75231445e-01 -4.89210099e-01 1.24573064e+00 -6.24644578e-01 1.83155779e-02 -3.89397936e-03 -6.56663120e-01 -1.60374697e-02 2.24553376e-01 -1.37744761e+00 7.01690972e-01 4.47441339e-01 1.86125621e-01 1.00241876e+00 9.91231680e-01 -7.24234164e-01 -1.20150030e+00 -9.28498268e-01 5.09481728e-01 -1.00228953e+00 8.19812119e-01 -8.70065749e-01 -8.31121743e-01 1.42241251e+00 3.12272608e-01 -2.65547365e-01 3.11113238e-01 6.88209832e-02 -5.79355359e-01 2.81415880e-01 -1.12396824e+00 1.02173877e+00 1.50695264e+00 -1.06852758e+00 -1.56182480e+00 6.30734384e-01 9.54251349e-01 -6.09950125e-01 -5.35142958e-01 2.47081593e-02 5.44876575e-01 -8.91979635e-01 7.63946712e-01 -1.15883040e+00 5.66532135e-01 -3.10669810e-01 -3.97695780e-01 -1.69203460e+00 -4.69851524e-01 -7.04708874e-01 1.25827305e-02 6.06413543e-01 5.85295796e-01 -7.71618366e-01 5.97808003e-01 6.75335646e-01 -4.49953258e-01 -3.03254187e-01 -5.26386201e-01 -9.56233144e-01 2.41661578e-01 -2.81943321e-01 8.11377823e-01 6.67668521e-01 6.39295638e-01 4.90911335e-01 3.64691764e-01 2.60621071e-01 5.35527647e-01 1.80739537e-01 8.99541080e-01 -1.06225610e+00 -5.74435830e-01 -4.34013784e-01 3.20711993e-02 -1.09522116e+00 6.95351064e-01 -1.29704845e+00 4.72453862e-01 -1.50056946e+00 2.56238759e-01 -4.06085402e-01 1.15852915e-01 6.41266882e-01 2.11136833e-01 -3.21990937e-01 3.65340799e-01 3.28184456e-01 -7.73099899e-01 3.55463564e-01 1.18746984e+00 -1.28852934e-01 -3.51236343e-01 -3.48807275e-01 -6.08613372e-01 9.09359217e-01 8.72094035e-01 -3.05562615e-01 -8.02565515e-01 -6.84582293e-01 4.70651746e-01 1.57535464e-01 7.28018761e-01 -1.33196878e+00 7.51630187e-01 -3.28628838e-01 3.57308626e-01 -4.67101336e-02 4.21137244e-01 -8.15488815e-01 8.42771977e-02 5.88766038e-01 -9.78164315e-01 5.40536404e-01 4.02252227e-01 4.22185302e-01 2.46075261e-02 -2.20109984e-01 2.25160941e-01 -7.38936543e-01 -1.00880742e+00 4.16294634e-02 -5.69810450e-01 2.75721103e-01 1.10710883e+00 -1.37248605e-01 -6.24213338e-01 -3.71585608e-01 -1.08867621e+00 3.25683206e-01 8.83111775e-01 1.07027829e-01 5.32866061e-01 -9.91276443e-01 -8.09628308e-01 2.36427620e-01 3.85311544e-01 2.35342056e-01 -3.34808119e-02 4.11697268e-01 -6.02905512e-01 4.64910299e-01 -4.63325024e-01 -4.12240356e-01 -3.20044547e-01 8.83968294e-01 6.08547091e-01 -2.27153972e-01 -8.18845510e-01 7.64208198e-01 6.47884905e-01 -6.92752361e-01 1.56169385e-01 -7.00045228e-01 1.96259663e-01 -4.61620301e-01 4.85288233e-01 -2.89216544e-02 -1.23205051e-01 -3.76518965e-02 -1.39641643e-01 -1.40803820e-02 -5.20827621e-02 -6.57076001e-01 1.10566032e+00 2.16618955e-01 -7.17671514e-02 5.90993166e-01 5.70197642e-01 -2.16830149e-01 -1.65145206e+00 -1.10737972e-01 1.05609655e-01 -1.17668897e-01 -4.29173052e-01 -1.24488330e+00 4.04192694e-02 7.44875550e-01 -1.91984609e-01 3.35418284e-01 6.58667445e-01 3.03446084e-01 1.57545224e-01 1.07399464e+00 8.38312447e-01 -8.64946604e-01 4.35796559e-01 7.98160911e-01 1.26142311e+00 -9.84407544e-01 -3.25890891e-02 1.60224989e-01 -8.10154259e-01 8.53733301e-01 6.34837985e-01 -2.01362923e-01 -1.94044411e-01 3.87185663e-01 -3.71500671e-01 -5.48402190e-01 -1.15778112e+00 -3.69567275e-02 -4.58291434e-02 1.03457308e+00 -1.19131021e-02 1.20300293e-01 6.95246100e-01 3.98375213e-01 -8.03284168e-01 6.97552934e-02 5.46362877e-01 1.03481114e+00 -5.93193233e-01 -3.96030873e-01 1.74576603e-02 1.10611871e-01 3.29059362e-01 -1.67233348e-01 -3.35147917e-01 9.81560946e-01 -3.53522860e-02 6.44545138e-01 1.22611877e-02 1.64700225e-02 4.05582011e-01 5.56793630e-01 8.55003238e-01 -1.19583476e+00 -6.83794379e-01 -8.66690218e-01 1.35113597e-01 -8.83486927e-01 7.08931009e-04 -6.93568051e-01 -1.50010395e+00 -1.78385437e-01 6.03512943e-01 2.32301638e-01 5.73332429e-01 9.82497573e-01 1.40611380e-01 6.82385445e-01 6.16993792e-02 -1.20321512e+00 -9.60737884e-01 -1.03957057e+00 -4.70975876e-01 7.03897834e-01 4.95146662e-01 -5.83781958e-01 -4.14874941e-01 1.70644686e-01]
[4.320830345153809, 0.9621452689170837]
440baac7-dac3-4d4d-a2c7-5e30cadc6d1f
privacy-preserving-and-uncertainty-aware
2303.0434
null
https://arxiv.org/abs/2303.04340v1
https://arxiv.org/pdf/2303.04340v1.pdf
Privacy-preserving and Uncertainty-aware Federated Trajectory Prediction for Connected Autonomous Vehicles
Deep learning is the method of choice for trajectory prediction for autonomous vehicles. Unfortunately, its data-hungry nature implicitly requires the availability of sufficiently rich and high-quality centralized datasets, which easily leads to privacy leakage. Besides, uncertainty-awareness becomes increasingly important for safety-crucial cyber physical systems whose prediction module heavily relies on machine learning tools. In this paper, we relax the data collection requirement and enhance uncertainty-awareness by using Federated Learning on Connected Autonomous Vehicles with an uncertainty-aware global objective. We name our algorithm as FLTP. We further introduce ALFLTP which boosts FLTP via using active learning techniques in adaptatively selecting participating clients. We consider both negative log-likelihood (NLL) and aleatoric uncertainty (AU) as client selection metrics. Experiments on Argoverse dataset show that FLTP significantly outperforms the model trained on local data. In addition, ALFLTP-AU converges faster in training regression loss and performs better in terms of NLL, minADE and MR than FLTP in most rounds, and has more stable round-wise performance than ALFLTP-NLL.
['Lili Su', 'Fei Miao', 'Dongjin Song', 'Jiangwei Wang', 'Muzi Peng']
2023-03-08
null
null
null
null
['trajectory-prediction']
['computer-vision']
[-5.75733125e-01 4.46294665e-01 -8.28518748e-01 -8.06696951e-01 -1.33234596e+00 -6.08690500e-01 6.09860301e-01 1.93199962e-01 -5.45482159e-01 1.04454505e+00 6.68258369e-02 -5.72264552e-01 -2.71824986e-01 -8.39617074e-01 -1.02622306e+00 -8.02695096e-01 -1.21542469e-01 7.05174625e-01 2.30541199e-01 1.85173973e-01 -3.45794946e-01 5.02933741e-01 -1.19325459e+00 6.49573132e-02 1.03467000e+00 1.38749051e+00 -4.70518440e-01 5.64185269e-02 1.11873269e-01 1.11245906e+00 -5.39268069e-02 -9.82810259e-01 5.56816578e-01 2.24736035e-01 -7.12504506e-01 -3.83861959e-01 1.21521160e-01 -7.27285147e-01 -5.70444942e-01 7.31824696e-01 2.81299680e-01 2.88002700e-01 4.98194039e-01 -2.12495303e+00 -7.94394314e-02 8.20085526e-01 -3.07692021e-01 -2.82147557e-01 -1.10568546e-01 3.37297201e-01 9.64810014e-01 -8.14208865e-01 5.10899842e-01 1.10699320e+00 8.16651344e-01 4.70847130e-01 -1.19569683e+00 -1.01986897e+00 3.59800905e-01 6.03723466e-01 -1.36503398e+00 -7.44955778e-01 4.70544130e-01 -1.67261988e-01 8.92207801e-01 1.99414372e-01 1.84468284e-01 1.24560261e+00 -4.79268795e-03 1.21933413e+00 5.24374962e-01 2.73931950e-01 8.27808619e-01 4.65617329e-01 -1.60675764e-01 3.30117524e-01 3.41777712e-01 4.36932951e-01 -6.22520804e-01 -6.40439868e-01 9.27213430e-02 1.44836321e-01 1.32388296e-02 -9.70966935e-01 -7.63230443e-01 9.29344654e-01 3.18149865e-01 -5.53673923e-01 -5.09501934e-01 4.03175950e-01 6.08927488e-01 5.25716305e-01 4.40721154e-01 4.20956314e-02 -8.32331777e-01 -2.35419825e-01 -7.08382607e-01 1.54625416e-01 8.40730906e-01 1.25185072e+00 8.48169923e-01 -2.04881415e-01 -1.96195349e-01 3.17984462e-01 6.89097703e-01 5.26993513e-01 -2.45034412e-01 -1.41077924e+00 7.80796647e-01 5.80442727e-01 4.02001649e-01 -7.20546901e-01 -2.48784930e-01 -3.94819826e-02 -6.03707671e-01 4.78545636e-01 1.45983204e-01 -5.06228924e-01 -5.00062525e-01 1.72399163e+00 5.69814503e-01 1.44855389e-02 2.51371592e-01 7.70370066e-01 5.29866636e-01 6.14960730e-01 1.88268840e-01 -4.41106975e-01 4.26236033e-01 -9.26219702e-01 -7.13652372e-01 6.49999529e-02 8.50672007e-01 4.30249423e-02 6.84869409e-01 3.44664544e-01 -7.47899711e-01 3.38697731e-01 -8.11739445e-01 1.89470530e-01 -3.19469452e-01 -2.01955035e-01 7.60479391e-01 6.78497851e-01 -9.02300954e-01 3.44595045e-01 -1.05850232e+00 -9.95270535e-02 1.18112493e+00 5.69473803e-01 -5.22338152e-01 -9.36805084e-02 -1.14794481e+00 9.28196609e-01 3.13488275e-01 -2.85003394e-01 -1.30606079e+00 -9.01338577e-01 -6.75984681e-01 -3.87080833e-02 8.04850578e-01 -2.79823244e-01 1.35323906e+00 -4.78692114e-01 -1.66092157e+00 2.73720503e-01 1.28078982e-01 -9.85803783e-01 1.11897695e+00 -1.61471009e-01 -3.22980732e-01 -1.05915979e-01 -1.55739725e-01 5.07825851e-01 6.38671041e-01 -1.14150190e+00 -7.57579982e-01 -2.89036214e-01 5.99072725e-02 9.96474475e-02 -2.31774956e-01 -1.99980974e-01 -1.95146337e-01 2.10614637e-01 -3.31534058e-01 -6.74520850e-01 -5.14194965e-01 6.23002231e-01 -1.39800459e-01 -5.24786711e-01 1.21403027e+00 -3.50434095e-01 1.02716887e+00 -2.12889886e+00 -4.40215647e-01 5.71829438e-01 4.13636267e-01 4.29667562e-01 -1.21093988e-01 6.26043320e-01 5.32428861e-01 -1.29132211e-01 -2.60105819e-01 -5.92210174e-01 3.95974278e-01 3.24176043e-01 -5.93305349e-01 6.33681715e-01 2.28125840e-01 8.92045140e-01 -9.39391971e-01 -5.33877552e-01 2.16326177e-01 2.52318829e-01 -5.80486357e-01 1.01077579e-01 -6.30822182e-01 5.90620577e-01 -7.07538962e-01 7.43985057e-01 8.52590382e-01 -1.40921801e-01 8.29591509e-03 2.75534719e-01 -2.74390839e-02 2.12121338e-01 -6.51258886e-01 1.36880469e+00 -5.05010307e-01 5.84572315e-01 9.09905210e-02 -7.82038450e-01 8.90691638e-01 3.61098200e-01 7.77507842e-01 -6.38697147e-01 1.52726129e-01 2.85073340e-01 -6.08350575e-01 -3.92305493e-01 2.90419370e-01 5.41281581e-01 -2.09774956e-01 3.95373970e-01 -1.42589346e-01 4.53469068e-01 -2.84307331e-01 4.41180944e-01 1.30178761e+00 5.94611801e-02 1.65341139e-01 7.68725798e-02 2.39675850e-01 1.85250025e-02 8.14037144e-01 5.57633758e-01 -7.99660325e-01 2.40353234e-02 6.98652148e-01 -3.71403933e-01 -8.03643763e-01 -9.69488680e-01 -1.82223529e-01 9.98103499e-01 2.95687765e-01 -3.24070424e-01 -4.72468466e-01 -1.39129436e+00 5.29962122e-01 1.25929248e+00 -4.21865493e-01 -2.52376974e-01 -2.62021601e-01 -6.23182714e-01 6.50582254e-01 4.52531785e-01 4.80588108e-01 -4.88496363e-01 -3.64177257e-01 5.53557929e-03 -8.22338685e-02 -8.85475039e-01 -2.18905598e-01 5.67709394e-02 -4.57780421e-01 -9.44463134e-01 -1.44046500e-01 -5.67496009e-02 5.05882025e-01 -3.15539651e-02 8.30136716e-01 -4.21697021e-01 2.60502428e-01 2.92317808e-01 -2.76748806e-01 -6.61274672e-01 -1.84324548e-01 1.40625760e-01 2.18125790e-01 1.54645503e-01 6.52660191e-01 -5.07928312e-01 -5.28828800e-01 4.51806128e-01 -3.97040218e-01 -4.68786865e-01 4.90519643e-01 5.85684240e-01 6.16167843e-01 -8.33239853e-02 8.64848912e-01 -7.00248420e-01 3.08984101e-01 -8.42914939e-01 -1.08314717e+00 3.88495535e-01 -9.31731641e-01 3.67320515e-02 4.75031137e-01 -3.78421456e-01 -1.19867980e+00 3.17179978e-01 6.43403903e-02 -7.02866554e-01 9.89917666e-02 5.73804565e-02 -4.84060854e-01 -2.73949593e-01 7.08628476e-01 -2.56252110e-01 4.75414991e-02 -7.83822462e-02 5.65468550e-01 8.01717997e-01 1.08051986e-01 -2.57975608e-01 7.88051367e-01 6.34256840e-01 -6.42268136e-02 -2.61894822e-01 -4.74456519e-01 -1.64801419e-01 -2.19268635e-01 -2.73901492e-01 3.61823350e-01 -9.71358716e-01 -1.38684380e+00 2.68825710e-01 -9.75195944e-01 -5.26560605e-01 -6.46957517e-01 7.72083521e-01 -7.22603858e-01 1.72136083e-01 -1.46739602e-01 -1.33897197e+00 -3.12293977e-01 -1.00672221e+00 6.45549834e-01 -1.01936653e-01 3.82548049e-02 -7.27100968e-01 2.50738770e-01 3.27395022e-01 5.86782038e-01 2.43463337e-01 4.84587669e-01 -1.01150596e+00 -1.16160202e+00 -5.23717880e-01 -2.55457222e-01 1.18373163e-01 -3.27905715e-01 -2.07197785e-01 -1.09707117e+00 -3.45799804e-01 -1.96055517e-01 -6.78251445e-01 8.25943947e-01 1.69391409e-01 1.17546523e+00 -8.66340578e-01 -5.44299603e-01 6.07269824e-01 1.31388736e+00 1.22906864e-01 6.17944539e-01 1.55233294e-01 4.16762799e-01 5.91095626e-01 9.06187356e-01 8.04750860e-01 7.68853664e-01 5.65501332e-01 9.75310326e-01 2.76121169e-01 4.26845253e-01 -3.78324062e-01 5.85039377e-01 2.41435915e-01 3.93950373e-01 -5.23502231e-01 -9.32641625e-01 5.57543337e-01 -2.52013493e+00 -9.01078224e-01 5.81164621e-02 2.44355869e+00 7.83690453e-01 -2.45492473e-01 2.13474780e-01 -2.84729451e-01 2.85723388e-01 1.69655140e-02 -1.17672646e+00 -2.89207667e-01 -1.12222329e-01 -5.82908452e-01 1.09517908e+00 3.95099819e-01 -1.18674028e+00 7.63992429e-01 5.75268650e+00 1.01712739e+00 -8.55054259e-01 4.09271598e-01 7.94550240e-01 -4.71718520e-01 -4.28893030e-01 2.58281752e-02 -7.40185142e-01 5.31401396e-01 1.06801283e+00 -2.88471401e-01 4.38073158e-01 1.33173335e+00 7.80728906e-02 4.43487801e-02 -1.42862618e+00 8.65025878e-01 -4.48654771e-01 -1.27475452e+00 -2.94487953e-01 3.07837844e-01 8.93157780e-01 6.91165507e-01 1.97713390e-01 4.50621396e-01 9.51042354e-01 -1.04300594e+00 7.31547117e-01 6.62306011e-01 6.64928317e-01 -1.11012137e+00 8.02138686e-01 5.64719737e-01 -8.36683989e-01 -4.25389618e-01 -4.44059908e-01 4.80641007e-01 1.71402559e-01 4.70061451e-01 -8.89887333e-01 5.79798341e-01 6.81688011e-01 4.53669190e-01 7.85017610e-02 1.13526404e+00 3.58944014e-03 6.18418336e-01 -7.53104508e-01 -4.49289717e-02 2.37964898e-01 -7.22165257e-02 5.87933838e-01 7.83789039e-01 1.00804590e-01 1.22917242e-01 1.71471313e-01 8.04232180e-01 -3.99563491e-01 -5.40658943e-02 -6.45695090e-01 -9.70022976e-02 9.98992622e-01 1.10756922e+00 1.16924740e-01 -1.00272961e-01 -3.61571491e-01 5.57097971e-01 3.45494747e-01 4.31763977e-01 -1.03637779e+00 -6.98446408e-02 1.00689995e+00 -5.21856807e-02 3.59131515e-01 6.46942779e-02 -2.56171077e-01 -9.28927422e-01 1.16391547e-01 -5.49437940e-01 4.24498737e-01 -2.02056140e-01 -1.46129727e+00 5.56708694e-01 -6.65612295e-02 -1.37722754e+00 -4.53551918e-01 -2.69770116e-01 -5.55018246e-01 5.11398077e-01 -1.51315165e+00 -1.10513639e+00 1.69203952e-02 7.62460768e-01 1.74742118e-02 -4.64376211e-01 7.44511962e-01 4.07964647e-01 -7.01257169e-01 1.27979648e+00 6.48736119e-01 -1.34472743e-01 6.69877589e-01 -7.19969988e-01 3.15197438e-01 7.08579600e-01 -3.06926906e-01 1.49758026e-01 5.36876738e-01 -6.06515884e-01 -1.53565180e+00 -1.63314557e+00 8.34341228e-01 -7.95716166e-01 7.11770654e-01 -2.80148804e-01 -6.05204642e-01 8.49349022e-01 -8.62874240e-02 2.66382873e-01 4.79717076e-01 -6.93829358e-02 -5.23402810e-01 -4.83229339e-01 -1.84196997e+00 3.75492156e-01 9.10916448e-01 -5.00648439e-01 1.50153071e-01 4.88547415e-01 1.08251524e+00 -1.64219067e-01 -8.49924624e-01 4.66621906e-01 4.42144871e-01 -8.23874831e-01 5.91296792e-01 -7.73321033e-01 -3.05733949e-01 -1.33817568e-01 -3.14509839e-01 -8.29858303e-01 -8.24765339e-02 -9.51781511e-01 -6.41471505e-01 1.30624962e+00 9.08457100e-01 -1.12624633e+00 1.18034446e+00 1.36241293e+00 4.42338400e-02 -9.63382959e-01 -1.31164443e+00 -9.61439550e-01 -4.20868695e-02 -7.98094392e-01 1.05251646e+00 8.60787451e-01 1.27706289e-01 -2.89379358e-01 -4.25585389e-01 1.94825202e-01 9.89526689e-01 -2.14556545e-01 9.25942063e-01 -1.14623594e+00 9.23265368e-02 -4.82552797e-02 -4.90158111e-01 -7.35082030e-01 2.80816585e-01 -6.85749710e-01 1.69282228e-01 -1.29013979e+00 -6.15023896e-02 -7.55982935e-01 -3.56019080e-01 7.74684310e-01 5.30560076e-01 -2.24754602e-01 2.43214574e-02 2.17150837e-01 -1.16754544e+00 1.06876338e+00 5.18276155e-01 -2.17401683e-01 -2.00723767e-01 4.94955510e-01 -3.56192201e-01 5.41044235e-01 9.50064898e-01 -5.96711576e-01 -6.50469959e-01 -2.74408489e-01 2.91205645e-01 5.26376106e-02 3.41029316e-01 -5.03039658e-01 4.97590661e-01 -4.64801759e-01 5.27069457e-02 -9.40891623e-01 5.21864891e-01 -1.21156168e+00 3.47544134e-01 6.53868467e-02 -5.36674261e-01 -4.18959618e-01 -8.55570436e-02 9.58130538e-01 6.64283633e-02 2.61302739e-01 5.57984531e-01 2.96221435e-01 -5.56915641e-01 9.84045506e-01 -3.70666414e-01 -1.11630358e-01 1.29986668e+00 -4.24772874e-02 -3.87054294e-01 -6.33753240e-01 -3.45578432e-01 8.95838499e-01 6.22577488e-01 2.29288846e-01 5.49415708e-01 -1.42874491e+00 -6.16994381e-01 8.77864137e-02 4.07074332e-01 1.62292391e-01 1.57720432e-01 1.19499218e+00 -1.57612115e-01 5.84651053e-01 2.50119865e-01 -4.99738365e-01 -9.15785372e-01 6.29882812e-01 2.36184031e-01 -3.80557477e-02 -4.73730057e-01 9.38875437e-01 9.88252237e-02 -8.70393574e-01 1.02338898e+00 1.94686744e-02 1.83906347e-01 -9.01002288e-02 4.99304175e-01 9.39739525e-01 1.20695792e-01 -5.40025949e-01 -5.99445462e-01 -3.27759564e-01 -1.42976537e-01 -1.18748501e-01 1.25505161e+00 -4.22009230e-01 1.81761190e-01 1.78157419e-01 1.29746032e+00 -2.20915869e-01 -1.88941514e+00 -5.12052357e-01 2.52930373e-01 -4.31833446e-01 1.53331146e-01 -9.94385719e-01 -1.20958722e+00 5.55731773e-01 6.66146159e-01 -2.03069165e-01 8.54414046e-01 2.17836518e-02 8.64535868e-01 9.09594297e-01 9.98097122e-01 -1.21445858e+00 -4.21855122e-01 2.17993662e-01 7.19109774e-01 -1.55110204e+00 -8.07362944e-02 -1.87082753e-01 -9.45568800e-01 5.11588693e-01 6.14156187e-01 2.49383539e-01 7.15467453e-01 1.99219853e-01 1.46101743e-01 1.18236214e-01 -1.27854741e+00 9.55850407e-02 -7.36720264e-02 6.78815126e-01 -4.43715066e-01 2.25989670e-01 1.31071821e-01 7.14368045e-01 1.12736881e-01 9.87705309e-03 1.33820906e-01 6.61155343e-01 -2.84366012e-01 -1.12049806e+00 -2.58326735e-02 4.52274144e-01 -9.95731354e-02 3.70497972e-01 -5.35185933e-01 4.72083241e-01 3.65690626e-02 1.15914321e+00 3.58542874e-02 -6.26793742e-01 -6.78781122e-02 -2.48771831e-01 5.07695191e-02 8.36510956e-02 -3.35404187e-01 -6.35387599e-01 3.56650084e-01 -1.05397272e+00 -4.64352071e-02 -8.72598827e-01 -1.27160180e+00 -7.82939732e-01 -4.51674700e-01 2.45996565e-01 7.71081388e-01 7.57694602e-01 8.14664066e-01 -3.12489539e-01 1.16077721e+00 -4.06086475e-01 -1.01505601e+00 -5.15985727e-01 -3.32490683e-01 -2.16842487e-01 5.62411368e-01 -5.93414843e-01 -3.64484966e-01 -6.15064383e-01]
[5.86344051361084, 6.33668851852417]
cc7caff8-7b5f-496c-91b4-c733d28e6c48
exploring-the-use-of-foundation-models-for
2304.05336
null
https://arxiv.org/abs/2304.05336v1
https://arxiv.org/pdf/2304.05336v1.pdf
Exploring the Use of Foundation Models for Named Entity Recognition and Lemmatization Tasks in Slavic Languages
This paper describes Adam Mickiewicz University's (AMU) solution for the 4th Shared Task on SlavNER. The task involves the identification, categorization, and lemmatization of named entities in Slavic languages. Our approach involved exploring the use of foundation models for these tasks. In particular, we used models based on the popular BERT and T5 model architectures. Additionally, we used external datasets to further improve the quality of our models. Our solution obtained promising results, achieving high metrics scores in both tasks. We describe our approach and the results of our experiments in detail, showing that the method is effective for NER and lemmatization in Slavic languages. Additionally, our models for lemmatization will be available at: https://huggingface.co/amu-cai.
['Artur Nowakowski', 'Gabriela Pałka']
2023-04-11
null
null
null
null
['lemmatization']
['natural-language-processing']
[-6.65443301e-01 9.41122770e-02 -1.79363176e-01 -3.50993812e-01 -9.30644214e-01 -9.77621138e-01 8.82208407e-01 2.02171840e-02 -6.56624794e-01 4.13410127e-01 6.29543722e-01 -5.80986738e-01 1.65513694e-01 -4.61892337e-01 -2.49644086e-01 7.57297203e-02 2.73400873e-01 8.32074881e-01 7.64412060e-02 -4.42079991e-01 1.52550891e-01 6.05262101e-01 -7.75184929e-01 6.50729299e-01 5.97812712e-01 4.01267946e-01 -8.52305889e-02 5.74320793e-01 -4.54271287e-01 7.25536764e-01 -4.81145203e-01 -1.05683529e+00 2.45466784e-01 1.81408286e-01 -1.64196408e+00 -5.02719164e-01 4.28724974e-01 1.81252763e-01 -2.29202164e-03 7.48815775e-01 3.46981347e-01 1.77067339e-01 7.55012393e-01 -9.16378081e-01 -3.34203511e-01 1.34743500e+00 -4.88857664e-02 7.52593875e-02 4.40196216e-01 -4.38556857e-02 1.44974577e+00 -1.05746210e+00 1.16299570e+00 1.17915821e+00 8.19601893e-01 5.26791215e-01 -9.06515658e-01 -6.38605237e-01 5.02833985e-02 2.22189710e-01 -1.47982335e+00 -7.57849753e-01 4.18970019e-01 -5.69713593e-01 1.12473619e+00 1.88253596e-01 3.14619094e-01 9.04241621e-01 -1.66738838e-01 1.22326851e+00 1.03231061e+00 -6.85727954e-01 -2.00178787e-01 3.72048914e-01 4.66320902e-01 6.62324786e-01 2.40362063e-01 -2.99943060e-01 -3.88627470e-01 -1.83761626e-01 4.99363989e-01 -6.54572725e-01 1.93477511e-01 -4.18492228e-01 -1.11688828e+00 7.44847834e-01 -3.73225361e-02 8.78945351e-01 -6.00923710e-02 1.63778439e-01 6.19761527e-01 1.38126343e-01 6.51821196e-01 1.04823136e+00 -9.13753688e-01 -1.67718843e-01 -1.04303741e+00 1.90837041e-01 1.07111156e+00 1.10731494e+00 4.98512089e-01 -2.49127626e-01 -2.31495146e-02 8.33316088e-01 2.51209378e-01 1.95312873e-01 3.15504164e-01 -9.47991490e-01 6.31292820e-01 4.59553242e-01 6.89472109e-02 -6.94488823e-01 -3.94567996e-01 -1.24430180e-01 6.37720898e-03 -2.47808307e-01 5.70920050e-01 -2.11100429e-01 -8.85053098e-01 1.44940209e+00 1.32579446e-01 -5.18895127e-02 1.40031055e-01 5.23149014e-01 1.03884566e+00 1.91702023e-01 5.32581925e-01 3.48680645e-01 1.53279519e+00 -8.80220592e-01 -6.38606668e-01 -1.03496812e-01 1.21203327e+00 -1.23768377e+00 9.01701033e-01 1.02215610e-01 -1.23697960e+00 -2.51287460e-01 -5.75585485e-01 -4.52248126e-01 -7.53014743e-01 5.41364789e-01 8.94312918e-01 8.11089814e-01 -1.02618802e+00 5.67845464e-01 -9.01162744e-01 -8.54296327e-01 5.27543016e-02 1.02780029e-01 -5.91427684e-01 2.94331193e-01 -1.21548498e+00 1.05761683e+00 4.47223276e-01 -1.24315225e-01 -5.04879594e-01 -5.19817829e-01 -1.09701562e+00 -5.08387983e-02 1.18600458e-01 -2.55793482e-01 1.52703202e+00 -5.00299156e-01 -1.39623606e+00 1.46001089e+00 -1.89653322e-01 -6.26168907e-01 5.32602966e-01 -4.91405874e-01 -3.79742295e-01 1.25072479e-01 1.74091145e-01 4.43956852e-01 1.16134502e-01 -1.03224075e+00 -5.84434569e-01 -1.42508730e-01 2.35274479e-01 1.16739519e-01 -2.93145150e-01 6.15146101e-01 -4.22878236e-01 -5.20505548e-01 -2.54963133e-02 -7.27462411e-01 -7.63846785e-02 -9.27082658e-01 -2.50698775e-01 -3.35128665e-01 2.43891805e-01 -1.02750003e+00 1.30561459e+00 -2.10011888e+00 -2.13686883e-01 2.49379724e-01 2.75113824e-04 4.93733078e-01 -2.19070733e-01 9.47918832e-01 -1.99838370e-01 8.53495300e-01 -5.61052747e-03 -6.44940615e-01 8.95531178e-02 -4.99820821e-02 -4.02240157e-01 4.21690762e-01 4.89603076e-03 1.07172143e+00 -7.54232109e-01 -6.53000295e-01 8.64564478e-02 1.26132503e-01 -3.15851212e-01 -1.07551478e-01 5.80457970e-02 1.00612953e-01 -4.92583662e-02 5.97510278e-01 3.90449882e-01 2.38358036e-01 4.50212449e-01 -1.22751467e-01 -3.64524215e-01 8.14764321e-01 -1.11664271e+00 1.80552363e+00 -7.72142291e-01 4.47353989e-01 3.45679730e-01 -4.38770503e-01 8.10289323e-01 7.01130331e-01 3.17542642e-01 -5.12751192e-02 3.77358407e-01 4.02301937e-01 -2.08184317e-01 -2.44942233e-01 6.89768672e-01 -1.40334610e-02 -3.01656008e-01 5.35377383e-01 3.34760696e-01 -4.29003656e-01 5.34262538e-01 6.06217921e-01 8.76428127e-01 3.18497300e-01 6.12022817e-01 -5.32026172e-01 7.26463377e-01 2.09444761e-01 5.69160700e-01 5.20023763e-01 -4.46585238e-01 4.76214409e-01 3.73131692e-01 -2.34283820e-01 -1.03052747e+00 -9.27637041e-01 -1.38914615e-01 1.12717426e+00 -3.57897878e-01 -8.69397283e-01 -6.67664349e-01 -1.07273519e+00 -2.08844095e-01 1.03690660e+00 -4.43509281e-01 3.90593350e-01 -7.85866559e-01 -1.44978419e-01 1.28591263e+00 4.48131204e-01 2.46668205e-01 -1.11055672e+00 -1.37606561e-01 5.13031892e-03 -3.77962023e-01 -1.37289989e+00 -3.71793956e-01 -5.10942303e-02 -6.33020878e-01 -1.21581340e+00 -3.79367411e-01 -9.83255982e-01 1.60657495e-01 -2.32232541e-01 1.06338537e+00 -9.52019468e-02 -2.48278797e-01 5.36818922e-01 -5.22593856e-01 -3.87141407e-01 -4.22994137e-01 7.55411088e-01 -6.93187043e-02 -3.71651262e-01 3.94042790e-01 -1.45913407e-01 2.14946340e-03 1.04597896e-01 -7.42084920e-01 -1.73963215e-02 2.84416169e-01 6.10307515e-01 -7.52928182e-02 -6.84182823e-01 2.51316696e-01 -1.43250358e+00 5.20619333e-01 -2.03278482e-01 -4.69469875e-01 2.23189324e-01 -5.24796963e-01 -1.24897286e-01 4.82438624e-01 8.80612731e-02 -1.35759676e+00 1.46197647e-01 -4.57166523e-01 -9.74329039e-02 -4.07995433e-01 5.10504901e-01 -1.70002386e-01 6.47323504e-02 4.64843571e-01 -2.84523159e-01 -3.90891194e-01 -9.28978682e-01 7.27350712e-01 9.40835476e-01 3.88795853e-01 -7.92857230e-01 7.55710363e-01 4.71026361e-01 -4.16656673e-01 -9.22476351e-01 -7.15858042e-01 -1.05384350e+00 -1.06043708e+00 9.58306864e-02 8.46618474e-01 -1.04961073e+00 -3.12197626e-01 3.36793125e-01 -1.23925710e+00 -4.32665348e-01 -3.73540223e-01 5.26739478e-01 -4.40706700e-01 6.44380271e-01 -1.24352837e+00 -9.28358316e-01 -4.98834908e-01 -7.21446872e-01 9.28641319e-01 -6.70524538e-02 -6.21981740e-01 -1.51941943e+00 3.58322382e-01 7.03953743e-01 2.84560204e-01 -1.75881073e-01 1.07008302e+00 -1.34778035e+00 -5.29254274e-03 -2.61694133e-01 -5.46403117e-02 1.95701927e-01 -1.62775025e-01 1.57223731e-01 -9.41305876e-01 -1.13505803e-01 -4.34031665e-01 -2.96989113e-01 6.96396291e-01 -1.25215381e-01 4.83321309e-01 -3.16275395e-02 -2.09508091e-01 6.08835459e-01 1.11396873e+00 -2.66510725e-01 4.73215312e-01 5.75219691e-01 5.40594935e-01 7.63966024e-01 6.74487054e-01 3.88924368e-02 6.23436391e-01 4.62008625e-01 -8.33099931e-02 3.39402887e-03 -3.99149269e-01 -4.02325541e-01 3.95980269e-01 1.02086127e+00 -9.28999633e-02 7.33480826e-02 -1.45150805e+00 1.08405173e+00 -1.81875288e+00 -7.86070645e-01 -4.51028734e-01 1.78306794e+00 6.74603403e-01 -1.77685410e-01 -7.19612092e-02 -3.14315259e-01 6.32780254e-01 1.49636760e-01 3.60226125e-01 -8.06254566e-01 -2.63615191e-01 6.40374422e-01 5.51187754e-01 8.66802752e-01 -1.42526186e+00 2.00058007e+00 6.83115578e+00 8.64567697e-01 -6.48570955e-01 3.52919400e-01 2.38160729e-01 -9.66277253e-03 -3.55563968e-01 3.89253974e-01 -1.12412679e+00 -1.37836158e-01 1.04101408e+00 -2.17899948e-01 1.90333426e-01 7.42668927e-01 1.83016554e-01 -6.52470738e-02 -8.63300920e-01 5.07920623e-01 1.20624848e-01 -1.18971145e+00 -4.27737199e-02 -5.57491556e-02 6.45779073e-01 3.20511043e-01 -3.12461942e-01 5.67390561e-01 8.27320576e-01 -7.59532511e-01 8.08224559e-01 1.89249262e-01 7.78343976e-01 -6.15316868e-01 1.07542527e+00 1.48892522e-01 -1.26210737e+00 3.08365017e-01 -9.46320295e-02 -7.32488707e-02 2.61500329e-01 3.71424735e-01 -1.14785469e+00 8.76420200e-01 4.21278775e-01 6.33987069e-01 -6.92602873e-01 9.83602166e-01 -7.47904122e-01 9.28737223e-01 -2.88210779e-01 6.04879111e-02 3.30170989e-01 -2.26162449e-01 5.15044630e-01 1.81910360e+00 -7.51377568e-02 -8.36851299e-02 2.36225009e-01 3.66242647e-01 -1.79477543e-01 8.77532840e-01 -5.26838899e-01 -1.69443607e-01 3.94599974e-01 1.52834165e+00 -8.31099808e-01 -4.80948478e-01 -1.26794845e-01 8.32472146e-01 6.06481075e-01 2.37827271e-01 -5.35381377e-01 -6.33685410e-01 5.86862683e-01 9.43325162e-02 6.11848906e-02 -4.87758011e-01 -6.26277447e-01 -1.43303716e+00 -2.68860042e-01 -8.27328801e-01 7.92637527e-01 -4.28554893e-01 -1.08197379e+00 5.14604151e-01 1.65074512e-01 -5.00796080e-01 -1.97906360e-01 -7.00733125e-01 -4.84526187e-01 8.05714130e-01 -1.18723226e+00 -1.72122920e+00 3.26821297e-01 3.50131720e-01 3.69267792e-01 -1.83835924e-01 8.72342229e-01 3.55057687e-01 -4.64673698e-01 5.11941254e-01 -7.65643418e-02 7.22829700e-01 1.19658756e+00 -1.33733928e+00 9.86824811e-01 1.03900826e+00 5.45496643e-01 9.42154408e-01 4.97282892e-01 -8.28893363e-01 -7.71670878e-01 -8.15048099e-01 1.76419461e+00 -7.24871576e-01 1.20689988e+00 -5.46573579e-01 -5.14820576e-01 1.19025171e+00 4.93625492e-01 -6.60978436e-01 9.11950111e-01 6.27301097e-01 -5.41946113e-01 4.26197201e-01 -1.03350949e+00 6.79614604e-01 1.26954591e+00 -6.39709473e-01 -9.55117464e-01 4.50295389e-01 3.75773907e-01 -2.92699039e-01 -9.37078953e-01 1.28777355e-01 4.87026334e-01 -5.99777520e-01 5.07455945e-01 -1.00004482e+00 1.23829231e-01 -5.67287393e-02 -4.16012853e-02 -1.19718480e+00 -2.35777497e-01 -4.55982417e-01 3.53917956e-01 1.61303580e+00 7.40407825e-01 -6.19681776e-01 6.99892282e-01 5.82515180e-01 -1.14133880e-01 -3.13491672e-01 -7.82509327e-01 -8.79914463e-01 4.85470116e-01 -8.75640333e-01 4.28142101e-01 1.11040759e+00 4.69310373e-01 4.01626587e-01 4.28084210e-02 4.70905937e-02 1.96951509e-01 -8.82488303e-03 8.25584173e-01 -1.11611044e+00 9.06888954e-03 -3.17992777e-01 -3.20653349e-01 -5.81023514e-01 6.46755219e-01 -1.47652173e+00 -1.79690212e-01 -1.72936463e+00 -7.30236247e-02 -4.31839556e-01 1.51629299e-01 8.47921550e-01 1.55579835e-01 2.15707257e-01 6.26814127e-01 3.06648761e-01 -5.78566551e-01 1.35693654e-01 4.79964316e-01 8.01507086e-02 -1.13073222e-01 -6.70930222e-02 -6.11381412e-01 9.67469037e-01 1.14486015e+00 -2.90634394e-01 3.23825598e-01 -5.78165472e-01 2.52031386e-01 -5.88597775e-01 -6.03086203e-02 -8.53752851e-01 2.32915774e-01 -1.85979716e-02 -1.06679862e-02 -4.68424141e-01 3.10147256e-01 -3.76981407e-01 -6.06328100e-02 3.69698226e-01 -3.42938632e-01 1.37065515e-01 1.72368810e-01 -2.59310335e-01 -1.73523724e-01 -5.78704834e-01 7.43669868e-01 -3.20231676e-01 -7.90124238e-01 -5.01929261e-02 -5.93818128e-01 2.30599388e-01 8.38566065e-01 4.49792922e-01 -1.55616507e-01 -3.87846649e-01 -9.72058654e-01 3.47359538e-01 6.07959092e-01 2.89383978e-01 2.29935154e-01 -8.25484097e-01 -7.59761572e-01 -1.52093619e-01 1.27142265e-01 -4.90780085e-01 -2.46442407e-01 6.83661938e-01 -9.60056603e-01 7.72877514e-01 1.09535359e-01 -7.39500998e-03 -1.57025564e+00 1.76694199e-01 2.25383505e-01 -7.78061152e-01 -3.04299802e-01 7.47516155e-01 -2.39684895e-01 -9.96782064e-01 -1.17438182e-01 -9.40015614e-02 -5.16018331e-01 2.78982788e-01 2.29651555e-01 5.50632060e-01 2.52750665e-01 -7.69681931e-01 -5.81278026e-01 3.12615037e-01 -3.23249102e-01 -6.40016317e-01 1.20275819e+00 -1.86203167e-01 -3.23163122e-01 5.71183562e-01 9.23284411e-01 6.69953585e-01 -1.94554761e-01 -1.11751601e-01 6.60745800e-01 -5.33250906e-02 -2.80040056e-01 -9.55343783e-01 -6.20554686e-01 7.42239833e-01 6.70794621e-02 -1.03516266e-01 4.41278368e-01 3.19749378e-02 7.75624871e-01 5.65480590e-01 5.65862060e-01 -1.41543853e+00 -6.97816372e-01 1.25746834e+00 4.42859769e-01 -8.35351288e-01 4.08020765e-02 -6.47284925e-01 -8.46455514e-01 1.00725818e+00 3.58103156e-01 -9.22461972e-02 6.04029298e-01 3.58278900e-01 6.85104549e-01 -2.03815103e-01 -5.48508883e-01 -7.06756473e-01 1.05494469e-01 3.90372545e-01 9.50791061e-01 3.19109291e-01 -1.00601220e+00 6.53337359e-01 -5.99196672e-01 -2.89756030e-01 6.80713117e-01 8.50938976e-01 -7.35880360e-02 -1.46710658e+00 -3.16136986e-01 8.85174870e-02 -8.58488142e-01 -5.63279867e-01 -9.66302812e-01 1.14790916e+00 3.99637185e-02 9.61339295e-01 -2.60027796e-01 -3.26916188e-01 4.15017098e-01 6.61250830e-01 2.12240741e-01 -1.10719693e+00 -1.21518409e+00 1.13654546e-02 7.31593430e-01 -5.48113704e-01 -3.52662280e-02 -9.68097687e-01 -1.27591658e+00 -3.28461856e-01 -1.90449581e-01 6.72590137e-01 8.13786149e-01 8.51159036e-01 3.02012563e-01 -6.85840175e-02 1.69112012e-01 -4.43483889e-01 -4.13847715e-01 -1.15710378e+00 -6.29491806e-01 4.88925993e-01 -4.31391716e-01 -1.19738035e-01 -4.29384679e-01 1.49200723e-01]
[9.875004768371582, 9.767131805419922]
017d076b-3b8a-4fc0-a687-98c53fbf10d0
investigating-cross-domain-behaviors-of-bert
2306.15123
null
https://arxiv.org/abs/2306.15123v2
https://arxiv.org/pdf/2306.15123v2.pdf
Investigating Cross-Domain Behaviors of BERT in Review Understanding
Review score prediction requires review text understanding, a critical real-world application of natural language processing. Due to dissimilar text domains in product reviews, a common practice is fine-tuning BERT models upon reviews of differing domains. However, there has not yet been an empirical study of cross-domain behaviors of BERT models in the various tasks of product review understanding. In this project, we investigate text classification BERT models fine-tuned on single-domain and multi-domain Amazon review data. In our findings, though single-domain models achieved marginally improved performance on their corresponding domain compared to multi-domain models, multi-domain models outperformed single-domain models when evaluated on multi-domain data, single-domain data the single-domain model was not fine-tuned on, and on average when considering all tests. Though slight increases in accuracy can be achieved through single-domain model fine-tuning, computational resources and costs can be reduced by utilizing multi-domain models that perform well across domains.
['Meng Jiang', 'Albert Lu']
2023-06-27
null
null
null
null
['text-classification']
['natural-language-processing']
[-1.32266104e-01 1.65908292e-01 -5.64798355e-01 -6.92957878e-01 -8.70149851e-01 -8.10035825e-01 5.40886104e-01 3.73627633e-01 -4.64230597e-01 6.37221336e-01 -5.05362265e-02 -6.63022161e-01 -2.93750077e-01 -6.33594513e-01 -4.61191922e-01 2.06110656e-01 5.80379009e-01 7.46732295e-01 9.37592313e-02 -2.90160090e-01 2.96795845e-01 -1.89837351e-01 -1.18722975e+00 6.73719227e-01 9.17748690e-01 6.74901962e-01 1.07014313e-01 6.81965649e-01 -3.29388499e-01 4.86032158e-01 -7.13043511e-01 -8.87345612e-01 5.45850158e-01 -2.71177232e-01 -4.82260942e-01 8.49820748e-02 2.69710898e-01 -4.39974070e-01 2.80548871e-01 7.44169235e-01 1.04875825e-01 -2.50478964e-02 1.10506117e+00 -1.13719451e+00 -1.15746880e+00 6.31265640e-01 -9.23751295e-01 2.36526445e-01 4.34842855e-01 -7.11294934e-02 1.53238225e+00 -7.41521657e-01 6.00090683e-01 1.19871926e+00 7.07298636e-01 2.62895703e-01 -1.26953125e+00 -6.99863315e-01 4.05098081e-01 -2.32003018e-01 -9.55697358e-01 -1.12946264e-01 3.54264468e-01 -3.90697688e-01 1.25046217e+00 -4.36947733e-01 4.37045097e-01 6.53453350e-01 6.45798743e-01 6.88619614e-01 1.13374233e+00 -4.90268797e-01 2.95809329e-01 8.09173405e-01 4.86807138e-01 -2.21079793e-02 6.70452774e-01 -2.10554004e-01 -5.59053600e-01 -2.57190287e-01 2.95080394e-01 -6.54071122e-02 5.10103166e-01 -2.03169510e-01 -5.75150251e-01 1.19459021e+00 -3.76179442e-02 1.92806050e-01 -5.22541046e-01 -3.25099617e-01 6.68487251e-01 8.08680832e-01 1.13287175e+00 7.71971822e-01 -1.24957800e+00 -5.45313418e-01 -9.80075240e-01 5.28825819e-01 1.20542133e+00 1.02908981e+00 8.61457109e-01 -2.78976023e-01 4.79594201e-01 1.34021163e+00 1.53335631e-01 4.36514825e-01 7.45599747e-01 -4.63859648e-01 5.14338613e-01 8.65388334e-01 2.09443033e-01 -8.99615169e-01 -4.19210374e-01 -5.14302365e-02 -2.00162187e-01 1.75236799e-02 4.96887773e-01 -5.34183443e-01 -9.46342409e-01 1.18189538e+00 5.71445115e-02 -5.63762426e-01 2.02170879e-01 6.64056063e-01 6.69185758e-01 5.81341207e-01 4.91739333e-01 -1.04644587e-02 1.48378062e+00 -8.98231447e-01 -4.90608484e-01 -8.06913912e-01 1.01843560e+00 -1.05200934e+00 1.26250863e+00 7.90801108e-01 -9.38864768e-01 -7.49195337e-01 -1.29506278e+00 -2.06901953e-01 -7.74973869e-01 9.61144790e-02 7.61451840e-01 8.97488534e-01 -6.47904515e-01 3.82583171e-01 -1.42180055e-01 -6.60665095e-01 1.50898933e-01 3.45582396e-01 -4.36813712e-01 -4.55842316e-01 -1.32976186e+00 1.35063136e+00 -9.59008113e-02 -4.45552975e-01 -6.28212869e-01 -8.86557162e-01 -7.50843287e-01 -1.83814932e-02 1.04187541e-01 -4.68460381e-01 1.77913713e+00 -1.28199267e+00 -1.19066167e+00 8.61449897e-01 -3.24039668e-01 -4.89355236e-01 1.04494333e-01 -4.87143815e-01 -7.38666296e-01 -2.24107727e-01 3.67174387e-01 5.48009574e-01 8.52044046e-01 -9.38404322e-01 -8.98324490e-01 -2.75219679e-01 4.15477425e-01 3.54253441e-01 -2.87777573e-01 1.47864401e-01 -1.77762639e-02 -2.78598011e-01 -5.51651001e-01 -9.41584885e-01 -1.02023751e-01 -3.36758643e-01 3.56475681e-01 -3.30807239e-01 5.29089451e-01 -4.83256519e-01 1.14320898e+00 -2.09993052e+00 -5.89894176e-01 -7.50482082e-02 1.58247381e-01 4.95094925e-01 -3.53679091e-01 5.32191515e-01 -2.32782140e-01 5.64069510e-01 2.13258564e-01 -2.63334394e-01 2.58783605e-02 1.38732299e-01 -2.08248824e-01 1.88190639e-01 4.49329197e-01 6.66473150e-01 -6.57077789e-01 -1.34019151e-01 1.21192954e-01 -4.41952795e-02 -4.79029447e-01 -2.76357353e-01 -4.45574492e-01 -2.89325565e-01 -5.04362345e-01 4.84956175e-01 8.17450285e-01 -3.63806784e-01 1.96828529e-01 2.50747323e-01 2.49473304e-01 5.92160046e-01 -9.84831989e-01 1.24433780e+00 -7.73224235e-01 6.48822367e-01 5.35778441e-02 -7.27293611e-01 1.24037552e+00 5.27188964e-02 3.04192036e-01 -1.08967364e+00 4.30523045e-02 2.81951755e-01 3.02681476e-01 -2.05200046e-01 9.21598256e-01 -5.66555381e-01 -2.51103699e-01 9.28903997e-01 -2.05338508e-01 -5.34498453e-01 3.67507905e-01 3.08965951e-01 1.15966308e+00 -3.02935541e-01 4.24460262e-01 -3.01519722e-01 1.40244141e-01 6.61917686e-01 1.34198591e-01 5.75889766e-01 -3.96810383e-01 4.90425289e-01 7.68947899e-01 -3.95601809e-01 -1.09642088e+00 -5.73605418e-01 -3.45578462e-01 1.48049426e+00 1.60859991e-02 -5.46590388e-01 -4.73300606e-01 -9.15804148e-01 6.08036816e-01 9.20838296e-01 -7.54933715e-01 -1.38431191e-01 1.98529959e-01 -7.40827262e-01 2.70935297e-01 3.38510334e-01 1.15990415e-02 -7.26322055e-01 -6.01475053e-02 4.85112339e-01 2.51866341e-01 -1.06699431e+00 -3.96472067e-01 4.50460851e-01 -7.82604337e-01 -1.09403479e+00 -8.10508311e-01 -6.36278450e-01 2.55221188e-01 4.33256745e-01 1.70843995e+00 -1.66765183e-01 -1.54225320e-01 5.36765695e-01 -5.86076736e-01 -1.13441753e+00 -4.63744849e-01 5.12746453e-01 1.46144722e-02 -5.24955809e-01 1.48108017e+00 5.37076518e-02 -3.81655544e-01 7.15565145e-01 -8.92176330e-01 -3.79545212e-01 6.41714334e-01 7.24197328e-01 1.59947261e-01 1.28933132e-01 1.43145597e+00 -1.37263525e+00 1.41842437e+00 -7.63168514e-01 -3.07081908e-01 2.92571485e-01 -1.19944131e+00 -1.50833845e-01 4.45140898e-01 -5.33252835e-01 -1.22680247e+00 -2.42155388e-01 3.22961628e-01 2.73629725e-01 -8.22042748e-02 8.37502956e-01 4.16709095e-01 2.86428779e-01 1.11415064e+00 -4.75359440e-01 2.01992527e-01 -2.84331083e-01 3.79071653e-01 9.35318053e-01 -4.01488334e-01 -5.30061543e-01 4.12001044e-01 4.77587432e-02 -7.38069296e-01 -6.36020720e-01 -1.07771134e+00 -8.12502146e-01 -6.63627446e-01 9.80353877e-02 6.93926871e-01 -1.34601629e+00 -8.56688097e-02 2.14205563e-01 -8.74147475e-01 -3.68523985e-01 -1.66098207e-01 3.89504433e-01 -1.26681358e-01 1.52450621e-01 -5.14613450e-01 -6.13210499e-01 -2.39793494e-01 -8.23593378e-01 1.05129218e+00 3.57810467e-01 -9.86959457e-01 -1.31604958e+00 2.52167225e-01 4.40759510e-01 4.19621795e-01 -3.80702406e-01 1.06527507e+00 -1.22839069e+00 1.25183925e-01 -5.89720607e-01 -3.76632452e-01 8.10844481e-01 3.40129614e-01 2.03649223e-01 -8.20473850e-01 -5.29523119e-02 -4.08910923e-02 -6.29492819e-01 3.88875097e-01 6.01262331e-01 4.35479790e-01 3.72886568e-01 -2.68095195e-01 -3.30206633e-01 1.22520626e+00 3.75639796e-01 2.68858701e-01 4.56901640e-01 9.32152718e-02 8.25177908e-01 1.41077995e+00 5.73285401e-01 5.54301381e-01 1.87296733e-01 -2.10164547e-01 -1.31871387e-01 3.09484512e-01 -1.37653515e-01 3.72274280e-01 3.70202065e-01 5.95990717e-01 -1.81829751e-01 -9.46991682e-01 8.50436747e-01 -1.65385675e+00 -5.02533853e-01 -2.12025359e-01 1.86994576e+00 6.86202765e-01 7.19067276e-01 3.02761257e-01 -2.50066102e-01 4.87836301e-01 -2.02357620e-02 -1.01746929e+00 -1.28540254e+00 1.03257395e-01 3.45893145e-01 5.45273006e-01 3.83912116e-01 -7.72391319e-01 8.49503338e-01 7.19130850e+00 6.26718640e-01 -7.85996199e-01 -1.69942789e-02 7.90412068e-01 -2.81626433e-01 -4.54849780e-01 5.88679239e-02 -1.11240590e+00 2.62632906e-01 1.34769249e+00 -9.05371726e-01 1.28679544e-01 1.47273481e+00 2.37638459e-01 -5.41636109e-01 -1.47319448e+00 5.97595513e-01 1.28529873e-02 -1.01916945e+00 -1.02879077e-01 2.91621596e-01 1.05166078e+00 4.41994965e-02 -9.56024155e-02 1.00328398e+00 8.02770436e-01 -9.12905872e-01 3.18661094e-01 -1.76124096e-01 7.55250692e-01 -8.27594519e-01 8.83168459e-01 3.98904353e-01 -8.13970268e-01 -1.02298446e-01 -8.06198418e-01 -5.38221896e-01 1.53705686e-01 8.89989674e-01 -1.09563386e+00 3.18653345e-01 6.24513268e-01 1.11292958e+00 -6.27009451e-01 5.61798990e-01 2.63775885e-02 6.00916445e-01 -1.78152844e-01 -3.57318372e-01 2.33022064e-01 -1.90418199e-01 -3.19370449e-01 1.17708421e+00 -1.01423576e-01 -1.76140293e-03 8.77434239e-02 3.33238542e-01 -8.49959031e-02 8.58242214e-02 -8.61007929e-01 -3.42433035e-01 2.86458194e-01 1.27013385e+00 -4.47828472e-01 -5.08397996e-01 -7.79908657e-01 5.01274049e-01 1.42020017e-01 3.01099747e-01 -7.39632130e-01 -4.14810300e-01 8.92696142e-01 5.19298255e-01 2.98451781e-01 -8.71645659e-02 -9.68024194e-01 -7.86400497e-01 -9.74219218e-02 -1.25055158e+00 3.58084977e-01 -8.37601721e-01 -1.81270564e+00 3.86353374e-01 9.27885026e-02 -9.97669458e-01 -3.50028723e-01 -9.56435442e-01 -3.07100236e-01 1.09705842e+00 -1.62619030e+00 -7.05680788e-01 5.25152832e-02 3.51808667e-01 8.56796622e-01 -1.36601806e-01 7.86335826e-01 2.75685728e-01 -1.02842070e-01 6.02307975e-01 2.94623435e-01 -2.40270182e-01 1.48652160e+00 -1.28503001e+00 6.31224275e-01 3.14003527e-01 -3.28188911e-02 9.64465141e-01 4.90952998e-01 -8.31276000e-01 -9.60840225e-01 -5.54504871e-01 9.59508181e-01 -7.43288517e-01 9.95440483e-01 -2.79584050e-01 -1.01583374e+00 6.88330889e-01 4.19443339e-01 -7.54394710e-01 1.14309227e+00 9.42772508e-01 -4.14463073e-01 -3.04315418e-01 -1.28053832e+00 2.74097949e-01 3.79489034e-01 -7.74837196e-01 -8.22343588e-01 3.44091743e-01 4.73140687e-01 -1.89084962e-01 -1.21977079e+00 3.44676264e-02 7.35385895e-01 -6.55189991e-01 4.48207170e-01 -7.19367504e-01 1.02330291e+00 3.97107862e-02 2.67977506e-01 -1.68512642e+00 -3.43697160e-01 -1.13173842e-01 4.57900465e-01 1.25842357e+00 1.06518495e+00 -7.36619115e-01 8.92390549e-01 1.55139709e+00 1.78775713e-01 -5.62370002e-01 -3.69232267e-01 -5.73891342e-01 5.45296669e-01 -6.39373958e-01 3.61369342e-01 7.09240019e-01 2.02856541e-01 6.48834765e-01 4.54692207e-02 -2.78007060e-01 -3.13718542e-02 -3.85753028e-02 1.21838856e+00 -1.26060677e+00 -3.03033113e-01 -2.46893078e-01 -9.34928842e-03 -1.18363547e+00 1.98094413e-01 -3.58197242e-01 -2.81968359e-02 -1.85071230e+00 1.16359688e-01 -5.28005064e-01 -2.89959192e-01 4.70300078e-01 -1.64375395e-01 -1.53491557e-01 1.22249700e-01 2.63032652e-02 -4.25381601e-01 8.40916336e-02 1.15690196e+00 -1.16232261e-01 -2.55183280e-01 1.86774254e-01 -1.51383531e+00 5.53916633e-01 7.78065979e-01 -7.24678397e-01 -9.77174520e-01 -5.58478951e-01 4.39088672e-01 -1.56605616e-01 -2.94299304e-01 -4.69384432e-01 1.74451724e-01 -1.03987589e-01 5.25096416e-01 -5.76287031e-01 2.24317923e-01 -7.59466410e-01 -1.61855072e-01 -7.26604089e-02 -5.02424717e-01 4.23842072e-01 7.27045655e-01 6.78874850e-01 -3.82499337e-01 -5.90297282e-01 6.26685381e-01 -2.67520130e-01 -7.04293668e-01 -2.59952635e-01 -6.68063700e-01 2.89447755e-01 1.15526068e+00 -5.07770538e-01 -4.03503716e-01 -5.96041441e-01 -4.79044467e-01 2.46092156e-01 5.14305294e-01 9.12668645e-01 2.36574695e-01 -6.83438957e-01 -6.18186116e-01 7.98786132e-05 4.27872926e-01 -4.44758348e-02 1.34566575e-01 2.02184975e-01 -2.37069249e-01 6.99240506e-01 -1.32827267e-01 -5.55186152e-01 -1.21756339e+00 3.76266569e-01 6.29427060e-02 -8.74812543e-01 1.10251188e-01 7.00330794e-01 3.76466542e-01 -4.80200827e-01 -2.00086027e-01 -4.38243359e-01 -8.42731670e-02 4.86493677e-01 5.48194647e-01 1.23687305e-01 4.35232699e-01 -1.02962427e-01 -2.79111236e-01 5.04894674e-01 -9.91831362e-01 -2.70266265e-01 1.25495517e+00 -1.68012455e-01 2.99204171e-01 6.19239628e-01 9.99269485e-01 -2.35562921e-01 -1.09615088e+00 -1.82452068e-01 2.10178718e-01 -2.87188232e-01 -1.24743052e-01 -1.21712863e+00 -5.73200464e-01 6.64891362e-01 1.53044805e-01 4.91816729e-01 9.39715981e-01 -1.83227494e-01 3.93720090e-01 3.72918338e-01 4.14485216e-01 -1.43367207e+00 4.57479469e-02 5.74412584e-01 6.97422147e-01 -1.53934646e+00 4.32242721e-01 -1.25606343e-01 -1.43399465e+00 7.71449089e-01 1.00142539e+00 -3.39613110e-01 1.10031104e+00 1.04100212e-01 3.27453703e-01 -3.56856793e-01 -1.14305711e+00 1.45619407e-01 1.66941524e-01 7.31687665e-01 7.61350989e-01 1.48679093e-01 -4.91541326e-01 8.95822167e-01 -8.60024840e-02 3.67672473e-01 7.55482495e-01 8.44433725e-01 -3.19754601e-01 -1.39114857e+00 -3.34953144e-02 1.04852676e+00 -6.23677433e-01 -3.38688076e-01 -7.26076066e-01 1.18582332e+00 -2.64506191e-01 1.47722375e+00 -7.16954321e-02 -2.65437990e-01 6.10953748e-01 1.24495745e-01 -5.07844985e-02 -1.39114928e+00 -9.66990590e-01 -1.90845668e-01 6.35449290e-01 1.78648776e-03 -2.31749728e-01 -7.19045520e-01 -9.23664331e-01 -5.25284827e-01 -6.68024898e-01 2.50692338e-01 9.22646224e-01 8.98496091e-01 6.37436926e-01 3.08812886e-01 5.58925629e-01 -3.30980450e-01 -9.31123793e-01 -1.21208107e+00 -1.02434766e+00 4.40072179e-01 2.76378561e-02 -5.96094310e-01 -4.32244867e-01 1.12633355e-01]
[11.207175254821777, 6.873745441436768]
2ee6f724-f46f-4227-b6dc-a1b7ef17ab5f
morphological-word-embeddings
null
null
https://aclanthology.info/papers/N15-1140/n15-1140
https://www.aclweb.org/anthology/N15-1140
Morphological Word-Embeddings
null
['Hinrich Schütze', 'Ryan Cotterell']
2015-05-01
null
null
null
hlt-2015-5
['morphological-tagging']
['natural-language-processing']
[-2.44508207e-01 3.89024585e-01 -2.65282035e-01 -2.15905145e-01 -8.60921741e-02 -7.76765764e-01 4.48510379e-01 -7.23253429e-01 -5.48377395e-01 1.31954515e+00 3.66348401e-02 -9.49533224e-01 -2.40340635e-01 -1.05564880e+00 -8.44053447e-01 -8.75781775e-01 -7.42435038e-01 6.86515033e-01 1.44298598e-01 -6.52004302e-01 8.47113907e-01 6.29996777e-01 -1.62287033e+00 5.77558100e-01 6.81926727e-01 5.49597681e-01 5.66466339e-02 1.04480565e+00 6.13576770e-02 1.59847176e+00 -6.05450153e-01 -4.56729174e-01 1.43710867e-01 -1.88022718e-01 -5.35770595e-01 -3.36825520e-01 7.91536197e-02 -5.88986814e-01 -3.10633808e-01 6.07356608e-01 1.14270830e+00 -5.53557873e-02 1.07025254e+00 -1.51067197e+00 -7.97060013e-01 5.44234335e-01 2.30399013e-01 1.20632313e-01 6.10446513e-01 -1.79472014e-01 3.51212233e-01 -1.36621380e+00 6.32680655e-01 8.64870071e-01 9.47046995e-01 5.99273622e-01 -1.20647264e+00 -5.60552299e-01 -7.17137158e-01 -2.57835120e-01 -1.59567785e+00 -6.30042374e-01 6.23617955e-02 -3.34735900e-01 1.69142139e+00 6.72587395e-01 1.35647905e+00 1.47341585e+00 1.36649036e+00 4.72517729e-01 1.13904178e+00 -3.16340059e-01 3.65351558e-01 6.43754780e-01 -2.29148138e-02 5.51889002e-01 1.19686353e+00 6.76312149e-01 -5.60917914e-01 -7.92779744e-01 1.11365557e+00 -3.12949389e-01 2.60771513e-01 -7.06989706e-01 -1.06566191e+00 7.88466871e-01 -1.33808568e-01 2.20882341e-01 -3.46203089e-01 -9.92989019e-02 2.24548295e-01 4.12961632e-01 -2.94082850e-01 4.77862865e-01 -8.89795244e-01 -1.87686309e-01 -1.01472771e+00 1.91863477e-01 1.30615628e+00 1.45811689e+00 4.72722203e-03 3.70100170e-01 1.66493103e-01 2.89523482e-01 5.29429734e-01 1.05927479e+00 4.36951250e-01 -1.33365822e+00 -9.55119133e-02 4.20878716e-02 6.12505674e-01 -1.13473701e+00 -7.12930143e-01 -2.41300568e-01 -9.54724610e-01 5.18771529e-01 -1.59270123e-01 -4.59180593e-01 -7.59686172e-01 4.79171664e-01 -1.81984559e-01 -3.19489211e-01 5.45619786e-01 2.69257158e-01 7.28577793e-01 1.45816103e-01 9.25893057e-03 -4.73371416e-01 8.38296890e-01 -1.26899445e+00 -1.11426139e+00 2.33304992e-01 6.82289064e-01 -1.11704338e+00 4.06794518e-01 4.29345578e-01 -1.87200487e+00 -9.69799384e-02 -1.08317137e+00 1.19022481e-01 -7.97166348e-01 -2.57194549e-01 6.66624963e-01 1.49922979e+00 -1.63972652e+00 6.47176981e-01 -6.05021298e-01 5.40335439e-02 -4.73218322e-01 9.07614470e-01 -3.47142309e-01 2.90655226e-01 -1.28751957e+00 9.52449858e-01 -1.04219764e-01 3.03568155e-01 -3.30324143e-01 -1.84501112e-01 -1.03385210e+00 -6.24127567e-01 -2.59417385e-01 -1.07312346e+00 1.36302292e+00 1.09815992e-01 -1.30379498e+00 1.00962746e+00 -4.14315462e-01 -1.14109404e-01 6.61359608e-01 -1.52096570e-01 -8.18174660e-01 1.58321753e-01 2.87593510e-02 3.47163439e-01 5.30214727e-01 -9.69232559e-01 -6.31655395e-01 -1.00153096e-01 -2.54188716e-01 2.48816490e-01 1.43904146e-03 2.05205917e-01 6.38460100e-01 -9.53604504e-02 1.91062525e-01 -8.61961663e-01 -3.21419090e-01 -7.61404037e-01 -1.77353188e-01 -3.84736449e-01 5.59762537e-01 -2.92163283e-01 1.33848476e+00 -1.84365213e+00 -4.54683900e-01 2.94456817e-03 2.53527164e-01 3.89016122e-02 2.33101144e-01 1.23831403e+00 -2.43849114e-01 1.06667292e+00 3.98064762e-01 9.52541176e-03 3.13382268e-01 5.07273376e-01 2.86674555e-02 2.51206756e-01 9.43330750e-02 1.15280068e+00 -1.09214199e+00 -5.20940363e-01 5.57267785e-01 4.60968643e-01 -3.84588838e-01 6.24593735e-01 9.30012882e-01 2.99172718e-02 -8.94021094e-02 1.33241999e+00 1.05377018e+00 -2.90812284e-01 -2.46628653e-02 3.44544262e-01 -6.44541264e-01 6.00222833e-02 -7.41804183e-01 9.49429095e-01 1.31349964e-02 1.99757457e-01 1.74998537e-01 -6.80728197e-01 3.94654661e-01 1.10545933e+00 4.45659161e-01 -9.24297392e-01 -1.69569388e-01 9.00310338e-01 1.03920102e-01 -9.62983370e-01 6.40755475e-01 -1.72503099e-01 7.69825354e-02 1.40852645e-01 -3.61913115e-01 -3.84120375e-01 5.52458428e-02 4.19574350e-01 1.00224769e+00 -1.33707494e-01 5.48685491e-01 -8.72328460e-01 6.88946426e-01 -2.01589778e-01 3.00597936e-01 1.23978651e+00 -2.90862411e-01 9.21478033e-01 3.19022417e-01 -5.37574828e-01 -5.72780192e-01 -1.20098352e+00 -4.80556756e-01 9.86326516e-01 -1.58388391e-01 -3.87671381e-01 -7.10155964e-01 -2.63155133e-01 -1.85071185e-01 4.93483543e-01 -5.93591571e-01 5.23159266e-01 -5.14705896e-01 -6.85349941e-01 7.65277922e-01 3.17972392e-01 6.49624616e-02 -1.49228525e+00 -5.70813596e-01 3.34754556e-01 -1.07132711e-01 -5.39811492e-01 -1.57590687e-01 6.12205923e-01 -1.19366693e+00 -1.01526415e+00 -8.75530541e-01 -1.03603637e+00 7.18889177e-01 2.44572356e-01 1.18463767e+00 2.84042627e-01 -2.87709564e-01 6.49904788e-01 -1.31030381e-02 -2.67939389e-01 -3.16364132e-02 -5.78750819e-02 5.01899779e-01 -1.07851994e+00 6.46393359e-01 -3.92214775e-01 -1.00622165e+00 5.35368681e-01 -3.85821253e-01 -2.04611585e-01 7.85597622e-01 1.04281425e+00 1.42742008e-01 -1.33182362e-01 2.41971418e-01 -1.16735983e+00 1.11498392e+00 -4.85333532e-01 1.30341025e-02 2.62960047e-01 -8.53362501e-01 -4.53584343e-01 2.27881029e-01 -1.33037493e-02 -6.33760452e-01 -2.51423597e-01 -4.14545417e-01 3.22794706e-01 -4.06866640e-01 3.34393084e-02 2.24873364e-01 -5.26000679e-01 5.21005154e-01 1.50258854e-01 -4.07556817e-02 1.21735156e-01 6.30975887e-02 3.75316888e-01 3.67786407e-01 -7.01224983e-01 5.19809723e-01 1.93325520e-01 -3.00995037e-02 -1.28791058e+00 2.66786069e-01 -1.64763972e-01 -6.60750508e-01 -2.69126564e-01 7.75121748e-01 -9.50198710e-01 -6.35705650e-01 5.21708906e-01 -9.54387486e-01 -1.36340424e-01 -5.25335252e-01 6.30965471e-01 -7.02474535e-01 2.94757709e-02 -4.17852998e-01 -1.17976546e+00 -4.78847146e-01 -1.38869667e+00 9.59160328e-01 3.85633826e-01 -2.05811903e-01 -1.19790471e+00 3.38975668e-01 8.99848714e-02 4.29175526e-01 7.39965662e-02 5.03531873e-01 -2.80004859e-01 -2.98304528e-01 -2.85470933e-01 1.59573574e-02 -1.65876746e-01 -3.09365243e-02 4.80770379e-01 -5.48391879e-01 -4.37962860e-01 2.44738594e-01 -4.86500829e-01 -7.51581416e-02 7.11018145e-01 4.09077644e-01 -6.02920949e-01 -7.12544680e-01 5.56552589e-01 1.27709007e+00 4.72497940e-01 5.88340998e-01 6.77552521e-01 9.28869769e-02 5.17871320e-01 4.48484391e-01 2.08113909e-01 1.07506551e-01 1.08483166e-01 1.14217520e-01 -2.17137560e-01 9.45900455e-02 -1.75644696e-01 4.44304377e-01 1.09980536e+00 -8.23837519e-01 -2.15223029e-01 -4.88283783e-01 6.68056428e-01 -1.69156313e+00 -1.31570101e+00 -6.70152605e-01 9.76623476e-01 3.78074080e-01 3.38884830e-01 1.58041969e-01 1.21232189e-01 5.47049284e-01 -4.76693302e-01 -1.27305150e-01 -9.16352808e-01 -3.82233024e-01 4.21136379e-01 7.14838982e-01 6.51479006e-01 -6.10602558e-01 5.08361042e-01 1.25505590e+01 5.49717486e-01 -5.10860123e-02 1.27314463e-01 3.14815581e-01 1.99914098e-01 -2.24020526e-01 1.62632689e-01 -1.04165065e+00 9.26906094e-02 1.64754760e+00 -4.36735392e-01 3.93212438e-01 6.44444764e-01 3.94769877e-01 -1.96429595e-01 -9.78620768e-01 5.21999180e-01 1.12709433e-01 -1.50397491e+00 -3.76861334e-01 9.64888573e-01 5.80093563e-01 -1.81895152e-01 6.81941509e-01 4.33209062e-01 7.39425182e-01 -1.14519978e+00 7.88427055e-01 3.18626672e-01 8.87266219e-01 -8.18826377e-01 1.07315242e+00 2.76328743e-01 -1.04355454e+00 1.15487054e-02 -5.90443075e-01 -9.22581375e-01 1.62244976e-01 2.02188566e-01 -5.61583102e-01 4.06869203e-01 6.19966745e-01 2.15250537e-01 -8.53338003e-01 1.45581341e+00 -4.29231405e-01 4.22687940e-02 -3.00976187e-01 -4.39922899e-01 4.83290881e-01 5.28384298e-02 3.26606274e-01 1.02295005e+00 2.89785862e-01 5.45581102e-01 -2.62208551e-01 2.92286038e-01 6.63503349e-01 5.84383309e-03 -1.38496268e+00 -8.68683681e-02 3.54620427e-01 9.69217539e-01 -3.99494410e-01 -3.56748998e-02 -6.11456513e-01 6.62115753e-01 -1.66942239e-01 5.75146377e-01 -2.92761147e-01 -8.89406264e-01 4.33969021e-01 -2.30117049e-02 -2.88469017e-01 -1.35338321e-01 -3.23372036e-01 -1.02459288e+00 -6.06416404e-01 -3.90122980e-01 7.32179061e-02 -8.63867998e-01 -1.79313409e+00 6.85844719e-01 -3.50327820e-01 -1.03178740e+00 -9.29034173e-01 -9.72984850e-01 -2.90242791e-01 6.67755604e-01 -6.59234166e-01 -1.30476952e+00 1.77773401e-01 4.65253919e-01 1.82401046e-01 -5.88519633e-01 1.09945750e+00 -1.42670095e-01 -3.18711132e-01 1.08882773e+00 3.51253241e-01 -3.05974871e-01 8.32965672e-01 -1.16271269e+00 4.23785180e-01 1.09189451e-02 -6.59259140e-01 1.28460062e+00 6.81860268e-01 -6.81785107e-01 -1.16639841e+00 -1.52893141e-01 1.38110602e+00 -9.15031374e-01 5.27874053e-01 -1.02810413e-01 1.46300867e-01 6.57190859e-01 9.07351732e-01 -5.09461224e-01 1.08935249e+00 -1.13616258e-01 4.93123859e-01 8.23448122e-01 -1.38805819e+00 5.99272728e-01 1.21222687e+00 -4.81854081e-01 -8.27020168e-01 3.28937650e-01 9.94227231e-01 -5.60786903e-01 -1.34066856e+00 6.99696779e-01 1.14132881e+00 -1.04883564e+00 1.54237401e+00 -1.16912270e+00 4.08816934e-01 1.70770600e-01 -2.18455836e-01 -6.52935803e-01 -5.24122715e-01 -8.07741940e-01 -3.97186369e-01 3.40213925e-01 8.64144087e-01 -1.16800272e+00 5.53630650e-01 1.39166510e+00 -3.58625412e-01 -4.51715857e-01 -8.63314152e-01 -1.06149745e+00 -4.66870656e-03 1.81816339e-01 3.85523647e-01 8.18638563e-01 7.09820271e-01 2.02331603e-01 -2.05201268e-01 -4.37674336e-02 6.59316301e-01 -4.87512946e-01 5.82342803e-01 -8.74011815e-01 2.49149442e-01 -3.85976404e-01 -2.94489831e-01 -2.90531307e-01 -3.26795012e-01 -4.87107635e-01 -8.59293997e-01 -1.53084815e+00 8.49664882e-02 1.97367743e-01 -2.33256146e-01 3.44385266e-01 6.77823246e-01 2.45789096e-01 -2.27382466e-01 1.17054820e-01 -1.86927155e-01 -9.06788632e-02 9.18564737e-01 1.10557920e-03 -2.17724726e-01 3.92412305e-01 -1.01716614e+00 9.60534811e-01 7.85404518e-02 -6.09580100e-01 -5.51487744e-01 1.64939865e-01 5.71885824e-01 3.57775450e-01 4.49405432e-01 -9.70509052e-01 9.87445056e-01 -4.32846308e-01 6.12257063e-01 -1.43385470e+00 -2.65485227e-01 -9.00021493e-01 3.61826897e-01 6.29400074e-01 4.08042461e-01 3.53796333e-01 2.26811334e-01 -9.13986787e-02 -2.01378226e-01 -6.44658387e-01 4.43001747e-01 -7.45609999e-01 -4.55086857e-01 -3.90635371e-01 -1.02313221e+00 -9.67486948e-03 7.88939476e-01 -4.61360693e-01 -6.47767961e-01 -1.95199717e-02 -1.44608676e+00 7.30553791e-02 8.33678424e-01 1.34732248e-02 6.98022604e-01 -1.44444084e+00 -1.11748077e-01 6.18802965e-01 -5.42432606e-01 -4.98194307e-01 2.20674574e-01 1.06156552e+00 -1.31528914e+00 1.26915324e+00 -5.21896303e-01 -2.05424115e-01 -8.96585643e-01 7.19095051e-01 3.15837234e-01 1.75098896e-01 -3.93049836e-01 7.00151443e-01 1.12239346e-02 -6.58594191e-01 1.24011397e-01 2.97559589e-01 -4.87763733e-01 -2.52128989e-01 9.22277749e-01 1.17918742e+00 1.32526740e-01 -3.91715407e-01 -5.58191121e-01 4.39446121e-01 2.08604142e-01 -6.12734079e-01 1.38662541e+00 -2.20989168e-01 -7.44191706e-01 5.00318348e-01 7.50003338e-01 -7.33495504e-02 -3.28966863e-02 1.27011871e+00 4.80256975e-03 -5.47239244e-01 -3.11336160e-01 -7.82959402e-01 -3.42039734e-01 5.93214929e-01 6.17733955e-01 4.59080935e-01 9.69997942e-01 -6.47576571e-01 7.70057857e-01 7.64313757e-01 7.83133745e-01 -1.22997761e+00 -6.36951804e-01 3.69483382e-01 9.39836800e-01 -9.30032432e-01 1.30400375e-01 -3.14221263e-01 -5.55534840e-01 1.08689797e+00 5.38173854e-01 -1.62142932e-01 1.17091513e+00 6.96656525e-01 -1.75611794e-01 -4.53250527e-01 -1.04549897e+00 9.83817875e-02 2.10727617e-01 1.32608259e+00 8.26061428e-01 1.41275764e-01 -1.40426433e+00 8.46834600e-01 -6.34404242e-01 4.76996809e-01 9.47268784e-01 1.55018985e+00 -5.12773991e-01 -1.41692734e+00 -4.41162825e-01 2.31142953e-01 -2.82854587e-01 -3.49930972e-01 -8.62966061e-01 1.09929097e+00 1.97083559e-02 1.37330377e+00 -4.48758155e-01 -4.63731140e-01 8.27405751e-01 3.19600284e-01 4.41484600e-01 -1.25142381e-01 -8.94688129e-01 6.48750722e-01 3.84649158e-01 -1.05672014e+00 -7.94483483e-01 -1.27049911e+00 -1.08188879e+00 -1.16740942e+00 -3.07217389e-01 4.65777248e-01 2.31674016e-01 3.19533348e-01 -2.09303293e-02 1.56313553e-03 6.67447627e-01 -1.00111675e+00 -1.97397545e-01 -6.07395887e-01 -1.20775104e+00 -5.56946039e-01 3.94425273e-01 -5.85869551e-01 -1.15223050e+00 -6.24734610e-02]
[-1.5391836166381836, 15.869185447692871]
7d8f341e-9724-4c31-8e17-569f4a123a44
improving-speech-representation-learning-via
2210.13805
null
https://arxiv.org/abs/2210.13805v1
https://arxiv.org/pdf/2210.13805v1.pdf
Improving Speech Representation Learning via Speech-level and Phoneme-level Masking Approach
Recovering the masked speech frames is widely applied in speech representation learning. However, most of these models use random masking in the pre-training. In this work, we proposed two kinds of masking approaches: (1) speech-level masking, making the model to mask more speech segments than silence segments, (2) phoneme-level masking, forcing the model to mask the whole frames of the phoneme, instead of phoneme pieces. We pre-trained the model via these two approaches, and evaluated on two downstream tasks, phoneme classification and speaker recognition. The experiments demonstrated that the proposed masking approaches are beneficial to improve the performance of speech representation.
['Jing Xiao', 'Kexin Zhu', 'Ning Cheng', 'Jianzong Wang', 'xulong Zhang']
2022-10-25
null
null
null
null
['speaker-recognition']
['speech']
[ 7.11329520e-01 1.82967961e-01 -2.17383653e-01 -4.41949666e-01 -8.22437346e-01 -4.90971386e-01 6.08146727e-01 -4.49437946e-01 -2.92189389e-01 5.10869265e-01 6.68602347e-01 -6.87632978e-01 5.36151469e-01 -3.21021855e-01 -5.73040307e-01 -9.11190271e-01 7.18766227e-02 -1.78837270e-01 2.64715374e-01 -3.33207138e-02 3.24548811e-01 3.95072132e-01 -1.68895042e+00 9.02973831e-01 7.12136507e-01 6.48920834e-01 3.96768928e-01 7.41594017e-01 -3.91295552e-01 9.09991503e-01 -8.93093407e-01 1.73579097e-01 9.93782431e-02 -8.24402630e-01 -5.24394393e-01 2.19404951e-01 -1.49148434e-01 7.68314442e-03 -3.62558186e-01 1.00922000e+00 5.04705608e-01 3.39665532e-01 4.69425738e-01 -6.42961860e-01 -4.79056150e-01 1.03840029e+00 -3.75795811e-01 5.66791296e-01 3.11299384e-01 -1.17965095e-01 5.79015315e-01 -1.26462615e+00 -1.32634446e-01 1.50145686e+00 2.73214996e-01 9.01277363e-01 -7.68901110e-01 -8.62043202e-01 3.91376972e-01 2.65445560e-01 -1.41187620e+00 -1.19475675e+00 7.95137703e-01 -3.26589704e-01 1.22178030e+00 5.26180089e-01 2.37721205e-01 1.02082360e+00 -1.29739597e-01 6.31596446e-01 1.18297398e+00 -8.37347686e-01 1.04424037e-01 1.75411150e-01 3.19290578e-01 3.21691632e-01 -3.38424742e-01 5.49130440e-01 -7.30666757e-01 3.22729237e-02 5.74327171e-01 -2.12187633e-01 -4.27461088e-01 4.97995257e-01 -1.07420361e+00 6.02284133e-01 4.55890670e-02 5.64369082e-01 -2.73071051e-01 -6.42558932e-02 2.83595443e-01 2.53320038e-01 6.45004153e-01 -3.09943944e-01 -2.76110649e-01 -2.55675502e-02 -1.08336782e+00 -3.70179206e-01 4.97966975e-01 6.26793981e-01 6.02287948e-01 6.15425766e-01 -3.18852454e-01 1.14649284e+00 6.43712163e-01 1.74040154e-01 9.54576313e-01 -3.31524074e-01 8.72525871e-01 -1.17470967e-02 -8.22293162e-02 -4.39963788e-01 -6.46830648e-02 -3.26940656e-01 -8.51512372e-01 5.32660149e-02 1.26675554e-02 -2.50155330e-01 -1.45780361e+00 1.86462379e+00 7.89655298e-02 8.12934101e-01 3.59269857e-01 7.14969039e-01 1.01494575e+00 1.19794416e+00 8.35734680e-02 -6.22618794e-01 1.17387164e+00 -1.23325658e+00 -1.33826172e+00 -4.85099792e-01 4.30405915e-01 -1.08134937e+00 8.99870813e-01 3.11939806e-01 -1.20953119e+00 -1.02774334e+00 -1.05918324e+00 2.78913617e-01 -3.13156664e-01 2.52319276e-01 1.49223164e-01 1.10734224e+00 -9.21078563e-01 2.19005361e-01 -5.35342693e-01 3.06486130e-01 2.17542812e-01 3.34395438e-01 -8.91466960e-02 -2.19054930e-02 -1.31398249e+00 8.05114865e-01 4.59661603e-01 3.91792595e-01 -1.21926618e+00 -3.50238591e-01 -8.53261054e-01 3.86341244e-01 1.33391932e-01 -1.11959092e-02 1.31045997e+00 -1.13897312e+00 -1.81095111e+00 7.63660252e-01 -9.53159094e-01 -4.24565047e-01 1.65436581e-01 -2.58517683e-01 -8.48283231e-01 -1.97997838e-01 -4.51773435e-01 5.74096620e-01 1.45859051e+00 -1.36218858e+00 -4.53456610e-01 -4.23911959e-02 -4.09623146e-01 3.82351935e-01 -3.67504470e-02 3.37109268e-01 -2.53299862e-01 -9.92831588e-01 3.77410710e-01 -6.47303402e-01 -4.53821011e-02 -8.62792253e-01 -5.67697406e-01 -1.71358347e-01 9.86307263e-01 -1.04256511e+00 1.39741731e+00 -2.60506988e+00 5.05366661e-02 -1.37226135e-02 -3.76735359e-01 7.43593454e-01 -2.86784798e-01 4.84405398e-01 -5.79327822e-01 3.24349940e-01 -3.60162079e-01 -8.14301670e-01 -2.12434307e-01 2.76098162e-01 -8.54144275e-01 1.71411619e-01 2.14130938e-01 6.31032288e-01 -3.73034298e-01 -2.42367849e-01 4.35678363e-01 7.18199968e-01 -3.27622205e-01 4.37936932e-01 1.44898191e-01 6.68711901e-01 3.32332671e-01 4.87655193e-01 8.01755011e-01 6.35795534e-01 3.22975852e-02 5.37322275e-02 -6.36966676e-02 1.17638040e+00 -1.27097452e+00 1.26676810e+00 -3.29633325e-01 4.40610439e-01 1.79611236e-01 -9.45567906e-01 9.73960698e-01 8.53724957e-01 -1.41989114e-02 -2.80845940e-01 9.04890671e-02 3.26672234e-02 3.33159238e-01 -5.09044647e-01 2.10199356e-01 -4.48752224e-01 2.22964555e-01 2.61367798e-01 -1.65978000e-01 1.23570831e-02 -2.66511589e-01 -2.44556516e-01 8.29138517e-01 -2.26261571e-01 2.78798580e-01 6.44949526e-02 9.11767900e-01 -6.04417264e-01 6.09063804e-01 6.58843458e-01 -1.75855771e-01 7.53090680e-01 -2.47553317e-03 1.07232288e-01 -5.10584772e-01 -1.14647925e+00 1.45049870e-01 1.40193605e+00 -1.59744367e-01 -3.18814456e-01 -1.02741492e+00 -5.72653413e-01 -4.48811531e-01 8.58759463e-01 -2.06165344e-01 -4.88427252e-01 -9.54398930e-01 -6.09819233e-01 7.58377552e-01 4.85165715e-01 4.39475924e-01 -1.42144525e+00 6.32584393e-02 1.55897185e-01 -3.55437517e-01 -8.53326499e-01 -9.13608730e-01 5.36835313e-01 -8.11839402e-01 -4.59734797e-01 -6.37928307e-01 -1.23944962e+00 6.12494826e-01 6.89858794e-01 5.44015944e-01 2.81033933e-01 2.13440955e-01 -3.63807797e-01 -5.85917056e-01 -2.75170058e-01 -7.36865878e-01 -1.33049831e-01 1.31173819e-01 4.03215230e-01 2.67768472e-01 -5.51457584e-01 -1.54441223e-01 4.91211116e-01 -9.64567840e-01 -3.05603612e-02 7.19351113e-01 6.75365269e-01 4.91653264e-01 2.44958669e-01 8.92209291e-01 -7.94833660e-01 5.75796187e-01 -3.23706955e-01 -1.85013816e-01 -1.69102326e-02 -4.71206456e-02 -3.21957350e-01 6.02985203e-01 -8.34582508e-01 -1.19839275e+00 2.12313861e-01 -7.28333414e-01 -5.20300031e-01 -4.04692858e-01 5.57790279e-01 -6.51251018e-01 2.07642257e-01 3.10127765e-01 6.35192633e-01 -2.67665416e-01 -8.75576019e-01 3.44687492e-01 1.06283414e+00 5.21173894e-01 -1.26499265e-01 6.96790993e-01 6.40262812e-02 -7.57443964e-01 -1.06575716e+00 -5.94491124e-01 -6.56215131e-01 -5.18410504e-01 5.27944937e-02 7.45560706e-01 -1.03293562e+00 -9.72200856e-02 4.21781898e-01 -1.29962826e+00 -1.98282570e-01 -1.35084227e-01 6.92935944e-01 -1.02562733e-01 3.32848608e-01 -5.13225496e-01 -1.29297900e+00 -9.20589864e-02 -1.09374106e+00 7.43151307e-01 -7.83995166e-03 -1.33780286e-01 -5.55739641e-01 -2.75285337e-02 3.32314521e-01 3.25133443e-01 -5.80616295e-01 9.43584263e-01 -7.18649864e-01 -4.19361204e-01 1.86374169e-02 1.62858516e-01 7.99458802e-01 6.74646616e-01 -2.55909234e-01 -1.70267487e+00 -2.84266651e-01 5.24434805e-01 3.99745479e-02 1.14036298e+00 4.69416529e-01 1.36666346e+00 -4.76206332e-01 -3.15164030e-01 5.64125597e-01 8.07944715e-01 7.62393713e-01 1.04759419e+00 -3.00509274e-01 6.18991017e-01 6.46656930e-01 3.66858691e-01 -2.25127898e-02 -1.89638749e-01 5.80026984e-01 1.60412401e-01 -1.82552412e-01 -5.79923928e-01 -3.18494856e-01 8.54145050e-01 1.58631587e+00 1.31236508e-01 -4.91175950e-01 -6.46691740e-01 5.18925369e-01 -1.58176172e+00 -1.19452953e+00 9.43370014e-02 2.11277080e+00 8.75917733e-01 4.56718326e-01 -1.74174942e-02 6.64087534e-01 1.17692900e+00 5.73205709e-01 -3.18289667e-01 -6.74374998e-01 -2.28825748e-01 3.31207931e-01 1.04429349e-01 8.89187574e-01 -9.47331727e-01 1.17396164e+00 7.31695318e+00 9.56404150e-01 -1.09200609e+00 3.48461747e-01 5.09786665e-01 -6.65247962e-02 -3.29313338e-01 -6.31748838e-03 -9.92422700e-01 7.43099928e-01 1.08881342e+00 2.49245778e-01 6.89889789e-01 4.45774376e-01 5.11974037e-01 1.48006499e-01 -1.02749467e+00 1.02597475e+00 1.67613059e-01 -1.09198737e+00 2.57077605e-01 -1.19165517e-01 5.15204310e-01 -1.85908929e-01 -2.77956445e-02 5.97970665e-01 -2.30735764e-02 -1.44183218e+00 7.92358696e-01 2.33052135e-01 6.41504288e-01 -7.17565119e-01 5.14930904e-01 6.92656517e-01 -1.40588903e+00 -2.68626604e-02 -2.91800618e-01 -3.02475065e-01 2.87480235e-01 7.15994298e-01 -8.88586819e-01 3.31469178e-01 3.70054483e-01 2.84962773e-01 -2.77058065e-01 8.87843668e-01 -4.78993356e-01 1.36800945e+00 -7.59206191e-02 3.13602537e-01 -7.94698521e-02 -4.11481932e-02 4.26853299e-01 1.57674730e+00 1.89670518e-01 1.19313426e-01 6.60600439e-02 5.25095940e-01 -8.79443288e-02 -7.56557435e-02 -4.74085510e-01 -8.17879215e-02 8.54411364e-01 8.05765927e-01 -5.11886597e-01 -3.95012826e-01 -3.57681960e-01 8.85770082e-01 8.73238668e-02 5.50509930e-01 -1.01300478e+00 -4.04625922e-01 7.26325154e-01 -1.44764587e-01 5.89767337e-01 -2.69311935e-01 -3.90079439e-01 -9.51324582e-01 7.31298774e-02 -1.02745390e+00 1.17360994e-01 -5.24125755e-01 -9.38687384e-01 7.73411751e-01 -1.75327867e-01 -1.08691299e+00 -3.71531159e-01 -3.67677599e-01 -1.01718736e+00 1.28663671e+00 -1.64022064e+00 -7.54824758e-01 7.57342204e-02 4.78843212e-01 1.02128136e+00 -3.08769375e-01 7.24219382e-01 4.26428318e-01 -7.41058290e-01 6.32404685e-01 -1.28078982e-01 8.71902704e-02 4.88493562e-01 -7.24023402e-01 5.44114292e-01 1.10048544e+00 5.61847329e-01 8.04707706e-01 4.65721399e-01 -5.53555846e-01 -8.86817992e-01 -1.05792499e+00 1.11155736e+00 -1.05461106e-01 1.02764042e-02 -6.48829162e-01 -1.29784346e+00 8.70698929e-01 4.09038454e-01 -3.16289186e-01 9.55450356e-01 -1.66964814e-01 -2.12224036e-01 -2.35096947e-03 -9.17356670e-01 3.75638694e-01 8.99459243e-01 -8.57609868e-01 -1.21883810e+00 4.48247008e-02 1.06974530e+00 -3.19975615e-01 -1.05117016e-01 3.76925498e-01 2.15355068e-01 -7.99549878e-01 1.14928186e+00 -3.31656009e-01 4.60889284e-03 -5.73582292e-01 -5.35900354e-01 -1.50969195e+00 -2.01203451e-01 -7.58064091e-01 -2.97760457e-01 1.67301476e+00 4.70833808e-01 -7.05842912e-01 5.83899915e-01 -1.35859489e-01 -6.10443652e-01 -3.63354325e-01 -1.22019911e+00 -8.10088634e-01 -1.35848494e-02 -4.95609611e-01 7.56019890e-01 8.10825408e-01 -1.49329171e-01 4.03644830e-01 -4.70798910e-01 4.91711617e-01 -4.89084274e-02 -4.16488796e-01 3.74910176e-01 -6.81897640e-01 -4.26588774e-01 -4.31738138e-01 1.11367896e-01 -1.46157658e+00 2.44043157e-01 -8.36567402e-01 6.01981342e-01 -1.27953887e+00 -2.51953185e-01 -2.16240495e-01 -6.39428139e-01 5.39166152e-01 -3.97533178e-01 -7.11282110e-03 9.83908176e-02 3.23218703e-01 1.33698985e-01 5.61786413e-01 8.50485802e-01 -1.18286826e-01 -4.38381791e-01 3.31975907e-01 -6.24294817e-01 7.35386789e-01 9.71309125e-01 -7.30287194e-01 -6.39702141e-01 -4.23082262e-01 -5.49188972e-01 -4.44220454e-02 1.32693961e-01 -9.78768706e-01 6.47844421e-03 -1.19227506e-01 2.85902411e-01 -9.71650541e-01 7.72729397e-01 -5.54591596e-01 -7.48959742e-03 5.42514861e-01 -4.80847239e-01 -2.62209326e-01 4.65678185e-01 3.61898094e-01 -5.63420057e-01 -5.15319884e-01 9.60386157e-01 -1.37189701e-01 -5.27199924e-01 -2.35953346e-01 -8.53568256e-01 -3.26303184e-01 7.42674172e-01 -2.79397309e-01 -1.98558584e-01 -3.85023832e-01 -9.01272655e-01 -3.46160591e-01 -2.43149355e-01 5.12897372e-01 8.88220668e-01 -1.30108190e+00 -6.31497145e-01 7.11360812e-01 -3.90106708e-01 -1.46441951e-01 1.30883604e-01 6.16366625e-01 -6.64532706e-02 4.75221634e-01 2.11817265e-01 -4.90008593e-01 -1.52994740e+00 7.48852313e-01 2.58510441e-01 2.45336052e-02 -2.64249027e-01 1.10824978e+00 6.19064748e-01 -4.58100170e-01 7.98970461e-01 -4.48771477e-01 -4.51093167e-01 -9.01013240e-03 8.63769531e-01 2.40029514e-01 2.21512303e-01 -8.35525870e-01 -5.27746320e-01 4.92926985e-01 -2.12799937e-01 -5.03279507e-01 8.48014712e-01 -3.73406440e-01 1.58911236e-02 7.18532860e-01 9.69332218e-01 4.10883844e-01 -1.06334901e+00 -1.44752234e-01 3.66783887e-02 -4.71737653e-01 1.06136225e-01 -7.96000838e-01 -8.07538688e-01 1.11931193e+00 6.08480692e-01 4.70337749e-01 1.28220606e+00 -1.66900605e-01 7.36059785e-01 1.18936270e-01 7.33060613e-02 -8.37303162e-01 -8.70092586e-02 7.02277720e-01 8.81773293e-01 -9.12018716e-01 -4.66635048e-01 -8.54100168e-01 -6.10680282e-01 7.91685283e-01 4.81092036e-01 4.91608977e-02 7.49393880e-01 3.06374311e-01 4.12148774e-01 3.35492820e-01 -8.20515335e-01 -3.29188496e-01 4.45376068e-01 6.54076397e-01 6.03842854e-01 -4.75184880e-02 -8.12562332e-02 7.04235673e-01 -2.89085209e-01 -2.52692819e-01 2.81169921e-01 9.86762166e-01 -8.71968210e-01 -1.15729904e+00 -8.43131185e-01 1.06100596e-01 -4.03608203e-01 -4.52449471e-01 -6.58497155e-01 2.43709967e-01 3.80216330e-01 1.56487095e+00 -8.00061971e-02 -5.30792713e-01 2.43774474e-01 5.69374740e-01 1.48163646e-01 -1.08847320e+00 -6.63521588e-01 5.18086016e-01 1.07432278e-02 -1.58534721e-01 -2.77596980e-01 -4.55119073e-01 -1.32574916e+00 9.71401408e-02 -6.18997455e-01 3.26608092e-01 6.58740163e-01 1.23916185e+00 2.06717104e-01 9.16995645e-01 9.34449196e-01 -9.64395404e-01 -4.45227444e-01 -1.39113426e+00 -3.88337225e-01 1.03266776e-01 7.29499578e-01 -3.97488594e-01 -6.44239068e-01 4.23019499e-01]
[14.582365036010742, 6.263854503631592]
b3617fea-03a1-4521-a66c-c93d8fc2a8f4
dfki-dkt-at-semeval-2017-task-8-rumour
null
null
https://aclanthology.org/S17-2085
https://aclanthology.org/S17-2085.pdf
DFKI-DKT at SemEval-2017 Task 8: Rumour Detection and Classification using Cascading Heuristics
We describe our submissions for SemEval-2017 Task 8, Determining Rumour Veracity and Support for Rumours. The Digital Curation Technologies (DKT) team at the German Research Center for Artificial Intelligence (DFKI) participated in two subtasks: Subtask A (determining the stance of a message) and Subtask B (determining veracity of a message, closed variant). In both cases, our implementation consisted of a Multivariate Logistic Regression (Maximum Entropy) classifier coupled with hand-written patterns and rules (heuristics) applied in a post-process cascading fashion. We provide a detailed analysis of the system performance and report on variants of our systems that were not part of the official submission.
['Georg Rehm', 'Julian Moreno Schneider', 'Ankit Srivastava']
2017-08-01
null
null
null
semeval-2017-8
['rumour-detection']
['natural-language-processing']
[-5.79866096e-02 3.55804861e-01 1.83392204e-02 -2.21175313e-01 -4.00682390e-01 -4.35114056e-01 1.08827829e+00 5.22893846e-01 -4.73704994e-01 7.00664997e-01 4.12801772e-01 -7.57539392e-01 7.13193715e-02 -5.39869010e-01 -3.52125019e-01 -1.63609788e-01 -2.95786828e-01 4.57017243e-01 2.93848485e-01 -3.79780114e-01 7.57617176e-01 2.69779772e-01 -1.13930857e+00 8.23858619e-01 4.16051805e-01 9.75427687e-01 -4.34395403e-01 9.37606275e-01 1.59936070e-01 1.58366704e+00 -5.83119810e-01 -6.82200193e-01 4.64216061e-02 -3.76273751e-01 -1.14889848e+00 -1.79859236e-01 1.42450958e-01 -8.49937797e-02 -1.47936285e-01 9.03021038e-01 5.60087077e-02 -1.59630254e-02 6.34820879e-01 -1.43413699e+00 -5.78103602e-01 8.94367933e-01 -3.89676422e-01 5.23597658e-01 4.79763061e-01 1.01839870e-01 1.15870655e+00 -1.03122032e+00 9.41292584e-01 1.15807271e+00 9.20670927e-01 9.58712474e-02 -1.25421476e+00 -5.64053178e-01 -3.12350005e-01 6.10248327e-01 -1.13701773e+00 -7.05792427e-01 5.15272021e-01 -7.48648167e-01 1.35040069e+00 5.55126905e-01 3.89092326e-01 1.34146094e+00 4.09272701e-01 7.92744577e-01 1.63868809e+00 -7.28430003e-02 4.83682036e-01 6.57588363e-01 4.83758450e-01 4.16808248e-01 2.01960593e-01 -1.28002539e-01 -5.79149008e-01 -7.38525510e-01 6.83229491e-02 -3.70263815e-01 -6.33281469e-03 2.56893158e-01 -1.19489336e+00 1.18682289e+00 2.75554359e-01 3.86112720e-01 -9.51282561e-01 -2.26333722e-01 5.05927265e-01 7.91336179e-01 6.67374790e-01 6.09503090e-01 -5.20168781e-01 -3.62393230e-01 -1.25317967e+00 5.76774240e-01 1.67669296e+00 5.45760036e-01 4.60050583e-01 -3.04845035e-01 -1.63175061e-01 7.12999165e-01 1.99798092e-01 4.49735224e-02 5.01523316e-01 -7.80154943e-01 2.65152574e-01 3.31346571e-01 4.79844302e-01 -1.43619180e+00 -7.91203558e-01 -3.30304146e-01 -5.73785424e-01 -4.42224070e-02 3.19632500e-01 -4.47493494e-01 -3.92454594e-01 9.49334800e-01 1.56320766e-01 1.43488213e-01 7.35767558e-03 9.82607067e-01 8.29097271e-01 4.65637267e-01 -8.00673813e-02 -5.60127795e-01 1.37821102e+00 -8.71525824e-01 -7.40186334e-01 -4.97327000e-02 5.92646956e-01 -9.90809143e-01 3.57350379e-01 7.62531459e-01 -9.62776244e-01 2.46887863e-01 -1.06219172e+00 -1.34028941e-02 -3.55627060e-01 -3.88013154e-01 2.90164024e-01 3.11040103e-01 -9.93514895e-01 5.30318081e-01 -3.58810365e-01 -3.34328443e-01 1.68945730e-01 -3.02613139e-01 -1.45432591e-01 2.93784171e-01 -1.54130352e+00 1.16811097e+00 8.41257572e-02 -1.51187498e-02 -8.87830853e-01 -4.42493558e-01 -3.96364897e-01 -2.83066571e-01 3.95351291e-01 -3.07442069e-01 1.29572546e+00 -9.24948752e-01 -1.47924638e+00 1.08082974e+00 1.54890299e-01 -1.03796685e+00 1.02690053e+00 -1.15261801e-01 -5.45937717e-01 -5.50735146e-02 -2.73348782e-02 -7.37327412e-02 8.74812841e-01 -9.48554933e-01 -6.09294176e-01 -2.61868477e-01 -1.70104951e-01 -4.65588808e-01 1.53427850e-02 7.05488980e-01 3.08625162e-01 -5.27111828e-01 -1.19511597e-01 -8.71283948e-01 -4.97357398e-02 -7.65129387e-01 -5.62983751e-01 -4.11059588e-01 3.67322922e-01 -1.17060483e+00 1.72263014e+00 -1.47100341e+00 -4.28055748e-02 2.19735235e-01 4.22259718e-01 6.71334863e-02 8.04040954e-02 8.23347151e-01 1.58525005e-01 3.18584204e-01 5.88265657e-02 -3.21058959e-01 8.76620114e-02 -1.88811883e-01 -7.21537650e-01 6.18701816e-01 1.43764336e-02 6.14040971e-01 -1.06195998e+00 -3.89195144e-01 -3.75861585e-01 2.17486527e-02 -3.20129067e-01 1.77501030e-02 -4.69751298e-01 -4.27676253e-02 -3.80490899e-01 4.47560847e-01 3.44256431e-01 -4.28705662e-01 1.54523075e-01 2.18779936e-01 -3.67326230e-01 7.26517260e-01 -8.33221853e-01 7.76969433e-01 -3.19630444e-01 7.73420155e-01 1.18914202e-01 -4.45984304e-01 9.65752542e-01 4.07189637e-01 4.28294688e-01 -4.58022028e-01 1.04029894e-01 1.82099119e-01 -2.43385866e-01 -7.94792354e-01 8.59539866e-01 -2.07939342e-01 -1.73449725e-01 1.06143272e+00 -2.99702287e-01 5.36497198e-02 1.00967117e-01 5.21388769e-01 1.48788369e+00 -3.00672323e-01 7.33968198e-01 -1.20779850e-01 4.26301986e-01 3.44543517e-01 4.81845886e-01 9.23074007e-01 -3.62093776e-01 3.35579604e-01 1.05528343e+00 -6.96965814e-01 -1.41398442e+00 -3.51994008e-01 -2.16007516e-01 1.08180261e+00 -3.21216255e-01 -7.83039153e-01 -4.74174827e-01 -7.34679043e-01 2.93581754e-01 1.19181144e+00 -7.78679073e-01 3.26352447e-01 -3.95526707e-01 -8.92619193e-01 8.66721511e-01 -1.13236845e-01 2.84502029e-01 -1.16877615e+00 -5.64936161e-01 4.20542449e-01 -3.83276373e-01 -1.19116092e+00 1.81958854e-01 -1.13328919e-01 -4.51568455e-01 -1.08630574e+00 -1.21955790e-01 -1.07823588e-01 -1.66603848e-02 -7.86805004e-02 1.05180132e+00 4.95159060e-01 6.87586889e-02 -9.12701488e-02 -5.38273096e-01 -3.82499725e-01 -9.05419827e-01 -1.47012500e-02 6.21403717e-02 4.63898592e-02 5.22273064e-01 -6.18318737e-01 -4.20485616e-01 3.70770663e-01 -6.43236339e-01 7.65291005e-02 4.08179998e-01 7.74129987e-01 -4.45966832e-02 -2.85356045e-01 6.78578258e-01 -1.16725063e+00 1.17175245e+00 -1.44613516e+00 -3.81384850e-01 -4.04724404e-02 -9.73649621e-01 -3.55353683e-01 5.74521482e-01 -7.85621852e-02 -8.46606612e-01 -5.39528131e-01 -1.77279173e-03 -1.09655112e-01 -8.09588376e-03 7.51884818e-01 6.74926341e-01 3.32192242e-01 1.05567062e+00 2.92894095e-01 -8.11051112e-03 -6.04161382e-01 2.86932468e-01 1.31265068e+00 3.32553536e-01 -2.55888641e-01 6.02863312e-01 1.49200752e-01 -2.38115370e-01 -6.72642350e-01 -9.74922121e-01 -5.69577098e-01 -4.57299024e-01 -3.16872656e-01 4.57668394e-01 -8.48658323e-01 -7.36939430e-01 4.47310179e-01 -1.27699578e+00 -3.39997262e-01 2.12529987e-01 2.45177716e-01 -4.86105233e-01 2.19798490e-01 -1.02005339e+00 -1.00863421e+00 -5.51059902e-01 -4.84073877e-01 2.81753063e-01 -1.15548022e-01 -6.39208674e-01 -7.71184444e-01 3.66531193e-01 7.26521790e-01 7.27775633e-01 2.99950868e-01 5.60905933e-01 -1.42724526e+00 -1.90768927e-01 -4.18713480e-01 -3.27493966e-01 1.79120705e-01 -4.99259174e-01 9.38486457e-02 -9.23482776e-01 -1.04159489e-01 6.14073388e-02 -5.26668131e-01 9.84457493e-01 -2.81417280e-01 5.96353352e-01 -1.08203864e+00 -6.00016639e-02 -3.90439443e-02 9.48290169e-01 -7.43887842e-01 4.47932750e-01 9.23925042e-01 1.31734684e-01 6.67666495e-01 6.14917755e-01 1.01073754e+00 7.83883274e-01 7.09354818e-01 2.83785999e-01 6.67534649e-01 9.26465169e-02 -1.66376635e-01 6.40334427e-01 7.79826522e-01 -2.33050823e-01 -1.42927706e-01 -1.10269773e+00 5.32135665e-01 -2.11274958e+00 -1.15638483e+00 -8.15995753e-01 1.94137216e+00 1.09669673e+00 2.43969098e-01 3.78671050e-01 -8.82095844e-02 4.27723885e-01 2.76589990e-01 -6.78078085e-02 -8.63339782e-01 4.59523313e-02 -6.22894526e-01 4.07078117e-01 6.95506454e-01 -7.56203353e-01 9.36121047e-01 7.12792635e+00 4.67732280e-01 -8.60117733e-01 3.74351382e-01 3.66455853e-01 -2.40367413e-01 -3.77276808e-01 2.56456465e-01 -6.62495494e-01 6.64431393e-01 1.57306492e+00 -4.19953138e-01 7.17096865e-01 6.57966077e-01 3.88298064e-01 -2.11066917e-01 -9.08441782e-01 2.87304670e-01 2.27466583e-01 -1.49345112e+00 -4.42839742e-01 1.36977121e-01 6.67714059e-01 8.72044802e-01 -3.71727377e-01 4.67612207e-01 5.76172590e-01 -7.97981381e-01 1.08772945e+00 8.66780877e-01 1.05173238e-01 -3.08490098e-01 7.93973625e-01 9.23055828e-01 -5.06382324e-02 -1.73356444e-01 -2.69305319e-01 -3.97067249e-01 1.89484105e-01 1.13766015e+00 -1.42233253e+00 2.73392588e-01 5.99784613e-01 7.26319015e-01 -6.30881011e-01 8.05205584e-01 -4.87015396e-01 1.10301304e+00 -1.53388306e-01 -3.09397370e-01 1.93858489e-01 1.59382522e-01 1.30962861e+00 1.73267519e+00 -2.49582037e-01 -1.01117894e-01 9.96887609e-02 1.03253639e+00 -2.61248142e-01 4.96873036e-02 -4.25600596e-02 -9.94140804e-02 6.11358702e-01 1.44420576e+00 -4.17062968e-01 -4.67129827e-01 -2.34512508e-01 6.20087504e-01 4.70836520e-01 1.90022588e-01 -5.79489827e-01 -6.73985109e-02 4.68315810e-01 3.59570116e-01 2.62876093e-01 -1.08006671e-01 -5.90063870e-01 -1.23343563e+00 -3.13686877e-01 -9.79166269e-01 6.51733816e-01 -8.84084165e-01 -1.59475183e+00 6.72835350e-01 -2.14971587e-01 -6.56811893e-01 -5.43606043e-01 -1.01791479e-01 -6.49535716e-01 7.75718331e-01 -1.51939988e+00 -8.51418972e-01 -1.58102944e-01 4.16929692e-01 1.71136454e-01 -1.92925572e-01 7.19621241e-01 -6.05805367e-02 -6.42021000e-01 2.40521193e-01 6.54817298e-02 -4.07995470e-02 8.79423916e-01 -1.20363975e+00 4.58738118e-01 5.97204387e-01 -2.21498191e-01 5.93531013e-01 1.34175169e+00 -9.07452583e-01 -1.27155948e+00 -1.01227474e+00 1.68735015e+00 -5.85143864e-01 1.50352418e+00 -1.80618793e-01 -1.11664653e+00 9.24822807e-01 2.51645833e-01 -2.89844900e-01 5.76322794e-01 3.86667818e-01 -4.89297539e-01 3.76738042e-01 -1.10837615e+00 2.26253539e-01 2.96456009e-01 -3.29493046e-01 -8.91975164e-01 5.09660602e-01 5.59948444e-01 -2.56110609e-01 -1.11598372e+00 -4.56230901e-02 5.88687897e-01 -1.01369655e+00 3.39060277e-01 -9.32759881e-01 9.32699084e-01 -1.81575716e-01 -6.16336949e-02 -1.06440890e+00 -3.61542493e-01 -1.00467885e+00 -5.73200226e-01 9.39393342e-01 4.72014070e-01 -7.35858381e-01 2.05963865e-01 5.30417383e-01 2.60169059e-01 -6.18083298e-01 -6.99495077e-01 -5.14478147e-01 9.03878883e-02 -6.05635226e-01 2.91874051e-01 1.24134779e+00 6.26599729e-01 3.82171005e-01 -6.58888221e-01 -8.46824143e-03 5.77959061e-01 1.57221824e-01 8.36781323e-01 -1.11834586e+00 -2.06049547e-01 -6.24896407e-01 -2.97865808e-01 -4.59879279e-01 5.18173017e-02 -9.86414492e-01 -1.06016640e-02 -1.13053942e+00 3.92125160e-01 -3.64762455e-01 -3.36769432e-01 7.16687202e-01 -8.59073848e-02 2.69340664e-01 1.13839619e-01 8.26886535e-01 -8.72777402e-01 1.57530233e-01 4.76977587e-01 7.07039312e-02 -4.08531027e-03 2.62604088e-01 -7.88311422e-01 6.47570133e-01 7.19156265e-01 -8.68147254e-01 4.51752335e-01 1.17729411e-01 9.55923796e-01 2.14742541e-01 8.32575858e-01 -3.35072309e-01 4.93537962e-01 -2.56994694e-01 8.23067501e-03 -5.25740504e-01 -1.24156572e-01 -1.66930392e-01 1.65582057e-02 2.82206506e-01 -5.51871717e-01 1.11218272e-02 -1.65189683e-01 5.79285800e-01 1.59923779e-03 -3.15774947e-01 6.33804739e-01 1.80297103e-02 -3.04024339e-01 -2.29223847e-01 -8.07966411e-01 -5.98676614e-02 8.78466189e-01 2.89685935e-01 -8.40656757e-01 -4.88789171e-01 -7.85199106e-01 3.60099792e-01 4.98378366e-01 6.24998152e-01 3.80655110e-01 -8.48181069e-01 -1.43210042e+00 -2.00516701e-01 3.36918458e-02 -7.77970791e-01 -1.51484534e-01 1.64363122e+00 -3.89658481e-01 3.99505317e-01 -7.42517561e-02 -1.57883316e-01 -1.04652083e+00 4.80073243e-01 1.38787195e-01 -4.97860938e-01 -7.85309911e-01 4.78277057e-01 -7.71963716e-01 -2.18557373e-01 4.20253575e-02 3.86464447e-01 -4.60438997e-01 3.50029707e-01 1.02256823e+00 7.75412083e-01 2.77659029e-01 -6.08646810e-01 -4.16166216e-01 -5.69161475e-01 -3.17226231e-01 -1.77729458e-01 1.72566926e+00 -3.22097540e-01 -6.12772465e-01 7.37198710e-01 9.15375412e-01 7.54800960e-02 -7.28687644e-01 -6.00224793e-01 5.56431711e-01 -2.63373971e-01 4.81910855e-01 -1.39196908e+00 -3.86510223e-01 2.56939858e-01 -4.62790430e-01 8.42408955e-01 4.50677812e-01 -1.59064636e-01 8.99697363e-01 3.87360752e-01 2.70251721e-01 -1.18066990e+00 -3.23405445e-01 8.84892106e-01 1.37309027e+00 -1.13947880e+00 3.78816485e-01 2.85819452e-02 -9.72015381e-01 1.38485777e+00 5.97738139e-02 -1.01907142e-01 6.42358005e-01 1.87828615e-01 -2.62043881e-03 -4.02854949e-01 -1.63158405e+00 3.14211220e-01 2.39757299e-01 1.04765396e-03 4.66119111e-01 2.24038988e-01 -8.40987206e-01 1.00230277e+00 -4.77963150e-01 1.96686551e-01 1.06903684e+00 7.36377776e-01 -6.24408364e-01 -4.33533251e-01 -4.03557092e-01 4.61867869e-01 -5.54462731e-01 -2.73728758e-01 -8.27299118e-01 3.99727911e-01 -1.58178166e-01 1.17190528e+00 -2.37609133e-01 -7.74161100e-01 4.22319882e-02 9.60286483e-02 -1.57632411e-03 -2.82610625e-01 -1.04616821e+00 -2.76714712e-01 9.02998507e-01 -6.56738400e-01 -6.26249909e-02 -1.07843590e+00 -9.21621919e-01 -9.59629655e-01 -1.17276097e-02 2.41566777e-01 8.70287955e-01 1.09679317e+00 2.77828664e-01 9.48200896e-02 7.92770505e-01 -4.63260561e-01 -7.74902225e-01 -1.43477023e+00 -5.02699256e-01 2.62217820e-01 3.00195307e-01 -1.48727417e-01 -6.87320471e-01 5.31919673e-02]
[8.24385929107666, 10.124431610107422]
c5562fd6-5c55-4f2f-ae95-e617f2fe38e2
eco-evolutionary-tradeoffs-in-the-dynamics-of
2207.02014
null
https://arxiv.org/abs/2207.02014v1
https://arxiv.org/pdf/2207.02014v1.pdf
Eco-evolutionary tradeoffs in the dynamics of prion strain competition
Prion and prion-like molecules are a type of self replicating aggregate protein that have been implicated in a variety of neurodegenerative diseases. Over recent decades the molecular dynamics of prions have been characterized both empirically and through mathematical models, providing insights into the epidemiology of prion diseases, and the impact of prions on the evolution of cellular processes. At the same time, a variety of evidence indicates that prions are themselves capable of a form of evolution, in which changes to their structure that impact their rate of growth or fragmentation are replicated, making such changes subject to natural selection. Here we study the role of such selection in shaping the characteristics of prions under the nucleated polymerization model (NPM). We show that fragmentation rates evolve to an evolutionary stable value which balances rapid reproduction of $PrP^{Sc}$ aggregates with the need to produce stable polymers. We further show that this evolved fragmentation rate differs in general from the rate that optimizes transmission between cells. We find that under the NPM, prions that are both evolutionary stable and optimized for transmission have a characteristic length of $3n$, i.e three times the critical length below which they become unstable. Finally we study the dynamics of inter-cellular competition between strains, and show that the eco-evolutionary tradeoff between intra- and inter-cellular competition favors coexistence.
['Alexander J. Stewart', 'Saul Acevedo']
2022-07-05
null
null
null
null
['epidemiology']
['medical']
[ 4.39162970e-01 -1.09705485e-01 4.77602668e-02 1.54799938e-01 3.02156955e-01 -5.61447978e-01 5.79867423e-01 3.00906032e-01 -6.97891653e-01 1.19426787e+00 -1.43625259e-01 -1.73536241e-01 5.57480343e-02 -7.40924418e-01 -6.69322133e-01 -1.09876275e+00 -6.88070059e-01 6.00592017e-01 5.87128460e-01 -3.86625022e-01 1.50388464e-01 4.40766662e-01 -1.20767295e+00 -2.31964082e-01 1.04182732e+00 3.35901856e-01 4.72558945e-01 8.33946109e-01 1.28388554e-01 5.70524558e-02 -9.08331692e-01 -3.07897151e-01 1.28921255e-01 -4.69501853e-01 -2.52552897e-01 -4.42852937e-02 -5.28364122e-01 1.58902388e-02 -6.22023595e-03 7.93107688e-01 4.84281480e-01 -2.97399789e-01 1.00043571e+00 -8.68187487e-01 -1.00106013e+00 -6.55401871e-02 -6.17053926e-01 5.78100920e-01 1.97728038e-01 5.18916070e-01 6.99481606e-01 -1.85821429e-01 1.06420791e+00 1.06469655e+00 4.21286672e-01 6.60687447e-01 -1.50511563e+00 -7.12046400e-02 -2.79647738e-01 -2.30624393e-01 -8.49219501e-01 -4.41419333e-01 3.30520600e-01 -9.40968752e-01 1.22674084e+00 1.08444393e-01 1.23973370e+00 7.98777163e-01 1.32629108e+00 -5.19212224e-02 1.26094484e+00 -3.59024704e-01 3.95743847e-01 -4.12502855e-01 -6.38935203e-03 5.30525744e-01 9.36054826e-01 9.75974724e-02 -4.19764400e-01 -7.46094108e-01 9.33078825e-01 -2.24029630e-01 -5.56908727e-01 -1.79728493e-01 -9.72621441e-01 6.42773926e-01 -1.41384661e-01 3.43366325e-01 -5.62956095e-01 3.57416458e-02 1.70570761e-01 1.82868704e-01 2.09920540e-01 5.37301600e-01 -6.05452180e-01 -2.43342295e-02 -3.58562559e-01 2.50022858e-01 6.72889292e-01 3.79832298e-01 4.78785723e-01 -4.26207662e-01 1.83543399e-01 8.27845752e-01 2.49022797e-01 6.97083890e-01 3.20077002e-01 -9.18722808e-01 6.28586486e-02 4.97995645e-01 3.86976063e-01 -4.67380762e-01 -4.32084411e-01 -3.05573761e-01 -8.51985991e-01 3.04893076e-01 6.06680453e-01 -2.73967355e-01 -4.70760047e-01 2.01613832e+00 -7.50781000e-02 -4.79008287e-01 -6.07882440e-02 3.57071757e-01 -4.30852145e-01 7.60511339e-01 2.69467622e-01 -1.00870478e+00 1.28110564e+00 -1.81355491e-01 -3.42907250e-01 -1.73989192e-01 1.41981557e-01 -4.24486160e-01 7.28582621e-01 -4.02958766e-02 -1.17997849e+00 3.17030013e-01 -9.65738118e-01 5.91552377e-01 -2.36836344e-01 -8.34472418e-01 2.69903570e-01 6.96255505e-01 -1.18199694e+00 7.32698679e-01 -9.46137488e-01 -8.80441189e-01 2.19999909e-01 3.50177050e-01 8.89375284e-02 3.10898215e-01 -1.06457210e+00 1.07695699e+00 -1.20252125e-01 -2.72590280e-01 -3.18033278e-01 -3.31585616e-01 1.32834036e-02 -2.51338303e-01 -2.78563291e-01 -1.16081750e+00 6.42559826e-01 -5.74961722e-01 -1.18423593e+00 9.25897181e-01 -2.67491370e-01 -4.70550865e-01 2.52580404e-01 1.29658297e-01 -1.71625644e-01 3.00282747e-01 3.10133994e-01 4.30917263e-01 3.18968982e-01 -1.06730151e+00 -1.27983883e-01 -5.45521319e-01 -3.07175159e-01 1.39799893e-01 2.43019581e-01 4.01924193e-01 2.89058894e-01 -4.77627575e-01 -2.14224577e-01 -1.19284725e+00 -3.60730380e-01 -3.50529701e-02 -1.67224899e-01 -1.91461086e-01 3.30446541e-01 -4.10224885e-01 5.34180045e-01 -1.71594405e+00 3.42466474e-01 7.76490793e-02 5.28938591e-01 1.65886551e-01 1.01215601e-01 7.22048283e-01 3.19079101e-01 4.30114388e-01 -6.66650653e-01 4.02866185e-01 -2.78408647e-01 1.12164661e-01 8.68380517e-02 6.61133349e-01 2.80779839e-01 7.80416906e-01 -9.15046155e-01 -3.32765818e-01 -3.79906267e-01 5.91976583e-01 -3.58884156e-01 -8.38487521e-02 -3.49133164e-01 3.70744884e-01 -2.86807120e-01 4.68711078e-01 5.21970093e-01 -3.93475473e-01 5.81626952e-01 5.98310173e-01 -4.06782120e-01 7.07275420e-02 -1.44027546e-01 5.57674706e-01 3.17126215e-01 4.04485852e-01 2.83813030e-01 -4.60569382e-01 5.41287720e-01 1.52226746e-01 4.26040322e-01 -3.23050439e-01 1.02858089e-01 3.08483511e-01 6.79460943e-01 -1.86155930e-01 1.81091830e-01 -5.25344431e-01 2.52114624e-01 7.56611109e-01 -2.52390683e-01 1.38054281e-01 5.13474226e-01 1.86752856e-01 1.43950307e+00 -3.30596924e-01 3.51083905e-01 -4.77345139e-01 2.43612811e-01 5.11470437e-02 8.42869043e-01 6.43220901e-01 -6.49587274e-01 9.60822850e-02 6.61980569e-01 -4.40176809e-03 -1.40958190e+00 -1.28358734e+00 -2.97961622e-01 7.27505386e-01 3.69260103e-01 -1.17299922e-01 -8.25130105e-01 3.12019587e-01 -1.01084970e-02 3.13719302e-01 -4.98380721e-01 -5.50336301e-01 -6.95589244e-01 -1.55685318e+00 4.90813702e-01 6.69605359e-02 3.47524136e-01 -1.21065438e+00 -9.67952132e-01 3.86726111e-01 7.52325952e-02 -5.80516398e-01 -3.82745773e-01 3.56269300e-01 -1.09975708e+00 -1.06392550e+00 -9.92103517e-01 -3.25469106e-01 7.70440757e-01 1.25728071e-01 9.35153842e-01 6.56077504e-01 -4.41456854e-01 1.89237595e-01 -7.51410648e-02 -2.08578810e-01 -9.33768094e-01 -6.96572140e-02 5.47515929e-01 -6.52307451e-01 1.49000101e-02 -1.02645755e+00 -6.51167095e-01 2.82322705e-01 -6.64247274e-01 -2.38985628e-01 6.54130340e-01 5.58807611e-01 4.49911654e-01 6.34773225e-02 7.43046761e-01 -3.22487682e-01 1.21654713e+00 -5.21416426e-01 -3.70669544e-01 4.75420177e-01 -6.05144560e-01 8.83774906e-02 5.27737081e-01 -3.01837772e-01 -8.59587669e-01 -4.16360646e-01 3.18589061e-01 5.33878267e-01 1.71720073e-01 2.03788728e-01 -1.12956896e-01 -3.49228568e-02 4.30761516e-01 6.39765203e-01 4.46095377e-01 -1.23213172e-01 -5.88650368e-02 3.98423493e-01 -6.45910215e-04 -6.98203683e-01 5.68400562e-01 4.49071467e-01 1.50217757e-01 -1.32705092e+00 2.09816024e-01 2.25766659e-01 -4.07044023e-01 -3.09494168e-01 9.72681224e-01 -3.58530015e-01 -6.98403537e-01 9.23058629e-01 -1.28939879e+00 -5.58323920e-01 -2.15926126e-01 1.33865386e-01 -8.48610759e-01 5.77306151e-01 -1.04159939e+00 -9.73396122e-01 -2.01055497e-01 -7.86960244e-01 4.53361064e-01 1.76278248e-01 -1.58525229e-01 -7.41743445e-01 7.54076600e-01 1.86518088e-01 5.32491744e-01 3.41999084e-01 1.39243412e+00 -1.73514083e-01 -7.23532796e-01 3.30599695e-01 5.48722297e-02 -3.18133891e-01 2.17584491e-01 1.98765919e-01 -1.48645658e-02 -1.56497717e-01 2.17933878e-01 2.79337168e-01 7.29495764e-01 6.64789259e-01 -2.93447345e-01 -3.75254393e-01 -6.91653252e-01 2.62025774e-01 1.30612218e+00 7.43425906e-01 6.55674577e-01 5.98816633e-01 -1.57410905e-01 6.59696579e-01 7.13316724e-02 1.38298631e-01 4.04455513e-02 4.85269010e-01 1.09700887e-02 4.15255845e-01 2.69279122e-01 4.26591218e-01 5.57453096e-01 8.99441779e-01 -5.69813311e-01 -2.80182213e-01 -9.04168725e-01 3.34706843e-01 -1.56991839e+00 -1.23301041e+00 1.05133578e-01 2.19871187e+00 1.16505325e+00 4.26520467e-01 6.62414312e-01 -4.81819123e-01 1.08411324e+00 -1.75841659e-01 -9.13798094e-01 -4.57009733e-01 -7.31088758e-01 6.12676963e-02 8.55586648e-01 6.66349649e-01 -2.33804911e-01 6.13296807e-01 8.05064011e+00 5.92050664e-02 -8.21942747e-01 1.48809860e-02 4.24479783e-01 -1.62541881e-01 -1.62191585e-01 -9.85957906e-02 -4.04648721e-01 8.96367371e-01 1.05694199e+00 -5.47024608e-01 5.65496385e-01 1.28559038e-01 4.47689146e-01 -4.05680507e-01 -5.39617360e-01 1.86282426e-01 -4.30924773e-01 -1.38856709e+00 -1.04722977e-01 6.86526835e-01 4.86349732e-01 2.27600291e-01 -2.51914322e-01 -3.79546285e-01 5.87513566e-01 -5.63370049e-01 5.17534196e-01 7.15559483e-01 5.04553080e-01 -5.20733476e-01 3.63184869e-01 4.48938906e-01 -8.17731678e-01 7.12012649e-02 -4.37415093e-01 -1.46933287e-01 7.67899752e-01 9.51772094e-01 -6.29967928e-01 -2.21519962e-01 2.67440259e-01 5.61525337e-02 -5.12081385e-01 1.00046778e+00 1.32839888e-01 1.89205676e-01 -3.32741350e-01 -1.74613968e-01 -4.30135816e-01 -7.12505698e-01 1.10193586e+00 7.32360065e-01 9.18544922e-03 1.96097031e-01 -4.92569238e-01 1.13155258e+00 1.19243830e-01 -2.72762507e-01 -5.09042919e-01 -4.25962687e-01 6.20768964e-01 5.60986698e-01 -1.14156103e+00 -1.72744438e-01 1.70498461e-01 9.49595988e-01 2.46412292e-01 2.59239495e-01 -6.17012143e-01 -1.99141204e-01 1.11374950e+00 4.05619830e-01 2.10886180e-01 -6.73952579e-01 3.43144424e-02 -9.87359881e-01 -8.37796032e-02 -4.21675742e-01 -3.23368758e-01 -4.97351795e-01 -1.38340795e+00 4.54673856e-01 -2.32326820e-01 -3.33119959e-01 4.71885242e-02 -3.62280905e-01 -7.19743252e-01 6.60651624e-01 -8.55355442e-01 -3.93156350e-01 5.76863170e-01 1.68858737e-01 9.56757963e-02 5.77339195e-02 4.70007569e-01 -2.18685776e-01 -7.26408839e-01 -3.88219357e-02 6.84545815e-01 -4.86640722e-01 4.17423725e-01 -9.71232533e-01 6.35927379e-01 6.71498060e-01 -6.53701067e-01 1.24482036e+00 1.05916786e+00 -1.46989894e+00 -1.14719224e+00 -7.49374509e-01 7.47136533e-01 -2.42147699e-01 7.23929882e-01 -4.88306254e-01 -5.30826688e-01 3.61829430e-01 2.02293620e-01 -6.17477477e-01 8.04564118e-01 -1.52570829e-01 -5.34799807e-02 5.44143081e-01 -1.43099487e+00 9.15164411e-01 1.26839638e+00 -2.97724247e-01 -4.95307624e-01 4.55248892e-01 8.70942056e-01 5.42989790e-01 -6.58403277e-01 -6.29273206e-02 8.35788965e-01 -1.20119953e+00 8.78359497e-01 -5.72429597e-01 2.42505908e-01 -3.89434069e-01 -7.04223812e-02 -1.14047360e+00 -5.20297527e-01 -6.91256702e-01 1.44743666e-01 8.06634068e-01 3.86698067e-01 -1.13261545e+00 4.46123064e-01 4.08614337e-01 3.88691127e-01 -9.09739077e-01 -1.13758945e+00 -1.23100591e+00 3.56135428e-01 7.67442882e-01 3.05305332e-01 5.33321738e-01 1.90285593e-01 2.94684947e-01 2.39502769e-02 -3.91389698e-01 9.80729699e-01 -3.91025841e-01 1.25252277e-01 -1.40509129e+00 -3.10842842e-01 -5.42308867e-01 -4.49140996e-01 -3.95945042e-01 -1.67962343e-01 -4.17921633e-01 2.63000652e-02 -1.41939807e+00 5.06250262e-01 -4.36634362e-01 -2.91688088e-02 -2.42535695e-01 4.39409679e-03 -1.72868788e-01 3.81297708e-01 8.46511304e-01 -2.45956808e-01 3.10726911e-01 9.47519779e-01 2.10963681e-01 -5.12524605e-01 -3.73460442e-01 -6.98766708e-01 5.90102553e-01 1.08049130e+00 -4.64827001e-01 -7.72572756e-02 7.30506778e-02 4.66086358e-01 -1.87082659e-03 6.95641115e-02 -9.06728506e-01 -1.73260286e-01 -4.30436939e-01 7.52859935e-02 -1.85904980e-01 3.21790099e-01 -4.05028969e-01 4.05633181e-01 1.06175804e+00 1.24896578e-02 2.69691616e-01 -3.61624211e-01 9.23832119e-01 8.45836699e-01 -1.18720800e-01 1.03610504e+00 -3.11914414e-01 3.58482748e-01 4.05616947e-02 -1.45197427e+00 1.28889218e-01 1.18830013e+00 -6.23860538e-01 -8.68801236e-01 -1.32884935e-01 -5.92675030e-01 -9.96059254e-02 1.41047359e+00 -2.11905599e-01 2.01263234e-01 -8.07936490e-01 -4.73524421e-01 -3.23309809e-01 -4.20545787e-01 -7.46565044e-01 -9.25733298e-02 9.31418896e-01 -8.73509884e-01 3.43740195e-01 -7.20482230e-01 -4.00083989e-01 -1.04836953e+00 5.81820965e-01 4.31056052e-01 -4.34783667e-01 -3.75661343e-01 5.97114503e-01 1.58795059e-01 3.19331177e-02 -5.69828451e-01 3.65540422e-02 1.42991498e-01 -1.26192048e-01 3.97326291e-01 4.92414802e-01 -4.64972496e-01 -8.03166032e-01 -5.92071354e-01 5.12275040e-01 -1.92465097e-01 -1.85980424e-01 1.47126639e+00 -3.58842105e-01 -1.00780642e+00 5.11457503e-01 4.82125372e-01 1.81148738e-01 -1.18109608e+00 5.74253678e-01 1.30678266e-01 -6.87565580e-02 -8.79550874e-01 -5.94059348e-01 -5.82560539e-01 1.92380995e-01 3.23499262e-01 5.28010547e-01 6.74299240e-01 3.02756995e-01 1.06244814e+00 2.39660561e-01 6.55981958e-01 -8.99548292e-01 -3.75950933e-02 4.62439716e-01 7.01301038e-01 -2.96270281e-01 4.62132134e-02 -5.16521335e-01 -2.05304250e-01 5.98736763e-01 2.50743628e-01 -4.20628697e-01 4.74965721e-01 2.80397356e-01 -5.30587614e-01 -2.38750786e-01 -1.08318996e+00 1.54607043e-01 -6.97408199e-01 1.16316450e+00 3.44251394e-01 2.77093977e-01 -1.18239844e+00 3.72491598e-01 -5.89442924e-02 4.88779396e-02 9.16861117e-01 1.26815832e+00 -1.03581142e+00 -1.39564753e+00 -4.20091867e-01 6.67750776e-01 -3.78185689e-01 -1.21365324e-01 -1.06003749e+00 3.26849073e-01 3.05856615e-01 7.92815268e-01 7.94972256e-02 6.57733977e-02 -1.71191305e-01 2.77633220e-01 8.44208896e-01 -3.98595929e-01 -5.09899855e-01 -1.54691994e-01 2.39676517e-02 1.18412510e-01 -4.86549348e-01 -9.66812193e-01 -1.26657903e+00 -4.46610957e-01 -4.24623430e-01 2.31055990e-01 2.04024896e-01 8.73590589e-01 5.30357897e-01 1.10861279e-01 1.27335668e-01 -5.57122707e-01 -2.17451125e-01 -5.45829535e-01 -1.08384383e+00 -3.31904665e-02 2.10180417e-01 -7.01077044e-01 -5.47365963e-01 2.19591543e-01]
[5.631960391998291, 4.299342632293701]
2c1bc851-f452-4c18-8a61-d2f9a947595c
seeing-dynamic-scene-in-the-dark-a-high
null
null
http://openaccess.thecvf.com//content/ICCV2021/html/Wang_Seeing_Dynamic_Scene_in_the_Dark_A_High-Quality_Video_Dataset_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Wang_Seeing_Dynamic_Scene_in_the_Dark_A_High-Quality_Video_Dataset_ICCV_2021_paper.pdf
Seeing Dynamic Scene in the Dark: A High-Quality Video Dataset With Mechatronic Alignment
Low-light video enhancement is an important task. Previous work is mostly trained on paired static images or videos. We compile a new dataset formed by our new strategy that contains high-quality spatially-aligned video pairs from dynamic scenes in low- and normal-light conditions. We built it using a mechatronic system to precisely control the dynamics during the video capture process, and further align the video pairs, both spatially and temporally, by identifying the system's uniform motion stage. Besides the dataset, we propose an end-to-end framework, in which we design a self-supervised strategy to reduce noise, while enhancing the illumination based on the Retinex theory. Extensive experiments based on various metrics and large-scale user study demonstrate the value of our dataset and effectiveness of our method. The dataset and code are available at https://github.com/dvlab-research/SDSD.
['Jiaya Jia', 'Bei Yu', 'Jiangbo Lu', 'Chi-Wing Fu', 'Xiaogang Xu', 'RuiXing Wang']
2021-01-01
null
null
null
iccv-2021-1
['video-enhancement']
['computer-vision']
[ 1.65046439e-01 -7.28834510e-01 -1.07157156e-01 -3.68971527e-01 -4.66564208e-01 -6.58665776e-01 2.53852129e-01 -7.84780860e-01 -1.45205006e-01 4.50849950e-01 1.67579129e-01 5.91851994e-02 8.64980817e-02 -3.27429116e-01 -7.09679008e-01 -7.29315102e-01 3.26694809e-02 -5.28455734e-01 3.62315655e-01 3.24445143e-02 2.85286784e-01 1.33661255e-01 -1.66222680e+00 1.59624949e-01 5.70089519e-01 9.10161257e-01 3.37736279e-01 7.83314705e-01 6.70695305e-01 8.77076328e-01 -1.20936915e-01 -4.91567254e-02 8.08404863e-01 -5.36011994e-01 -3.35698158e-01 3.82188082e-01 6.46340311e-01 -6.24762058e-01 -7.64523089e-01 1.11424863e+00 7.65154839e-01 1.03459664e-01 -1.17082079e-03 -1.42669582e+00 -4.47127223e-01 -8.35056379e-02 -7.69586980e-01 3.40343505e-01 6.03141963e-01 6.35258198e-01 4.28407818e-01 -9.05530930e-01 5.77012300e-01 7.81404912e-01 5.51529527e-01 5.49423635e-01 -9.55432892e-01 -9.02073145e-01 -2.10672691e-02 3.32710594e-01 -1.39151466e+00 -1.00018191e+00 9.90962982e-01 -3.55807424e-01 2.18199566e-01 1.67972103e-01 7.90132761e-01 1.05252838e+00 1.62174463e-01 6.86840057e-01 1.15392363e+00 -4.86305267e-01 3.29916216e-02 -2.22056970e-01 -1.55138686e-01 7.48334706e-01 5.22467531e-02 4.50593114e-01 -6.20716751e-01 6.28977045e-02 1.10928380e+00 9.77021530e-02 -7.05553651e-01 -5.33333004e-01 -1.33862233e+00 8.44508559e-02 -8.87325406e-02 1.08490855e-01 -2.57205158e-01 1.98431149e-01 1.89209193e-01 2.12362304e-01 2.86840528e-01 9.00691897e-02 -3.77321243e-01 -3.23455185e-01 -8.19716275e-01 -8.72257128e-02 4.34563756e-01 1.31070769e+00 7.01623857e-01 -1.08685233e-01 -1.28649652e-01 8.08518350e-01 3.19105208e-01 4.99288201e-01 2.11961344e-01 -1.41651666e+00 2.33591586e-01 2.60522246e-01 2.95953333e-01 -1.00806558e+00 -8.31298381e-02 7.78292343e-02 -7.07463920e-01 3.74562025e-01 2.76310742e-01 -3.37239027e-01 -6.44721329e-01 1.71851003e+00 5.17635465e-01 8.24989259e-01 -1.36873931e-01 1.36284411e+00 6.33573115e-01 6.94157302e-01 -2.36164317e-01 -6.48201585e-01 1.10721564e+00 -1.06592607e+00 -1.08439696e+00 -3.86736095e-02 1.16858922e-01 -9.61555779e-01 1.15310264e+00 5.63606560e-01 -1.35307992e+00 -7.40390599e-01 -1.05665231e+00 -3.60113122e-02 2.64799088e-01 3.41837496e-01 3.59707773e-01 5.28232813e-01 -1.03517377e+00 4.14104819e-01 -7.49660194e-01 -2.65974253e-01 3.01149905e-01 1.45257235e-01 -2.64895827e-01 -2.45377555e-01 -9.38183606e-01 4.17610407e-01 -8.72145817e-02 1.83243051e-01 -1.02435434e+00 -5.01650572e-01 -8.65873933e-01 -4.06595618e-01 5.95910132e-01 -7.12714434e-01 1.28509593e+00 -1.09528887e+00 -1.81236815e+00 8.60648096e-01 -2.78954059e-01 1.48647696e-01 3.13335389e-01 -4.36719954e-01 -4.36795294e-01 4.65440780e-01 -1.11550689e-01 4.04841393e-01 9.95195389e-01 -1.37709546e+00 -6.49295568e-01 -2.52555400e-01 5.79368398e-02 3.73452127e-01 -4.12476301e-01 4.33918715e-01 -1.21722960e+00 -5.91650605e-01 -2.73716561e-02 -9.99456227e-01 -1.65903121e-01 2.04116777e-01 -1.33814797e-01 3.27767402e-01 1.05742836e+00 -7.15549767e-01 1.28147209e+00 -2.19623899e+00 -1.76043045e-02 -5.80126001e-03 2.49827951e-01 3.98363024e-01 -2.57176280e-01 1.86298385e-01 -1.72844246e-01 -3.22351336e-01 -8.98784213e-03 -3.01752925e-01 -3.48365664e-01 -1.65141761e-01 -8.12464952e-02 7.16610074e-01 -1.46816686e-01 6.29914761e-01 -9.88256693e-01 -5.39021850e-01 5.82963824e-01 3.94435078e-01 -4.83013242e-01 6.20916724e-01 2.29556873e-01 6.40104890e-01 -2.16049612e-01 8.98803711e-01 7.75253057e-01 -6.99227825e-02 1.91411555e-01 -6.61038578e-01 -3.10102552e-01 -2.30486199e-01 -1.40904212e+00 1.96045840e+00 -3.59539866e-01 8.74061882e-01 2.80415088e-01 -4.96313959e-01 6.58512294e-01 2.65300661e-01 7.90718377e-01 -8.46887708e-01 3.73052329e-01 -6.99022505e-03 -3.54836792e-01 -1.06878221e+00 4.20964926e-01 3.14876288e-01 3.49832982e-01 3.55271786e-01 -3.75639983e-02 5.95267825e-02 3.52840930e-01 1.28040329e-01 1.08189499e+00 5.18371999e-01 2.68146604e-01 -5.89909628e-02 4.75398302e-01 -3.23186487e-01 8.22923660e-01 3.17820549e-01 -5.93040586e-01 9.53095973e-01 6.56322911e-02 -1.93905830e-01 -1.06647193e+00 -9.26013529e-01 5.48357368e-02 8.08738887e-01 8.31130505e-01 -5.60071707e-01 -8.01828265e-01 -3.94504696e-01 -5.37229240e-01 2.65649050e-01 -2.27316901e-01 -8.18961337e-02 -6.02399468e-01 -4.29574013e-01 1.31008163e-01 3.10459286e-01 7.26254523e-01 -8.00577521e-01 -7.25073218e-01 -2.12340280e-01 -5.36861956e-01 -1.38581264e+00 -9.34461296e-01 -3.92012984e-01 -5.53146303e-01 -1.38063490e+00 -5.84818482e-01 -7.59287059e-01 6.45221293e-01 8.31092298e-01 8.67611587e-01 4.37465273e-02 -4.22190458e-01 7.01553941e-01 -3.38295072e-01 -2.33004931e-02 -1.10545017e-01 -6.15819871e-01 2.89544672e-01 2.75814593e-01 1.84334084e-01 -6.27711773e-01 -1.08491671e+00 7.82904208e-01 -8.11601460e-01 3.20847303e-01 3.97097439e-01 5.54293096e-01 6.86273038e-01 2.03893140e-01 1.15367137e-01 -2.91750938e-01 2.94652790e-01 -2.51876503e-01 -8.12410235e-01 1.25156149e-01 -2.89681584e-01 -5.75057805e-01 4.40930754e-01 -5.19891202e-01 -1.18052936e+00 2.71521896e-01 1.91053897e-01 -9.13767397e-01 -2.37024710e-01 -1.65121838e-01 -4.70413208e-01 -3.42651844e-01 4.46554184e-01 9.67680216e-02 1.26356214e-01 -1.55832753e-01 2.78278410e-01 7.40400255e-01 8.40187371e-01 -3.56845826e-01 8.72081101e-01 7.60690033e-01 -1.97526440e-01 -7.58650005e-01 -6.71723545e-01 -6.30583346e-01 -6.34809494e-01 -8.44514906e-01 7.10130990e-01 -1.09487772e+00 -8.15916836e-01 7.25133538e-01 -1.00608575e+00 -5.91823816e-01 -7.30867162e-02 7.26393223e-01 -7.02600121e-01 5.58544040e-01 -6.71177566e-01 -6.30737543e-01 -3.29696864e-01 -1.17087913e+00 1.10949612e+00 4.42904174e-01 1.73160493e-01 -6.43068135e-01 3.24194372e-01 3.76338869e-01 2.35083282e-01 9.06767175e-02 7.21744001e-02 3.61914814e-01 -9.40913022e-01 5.88723719e-02 -1.97653756e-01 5.82154989e-01 3.10598165e-01 3.26117486e-01 -9.61348891e-01 -4.29241776e-01 1.54585391e-01 -1.49337888e-01 4.62889969e-01 5.72434366e-01 1.24367368e+00 -3.87159102e-02 -1.46999508e-01 1.02784526e+00 1.59604394e+00 3.01540554e-01 9.74337757e-01 3.33381891e-01 7.34049976e-01 4.09181267e-01 1.04017925e+00 5.21011531e-01 2.46588841e-01 9.94373620e-01 4.41493720e-01 -2.50216931e-01 -3.54111373e-01 -1.67296082e-01 5.88606656e-01 7.10709870e-01 -3.51588398e-01 -3.02176982e-01 -6.19914174e-01 3.57145309e-01 -1.91386652e+00 -1.16608727e+00 -1.26718774e-01 2.27347159e+00 7.78474331e-01 -2.00145945e-01 1.24917850e-01 -6.61982894e-02 9.32938755e-01 1.30734041e-01 -4.14562374e-01 2.76053846e-01 -8.84452835e-02 -1.90721795e-01 3.64099413e-01 3.42277706e-01 -1.24139690e+00 6.61592484e-01 6.41958523e+00 6.21579170e-01 -1.17566168e+00 4.90974411e-02 4.35698956e-01 -4.08146054e-01 1.64911419e-01 1.06130913e-02 -5.22984684e-01 7.52172530e-01 6.57985151e-01 -1.27259567e-01 5.59723616e-01 6.46035671e-01 8.23756039e-01 -3.42765719e-01 -9.93480623e-01 1.41934192e+00 3.44717473e-01 -1.17137074e+00 -5.60321450e-01 -7.96396136e-02 7.78893948e-01 -1.31070256e-01 -9.05861706e-03 -1.42093375e-01 -1.30312636e-01 -3.52926463e-01 6.71009779e-01 6.62798822e-01 9.78273690e-01 -4.64467227e-01 2.85126597e-01 7.55291507e-02 -1.33004510e+00 -8.07374716e-02 -1.69221953e-01 -8.29096586e-02 3.82625878e-01 4.79984850e-01 -1.51110277e-01 4.01398361e-01 9.32350218e-01 1.10697734e+00 -4.58952129e-01 1.27995932e+00 -3.92570198e-01 6.00003421e-01 -9.29477513e-02 4.31297511e-01 -3.78867239e-01 -6.16323113e-01 6.89841032e-01 1.12149870e+00 2.69174784e-01 5.52022398e-01 1.21234782e-01 5.23951769e-01 -2.12401226e-02 -1.10250011e-01 -7.23438084e-01 3.88499856e-01 4.00162190e-01 1.62142384e+00 -4.23398405e-01 -1.45782247e-01 -5.33120751e-01 1.20752144e+00 -1.52987570e-01 5.27646542e-01 -1.26288950e+00 -4.89944011e-01 8.19745421e-01 1.17727891e-01 1.52278408e-01 -4.09541637e-01 3.03603988e-02 -1.34878659e+00 3.98015380e-01 -9.01694000e-01 2.15775982e-01 -1.33665156e+00 -9.71817255e-01 3.63722473e-01 -9.77882929e-03 -1.86127436e+00 -2.94261705e-02 -4.10956562e-01 -6.08079612e-01 3.96261007e-01 -1.50209856e+00 -9.14100945e-01 -8.54773462e-01 9.20764148e-01 6.12070262e-01 -8.80461633e-02 2.90696800e-01 7.73632884e-01 -9.32059109e-01 4.63677734e-01 1.55140251e-01 4.45052534e-02 1.23080432e+00 -6.63495123e-01 -9.12567601e-03 1.40237164e+00 -1.12351224e-01 5.58519244e-01 5.99658012e-01 -4.07714665e-01 -1.71764672e+00 -1.01865113e+00 2.64080763e-01 -4.26674455e-01 4.77079690e-01 -4.33021098e-01 -6.41682744e-01 5.76841593e-01 5.27061760e-01 3.02581698e-01 5.87912738e-01 -3.50608617e-01 -6.36029581e-04 -3.77923429e-01 -1.00605547e+00 7.36312687e-01 1.43639636e+00 -3.62279505e-01 -3.19937378e-01 3.60942453e-01 6.12527370e-01 -7.01500654e-01 -6.90885544e-01 4.63503987e-01 7.23956287e-01 -1.22677314e+00 9.47379351e-01 -1.54012099e-01 6.73724711e-01 -7.26115227e-01 -1.23398341e-01 -1.09348452e+00 -3.20740610e-01 -1.10067630e+00 -2.95595109e-01 1.30664980e+00 -2.31316417e-01 -3.18192959e-01 6.36504591e-01 5.20330012e-01 -1.57110661e-01 -7.11924314e-01 -4.48339760e-01 -8.56934488e-01 -8.76098275e-01 -5.39081872e-01 2.31371537e-01 8.86287570e-01 -4.37384099e-02 2.16292471e-01 -7.57113397e-01 4.36910838e-01 8.52203429e-01 2.28090867e-01 1.08160865e+00 -4.61184353e-01 -2.12249801e-01 3.02613731e-02 -4.28624421e-01 -1.23212957e+00 -2.05316525e-02 -2.20851108e-01 2.24651128e-01 -1.17441297e+00 4.05982763e-01 -1.63258299e-01 -1.85155347e-01 2.92574465e-01 -2.45272741e-01 6.20984733e-01 1.75857261e-01 3.68883044e-01 -9.99523103e-01 6.07369244e-01 1.15925884e+00 2.74597496e-01 -3.01378191e-01 -1.30465835e-01 -5.52477896e-01 8.83691430e-01 7.29061484e-01 -1.53827012e-01 -4.66368675e-01 -6.41097665e-01 -9.56684723e-02 2.01802012e-02 4.86964613e-01 -1.12906647e+00 3.65528166e-01 -3.57492328e-01 3.77182245e-01 -3.98299932e-01 4.91457313e-01 -1.14085066e+00 2.25826740e-01 1.80045187e-01 -8.77705365e-02 4.60165069e-02 9.27537456e-02 4.62983459e-01 -1.64290026e-01 9.62483808e-02 1.06414211e+00 1.95917815e-01 -1.01354563e+00 5.88531971e-01 -1.37377400e-02 -8.60084500e-03 1.45935190e+00 -2.73173779e-01 -4.16300178e-01 -5.41475058e-01 -3.65311742e-01 2.49082148e-01 7.95775354e-01 3.92660975e-01 8.44114244e-01 -1.49616122e+00 -4.86486226e-01 5.16158640e-01 1.19161092e-01 -3.70213062e-01 5.17789841e-01 1.01192474e+00 -6.21655703e-01 -1.54308140e-01 -3.65163207e-01 -8.00782561e-01 -1.70237577e+00 5.34889936e-01 3.57269794e-01 3.98137689e-01 -6.13399267e-01 5.43737948e-01 1.83200315e-01 4.97177243e-02 1.93300769e-01 -4.52167168e-03 -4.31497619e-02 -3.33049804e-01 9.31392193e-01 5.47842205e-01 -1.60866126e-01 -5.88235199e-01 -1.97590575e-01 8.35165739e-01 3.09986323e-01 -6.92599118e-02 1.27132034e+00 -7.48500049e-01 -1.41353244e-02 1.76849902e-01 1.19274402e+00 1.77574322e-01 -1.67855358e+00 -2.69214690e-01 -6.60748839e-01 -1.20341206e+00 1.04524128e-01 -4.18868124e-01 -1.32478166e+00 4.81540918e-01 9.27785337e-01 -1.85921952e-01 1.74925816e+00 -2.88497865e-01 7.97492862e-01 1.37137681e-01 4.27368253e-01 -1.12885594e+00 3.22095633e-01 3.10948193e-01 7.28808105e-01 -1.34445918e+00 6.53573275e-02 -5.38825870e-01 -6.02923214e-01 1.03707635e+00 8.17158282e-01 -2.33275443e-02 5.51695406e-01 5.25877357e-01 3.58566701e-01 -5.26263267e-02 -6.92357004e-01 -1.84220552e-01 2.49742821e-04 6.50001705e-01 2.22736180e-01 -4.15922761e-01 -6.71456009e-02 3.27303946e-01 2.48053312e-01 3.86217386e-01 6.61547720e-01 1.02757406e+00 -2.57827044e-01 -9.04440701e-01 -4.46796447e-01 1.54166454e-02 -2.13080570e-01 6.23360313e-02 -6.08765595e-02 5.69651246e-01 1.49129689e-01 1.18570459e+00 -1.17417775e-01 -6.22701526e-01 5.12622952e-01 -4.30836737e-01 6.36677325e-01 -2.98402041e-01 -1.63838759e-01 2.95281112e-01 -6.80962130e-02 -1.20625734e+00 -9.00541067e-01 -7.41224885e-01 -9.36822534e-01 -1.94037244e-01 -1.59169227e-01 -2.95419991e-01 5.13186097e-01 4.81059641e-01 5.49302459e-01 4.16026026e-01 1.23163903e+00 -1.27107680e+00 -7.84421340e-02 -6.23407543e-01 -5.32772660e-01 6.40999019e-01 3.58996212e-01 -6.00093126e-01 -4.13527936e-01 6.39846146e-01]
[10.691097259521484, -1.6686804294586182]
87babc91-9ad1-424d-9a53-bd8acd6fe388
bi-box-regression-for-pedestrian-detection
null
null
http://openaccess.thecvf.com/content_ECCV_2018/html/CHUNLUAN_ZHOU_Bi-box_Regression_for_ECCV_2018_paper.html
http://openaccess.thecvf.com/content_ECCV_2018/papers/CHUNLUAN_ZHOU_Bi-box_Regression_for_ECCV_2018_paper.pdf
Bi-box Regression for Pedestrian Detection and Occlusion Estimation
Occlusions present a great challenge for pedestrian detection in practical applications. In this paper, we propose a novel approach to simultaneous pedestrian detection and occlusion estimation by regressing two bounding boxes to localize the full body as well as the visible part of a pedestrian respectively. For this purpose, we learn a deep convolutional neural network (CNN) consisting of two branches, one for full body estimation and the other for visible part estimation. The two branches are treated differently during training such that they are learned to produce complementary outputs which can be further fused to improve detection performance. The full body estimation branch is trained to regress full body regions for positive pedestrian proposals, while the visible part estimation branch is trained to regress visible part regions for both positive and negative pedestrian proposals. The visible part region of a negative pedestrian proposal is forced to shrink to its center. In addition, we introduce a new criterion for selecting positive training examples, which contributes largely to heavily occluded pedestrian detection. We validate the effectiveness of the proposed bi-box regression approach on the Caltech and CityPersons datasets. Experimental results show that our approach achieves promising performance for detecting both non-occluded and occluded pedestrians, especially heavily occluded ones.
['Chunluan Zhou', 'Junsong Yuan']
2018-09-01
null
null
null
eccv-2018-9
['occlusion-estimation']
['computer-vision']
[-1.11871436e-01 4.12659496e-02 -1.03961572e-01 -2.89406508e-01 -4.27569836e-01 -9.03159529e-02 3.64893347e-01 1.37866050e-01 -6.15375876e-01 8.75416696e-01 -1.30777761e-01 -1.49442121e-01 1.00617254e+00 -7.89068699e-01 -8.04082930e-01 -9.30633903e-01 4.26443294e-02 2.06507564e-01 9.42773819e-01 1.56427193e-02 -3.32160652e-01 4.81012166e-01 -1.45397651e+00 3.44698429e-01 8.23172331e-01 1.00155389e+00 6.70769960e-02 4.70708013e-01 7.70706609e-02 6.51665568e-01 -6.86430335e-01 -6.48437977e-01 4.34729993e-01 6.65123016e-02 -7.00378567e-02 5.10109782e-01 6.30195379e-01 -5.85322380e-01 -2.55056530e-01 9.03643429e-01 6.31935477e-01 3.78144043e-03 5.91872692e-01 -1.29223454e+00 -6.74271733e-02 2.33688086e-01 -1.25914669e+00 3.99361819e-01 9.22389794e-03 4.65604752e-01 4.66260910e-01 -9.02964473e-01 3.64144951e-01 1.47009492e+00 5.97482443e-01 4.31809247e-01 -9.68666792e-01 -8.02030146e-01 5.57860196e-01 1.08956262e-01 -1.39366496e+00 -5.20663500e-01 7.93650329e-01 -6.00277424e-01 5.22868216e-01 1.30687594e-01 9.15064216e-01 7.38283396e-01 1.24891840e-01 1.13500226e+00 8.79599512e-01 -2.72514701e-01 4.73042428e-02 5.43097258e-01 3.41138184e-01 7.69191384e-01 6.54404938e-01 2.55804837e-01 1.14570104e-01 7.77051449e-02 6.78812623e-01 4.18781415e-02 -1.73140094e-01 -6.60173118e-01 -6.32137060e-01 7.83275247e-01 8.06819737e-01 -1.11657381e-01 -3.09888899e-01 1.59063898e-02 4.19701546e-01 -3.30007613e-01 3.96863133e-01 -6.13517702e-01 -3.32078516e-01 7.55238712e-01 -8.66587222e-01 2.96156496e-01 3.75602424e-01 9.06833231e-01 7.22970426e-01 2.67925225e-02 -6.91269994e-01 8.92246604e-01 6.23748720e-01 3.91268700e-01 -6.10185089e-03 -2.27781981e-01 7.67151058e-01 8.13767970e-01 5.26466131e-01 -8.44769061e-01 -4.01236504e-01 -9.33347285e-01 -8.54906142e-01 5.06099045e-01 8.34195793e-01 -3.55399162e-01 -9.51388359e-01 1.41297460e+00 9.24504340e-01 -1.95608601e-01 -1.85838789e-01 1.27906787e+00 1.22077608e+00 6.76627815e-01 6.57776952e-01 6.29743980e-03 1.81890559e+00 -1.38135672e+00 -3.58909756e-01 -6.96546555e-01 4.04140621e-01 -8.13888371e-01 3.94315213e-01 1.10878974e-01 -1.16587365e+00 -1.03327179e+00 -1.02666986e+00 -7.82371312e-02 -1.54946804e-01 1.01207924e+00 3.50936413e-01 7.84738004e-01 -5.87943852e-01 -4.89213206e-02 -6.42576694e-01 -2.43702114e-01 8.00570309e-01 2.72917390e-01 -3.32639456e-01 -9.61211622e-02 -8.67961526e-01 7.50208855e-01 4.86644804e-01 4.33103621e-01 -8.70379329e-01 -1.28626004e-01 -1.04019415e+00 1.36881322e-01 1.08694680e-01 -7.69028306e-01 7.73293018e-01 -8.59201550e-01 -9.68109012e-01 9.79230046e-01 -3.07802737e-01 -6.94665134e-01 9.71396565e-01 -1.85812384e-01 -2.84069985e-01 -1.47124082e-01 3.68308276e-01 1.02456450e+00 1.03837347e+00 -1.41657341e+00 -1.24345160e+00 -4.76772398e-01 -2.08749697e-01 1.71471030e-01 -2.62634158e-01 1.82331324e-01 -7.64460742e-01 -5.66128552e-01 2.56506652e-01 -6.66871428e-01 -3.13802183e-01 4.33013529e-01 -6.80690050e-01 -3.28053653e-01 1.01020944e+00 -8.92073989e-01 9.35332358e-01 -1.88680995e+00 -1.71083897e-01 -6.97523030e-03 2.69470274e-01 5.50898433e-01 -1.11633968e-02 -1.77172631e-01 -4.24139611e-02 -3.69853735e-01 1.17083356e-01 -6.10913754e-01 -3.54893774e-01 -1.96726117e-02 -7.66957551e-02 7.33931124e-01 6.54098809e-01 6.62871063e-01 -6.30943835e-01 -8.74730170e-01 4.88857687e-01 7.87430942e-01 -1.99839711e-01 2.48799846e-01 2.25815564e-01 5.36697567e-01 -3.88348043e-01 9.21069801e-01 1.27375257e+00 9.55899432e-02 -1.37548923e-01 -3.67278010e-01 -3.91536951e-01 -1.21931061e-01 -1.46362901e+00 5.80212533e-01 -1.54683396e-01 5.75756192e-01 2.35400364e-01 -8.93578470e-01 1.14044321e+00 1.40407830e-01 2.92267855e-02 -4.16961849e-01 3.43875349e-01 -1.06502645e-01 6.34464398e-02 -4.29852635e-01 4.01001304e-01 7.83908889e-02 2.77477354e-01 -1.93808109e-01 -2.05567062e-01 4.13158000e-01 5.54564834e-01 -6.72750622e-02 4.13648784e-01 2.54365176e-01 3.49931777e-01 -1.80782247e-02 1.23049712e+00 -2.30558336e-01 8.36878657e-01 5.68583190e-01 -8.48448038e-01 4.61705565e-01 3.92176956e-01 -9.00001347e-01 -9.98074472e-01 -1.02730227e+00 -2.21472681e-01 1.17545342e+00 4.07796085e-01 5.72144501e-02 -6.97140574e-01 -7.81911433e-01 4.24440652e-02 3.07339042e-01 -4.77129221e-01 1.59719855e-01 -1.00384665e+00 -7.94862568e-01 4.63444665e-02 7.99960256e-01 8.24789822e-01 -9.33519721e-01 -6.85351670e-01 1.94892034e-01 -2.73841053e-01 -1.17038214e+00 -3.96568000e-01 9.52192992e-02 -6.47145867e-01 -1.06170273e+00 -1.18448627e+00 -1.02945566e+00 8.17259610e-01 6.76357031e-01 8.83533061e-01 1.91054657e-01 -4.46442932e-01 -2.02846155e-01 1.31668895e-01 -3.90358537e-01 -2.79643267e-01 -2.61469454e-01 -1.17588758e-01 3.83839786e-01 2.24422798e-01 -3.48812677e-02 -1.05587351e+00 6.82134748e-01 -1.49698332e-01 1.16941449e-03 7.92126894e-01 7.64884233e-01 4.74055201e-01 1.16753422e-01 2.43554965e-01 -3.49762291e-01 -7.56752267e-02 -2.24911779e-01 -8.57162416e-01 1.78964406e-01 4.03605074e-01 -4.00469810e-01 4.72139835e-01 -5.58691680e-01 -1.14746904e+00 6.45663500e-01 -2.17055798e-01 -2.61476785e-01 -5.20996511e-01 -4.44663584e-01 -4.25621390e-01 -1.46854445e-01 5.29823184e-01 6.29415289e-02 -4.11223322e-01 -4.21971768e-01 2.75843471e-01 5.69662571e-01 6.01353765e-01 -4.58248109e-01 9.19586182e-01 6.38318419e-01 -7.98290074e-02 -8.58966768e-01 -6.00289285e-01 -8.00814152e-01 -8.22563827e-01 -5.89457214e-01 9.30296421e-01 -1.25488937e+00 -6.27924263e-01 3.04730058e-01 -1.35581744e+00 1.02308519e-01 -2.04140440e-01 3.39158684e-01 8.85716900e-02 4.69559342e-01 -5.70558310e-01 -1.34432304e+00 -4.68915522e-01 -1.34511864e+00 1.46740460e+00 6.29223168e-01 1.33610755e-01 -6.08948648e-01 -3.67036253e-01 4.54114705e-01 1.71169043e-02 2.93992758e-01 2.82210559e-01 -3.55473787e-01 -6.75891519e-01 -6.69521928e-01 -7.14138269e-01 1.89824522e-01 -3.47808212e-01 4.18397523e-02 -1.07748353e+00 -5.24332762e-01 -2.75373310e-01 -1.70387581e-01 1.27547860e+00 7.40730584e-01 5.60038567e-01 2.19649412e-02 -8.39638114e-01 3.96406561e-01 1.07466674e+00 -8.86849687e-02 5.39658904e-01 3.10358554e-01 5.64163685e-01 7.29787827e-01 6.91738009e-01 3.24099123e-01 2.46133894e-01 7.85570025e-01 4.94830757e-01 -5.98279476e-01 -5.94185412e-01 -6.68480843e-02 3.67312163e-01 3.07318382e-02 -1.42724097e-01 -1.29973337e-01 -6.05385721e-01 4.88802016e-01 -1.74893641e+00 -8.03568065e-01 -6.34814143e-01 2.36995244e+00 3.62894684e-01 4.46046263e-01 5.70239186e-01 1.43338606e-01 1.22795522e+00 9.08807945e-03 -3.64378393e-01 1.29898429e-01 -2.53092766e-01 -1.09854862e-01 4.08084750e-01 2.35377923e-01 -1.88616896e+00 8.65436018e-01 5.11647987e+00 7.62807488e-01 -8.16335917e-01 1.81264624e-01 9.69618797e-01 2.44534910e-01 6.19018674e-01 -1.11219823e-01 -1.49373865e+00 4.61223751e-01 1.21831065e-02 6.22332335e-01 -5.43162405e-01 1.14247119e+00 2.99186796e-01 -3.77014607e-01 -6.85176313e-01 8.62413466e-01 -2.07719892e-01 -7.44815767e-01 -1.50095373e-01 -1.37382135e-01 6.28307402e-01 -4.04270917e-01 5.34090288e-02 4.08585250e-01 6.51070848e-02 -5.16546011e-01 9.52372372e-01 1.70876086e-01 2.21397266e-01 -6.17667258e-01 8.64466131e-01 4.95416760e-01 -1.88384247e+00 -3.49493861e-01 -7.51650512e-01 -1.71552211e-01 2.83665150e-01 5.33438206e-01 -6.79537237e-01 2.68900275e-01 6.40394390e-01 5.72345376e-01 -6.72838688e-01 1.54045057e+00 -5.51107287e-01 2.90022314e-01 -2.66494632e-01 1.45149916e-01 1.01526722e-01 -1.03806630e-01 6.03407741e-01 1.37607956e+00 -3.07094119e-02 -1.09456740e-01 6.87477171e-01 9.69333589e-01 1.89512938e-01 2.92869329e-01 -2.46260405e-01 8.53146851e-01 1.41093403e-01 1.59097040e+00 -9.37829077e-01 -6.77099884e-01 -5.11392772e-01 8.54603887e-01 5.16452789e-01 5.02372205e-01 -1.11002302e+00 -5.69003448e-02 4.01954412e-01 3.49361062e-01 6.26042187e-01 -4.25814465e-02 -2.67811596e-01 -1.16328096e+00 1.56697124e-01 -2.76083052e-01 4.90968168e-01 -4.83075857e-01 -1.15427864e+00 5.14278173e-01 -2.88302451e-01 -1.29103029e+00 2.43604735e-01 -6.41273558e-01 -9.27170098e-01 1.18341160e+00 -1.66224360e+00 -1.47421873e+00 -6.80064142e-01 4.31394309e-01 6.42593622e-01 4.43109944e-02 7.97469392e-02 5.66589892e-01 -1.09444940e+00 5.49834669e-01 -4.24050450e-01 4.33791548e-01 5.59335589e-01 -9.13829327e-01 4.15514320e-01 1.22149301e+00 -4.23876673e-01 3.50329816e-01 6.57719374e-01 -9.17881727e-01 -6.12210035e-01 -1.49428129e+00 7.49010086e-01 -1.83274314e-01 1.72856841e-02 -3.18925977e-01 -6.92595005e-01 4.72982556e-01 -1.79980353e-01 4.65261012e-01 2.92651802e-01 -2.57786900e-01 -2.41212279e-01 -2.24385396e-01 -1.11788785e+00 5.25187135e-01 5.34278929e-01 2.80572355e-01 -3.40653360e-01 5.14559984e-01 2.90966272e-01 -2.97282428e-01 -4.12071526e-01 4.65218574e-01 7.82434762e-01 -1.14305067e+00 1.48821878e+00 -3.08639616e-01 2.26173192e-01 -6.93399668e-01 1.35509416e-01 -6.43170297e-01 -3.84101689e-01 1.03646092e-01 -3.28183353e-01 1.12803090e+00 -1.90105978e-02 -4.50913757e-01 1.03186560e+00 3.68706554e-01 -2.99224816e-02 -7.83423185e-01 -9.18429434e-01 -6.32420659e-01 -7.73651572e-03 4.68955226e-02 1.76945120e-01 3.19478184e-01 -4.69686300e-01 3.71033221e-01 -5.23831904e-01 4.26356167e-01 9.64340925e-01 3.18322890e-02 1.24017537e+00 -1.17766976e+00 -9.43218023e-02 -3.97329658e-01 -5.93646228e-01 -1.50834167e+00 -2.59514779e-01 -3.57214034e-01 1.38639495e-01 -1.46438384e+00 4.54878032e-01 -3.56140316e-01 -5.41076809e-02 2.08519608e-01 -6.58191383e-01 7.09872127e-01 2.44665608e-01 1.47549555e-01 -4.50034559e-01 4.38665867e-01 1.14570999e+00 -3.97513807e-01 -3.67924362e-01 6.44279659e-01 -2.50325143e-01 9.59158480e-01 5.76333821e-01 -1.34316608e-01 6.51313066e-02 -4.65634651e-02 -6.80785894e-01 -5.34552149e-02 8.51368845e-01 -1.30087268e+00 1.88209444e-01 3.27602565e-01 1.29883087e+00 -1.19100320e+00 5.51288962e-01 -6.90839767e-01 -3.18027854e-01 9.96581018e-01 4.36030030e-02 -4.63770688e-01 2.92572677e-01 5.12922347e-01 7.20604658e-02 -1.01807550e-01 1.26097369e+00 -9.56901312e-02 -6.00333273e-01 3.31766993e-01 -2.75433511e-01 -3.40268105e-01 1.32667112e+00 -3.84546220e-01 -2.46400356e-01 -1.51469231e-01 -8.88466239e-01 4.62180257e-01 5.49329259e-02 3.41230810e-01 7.27956533e-01 -1.13870358e+00 -9.39891398e-01 4.29490238e-01 5.78642711e-02 -1.71096668e-01 3.29813242e-01 9.40656960e-01 -3.57510209e-01 4.31642681e-01 -1.91086352e-01 -9.55195010e-01 -1.64907610e+00 7.80492544e-01 5.00082970e-01 -1.99457213e-01 -6.40066803e-01 9.55815315e-01 7.92641342e-01 -1.06042124e-01 5.96316755e-01 -3.17570686e-01 -5.74729025e-01 -3.65100764e-02 7.17895508e-01 4.46478695e-01 -3.54336470e-01 -1.21758103e+00 -3.32520396e-01 6.16557360e-01 -1.49407178e-01 4.21982199e-01 8.18686426e-01 -2.02597842e-01 2.92676166e-02 -1.47110209e-01 8.83080840e-01 -1.27044946e-01 -1.44097090e+00 -2.47193635e-01 -4.90588427e-01 -6.00235820e-01 -1.79355785e-01 -4.73155290e-01 -1.21037507e+00 1.13349104e+00 9.87780869e-01 -1.60685539e-01 8.30381930e-01 -8.79315287e-02 7.46848881e-01 -8.62577558e-02 1.39221057e-01 -7.81859338e-01 9.54723731e-03 2.88018528e-02 6.12763345e-01 -1.54531717e+00 2.54412502e-01 -8.86872649e-01 -4.13839102e-01 1.25311351e+00 1.08540964e+00 -7.33601078e-02 3.23378861e-01 6.81363344e-02 -2.16520071e-01 2.19890162e-01 -1.85061604e-01 -6.25269651e-01 5.74588835e-01 8.30209076e-01 3.97461712e-01 1.16030842e-01 -3.30795944e-01 5.37406027e-01 2.82133102e-01 -3.43035340e-01 1.19123437e-01 6.22837901e-01 -9.49503839e-01 -7.86589921e-01 -1.02055895e+00 1.49248376e-01 -1.22325055e-01 1.50406599e-01 -1.73663273e-01 9.38879490e-01 6.74369097e-01 9.61774886e-01 1.01070374e-01 9.70089436e-02 2.12683991e-01 -1.41598701e-01 2.94213206e-01 -3.24670792e-01 -4.97947872e-01 4.87277061e-01 7.73683190e-02 -1.68747723e-01 -2.17906386e-01 -6.51504695e-01 -8.42356324e-01 -1.96876619e-02 -7.16666639e-01 -9.83546972e-02 4.57666427e-01 6.80686355e-01 -3.14133883e-01 6.04736328e-01 3.56212735e-01 -1.23337710e+00 -3.11485648e-01 -9.38379109e-01 -2.17328519e-01 2.83693284e-01 3.19500506e-01 -9.45761383e-01 2.34916303e-02 9.33208019e-02]
[8.030909538269043, -0.5871753692626953]
15e2e7d5-35a6-44c0-93b7-3f9efc43187d
color-face-recognition-using-high-dimension
1712.01642
null
http://arxiv.org/abs/1712.01642v1
http://arxiv.org/pdf/1712.01642v1.pdf
Color Face Recognition using High-Dimension Quaternion-based Adaptive Representation
Recently, quaternion collaborative representation-based classification (QCRC) and quaternion sparse representation-based classification (QSRC) have been proposed for color face recognition. They can obtain correlation information among different color channels. However, their performance is unstable in different conditions. For example, QSRC performs better than than QCRC on some situations but worse on other situations. To benefit from quaternion-based $e_2$-norm minimization in QCRC and quaternion-based $e_1$-norm minimization in QSRC, we propose the quaternion-based adaptive representation (QAR) that uses a quaternion-based $e_p$-norm minimization ($1 \le p \le 2$) for color face recognition. To obtain the high dimension correlation information among different color channels, we further propose the high-dimension quaternion-based adaptive representation (HD-QAR). The experimental results demonstrate that the proposed QAR and HD-QAR achieve better recognition rates than QCRC, QSRC and several state-of-the-art methods.
['Qingxiang Feng', 'Yicong Zhou']
2017-11-19
null
null
null
null
['sparse-representation-based-classification']
['computer-vision']
[-1.30799010e-01 -8.21260691e-01 7.66132772e-02 -3.20995688e-01 -5.68564117e-01 1.31006524e-01 1.59484863e-01 -4.45167035e-01 -4.31571782e-01 6.45579159e-01 4.23772121e-03 4.71011549e-02 -3.26491892e-01 -7.94509888e-01 -2.38715187e-01 -7.93842137e-01 -4.33206648e-01 -1.09641336e-01 -3.00382197e-01 -6.12271070e-01 4.87939477e-01 7.68074691e-01 -1.56176686e+00 -3.15458551e-02 7.56608367e-01 1.08081770e+00 -3.09814095e-01 4.38220471e-01 7.21158832e-02 5.73273599e-01 -6.54954016e-01 -4.96668309e-01 2.70357490e-01 -6.92608297e-01 -1.88109383e-01 -9.43335444e-02 2.48825222e-01 6.43160641e-02 -4.67885494e-01 1.31069148e+00 5.57989717e-01 2.48570785e-01 8.35161328e-01 -1.36453736e+00 -1.10897744e+00 -1.63470992e-04 -9.84821439e-01 -2.80539636e-02 5.23921490e-01 -3.82039249e-01 6.91623807e-01 -1.32308257e+00 5.95015109e-01 1.55566871e+00 4.55311000e-01 3.08709890e-01 -8.77303481e-01 -1.11596346e+00 -6.94038197e-02 7.14974940e-01 -1.97604632e+00 -8.91912654e-02 9.16008174e-01 -7.46081397e-02 5.77988148e-01 4.61126357e-01 6.22558177e-01 3.64196509e-01 1.34199440e-01 5.92909455e-01 1.44474757e+00 -3.00051391e-01 7.76793957e-02 -1.77482799e-01 -2.84206301e-01 1.07849145e+00 3.72172415e-01 6.72094300e-02 -7.02288926e-01 1.44773489e-02 1.09484506e+00 1.26714006e-01 -4.55304652e-01 -1.13439105e-01 -1.30034912e+00 8.73457193e-01 5.71267426e-01 2.73515403e-01 -3.65215868e-01 -2.95356996e-02 5.09874038e-02 2.10100621e-01 1.27567545e-01 2.00546488e-01 6.65502921e-02 -9.86818373e-02 -8.14008594e-01 -2.99734324e-01 4.90624875e-01 9.05107439e-01 1.06019890e+00 6.65845454e-01 7.54676089e-02 1.23143637e+00 5.60479224e-01 1.17866099e+00 6.26821637e-01 -7.58122504e-01 2.87906408e-01 5.37711263e-01 -3.54279131e-02 -1.64324868e+00 -4.34262246e-01 -3.40901643e-01 -1.40619528e+00 1.54610887e-01 -1.21934079e-02 -9.39894319e-02 -6.54840827e-01 1.29779744e+00 3.51650976e-02 3.50247562e-01 2.46174544e-01 1.26576257e+00 8.44616771e-01 8.97695363e-01 1.99065227e-02 -5.87766945e-01 1.16985977e+00 -4.19260025e-01 -8.20856988e-01 2.69509673e-01 2.55628228e-01 -1.07829511e+00 5.46714544e-01 5.91201723e-01 -6.94360912e-01 -7.08863139e-01 -1.18140173e+00 4.66820419e-01 -2.35598579e-01 6.44904912e-01 6.93466961e-01 1.08169460e+00 -7.76401997e-01 3.39665532e-01 -3.76879096e-01 5.23262471e-02 1.71120107e-01 3.60332102e-01 -8.40369225e-01 -5.52490652e-01 -1.12924659e+00 8.00344884e-01 -1.61857158e-01 5.32356560e-01 -2.13737190e-01 -2.12851927e-01 -8.89357209e-01 -3.11244935e-01 -1.12065515e-02 3.09576660e-01 3.57208312e-01 -8.37520480e-01 -1.80071139e+00 3.55719954e-01 -2.67033458e-01 3.76203567e-01 -2.00487897e-02 4.70343791e-02 -9.41460192e-01 3.93026352e-01 -7.94128478e-02 3.10319990e-01 1.03717899e+00 -1.19696748e+00 -2.07081303e-01 -5.49149334e-01 -4.39639270e-01 1.85737893e-01 -2.40842640e-01 4.99991626e-02 -5.32474697e-01 -8.82408082e-01 9.14394200e-01 -9.37539637e-01 4.69564982e-02 4.68221121e-02 -1.30858704e-01 -1.50607213e-01 7.11391211e-01 -6.50733352e-01 1.06883800e+00 -2.09658098e+00 3.70239407e-01 5.14876127e-01 -3.00655603e-01 5.94832778e-01 -4.44623858e-01 5.30963838e-01 -4.04047906e-01 3.55615770e-03 -8.13758299e-02 1.11403234e-01 -2.06840619e-01 6.00623563e-02 -4.43499461e-02 9.43632841e-01 4.68818694e-01 3.71374220e-01 -9.15430665e-01 -5.01226902e-01 2.27498204e-01 8.01455796e-01 -4.08630073e-01 1.66846544e-01 6.13669693e-01 1.08888417e-01 -3.77751470e-01 1.01208472e+00 1.31133866e+00 7.04743937e-02 2.43231341e-01 -1.00305402e+00 -2.53619879e-01 -5.17919004e-01 -1.79295850e+00 1.16169596e+00 -4.17185694e-01 5.15553415e-01 2.28571519e-02 -1.00472248e+00 1.33834350e+00 1.32907882e-01 6.72664285e-01 -1.04713643e+00 -4.27587777e-02 2.10984826e-01 -6.10936880e-02 -9.80175883e-02 5.85615039e-01 -2.74648339e-01 1.59453154e-01 4.33261633e-01 9.36321728e-03 2.22469606e-02 1.93630576e-01 7.65823200e-02 5.59360027e-01 9.32056084e-02 2.76970267e-01 -1.18486091e-01 1.10406590e+00 -3.90535802e-01 1.03315079e+00 7.17857527e-03 -2.52886087e-01 8.12942922e-01 2.79097050e-01 -9.65370610e-02 -6.70637012e-01 -6.88805223e-01 -1.93056777e-01 6.35739863e-01 4.96557891e-01 -4.72080618e-01 -1.26252636e-01 -3.89974058e-01 3.25275101e-02 2.76610255e-01 -4.88027811e-01 -1.72260955e-01 -6.29081070e-01 -1.20543504e+00 3.12087387e-01 3.88224334e-01 1.06525958e+00 -4.98574615e-01 -1.01949155e-01 2.11224347e-01 4.00331952e-02 -6.41315579e-01 -3.28431219e-01 -3.98049951e-01 -5.03569543e-01 -1.20689428e+00 -1.15369725e+00 -5.16725659e-01 9.73263979e-01 6.31434917e-01 5.99873602e-01 9.42134410e-02 -4.34179246e-01 6.65717959e-01 -8.83990645e-01 -1.77126732e-02 1.69850171e-01 -7.74991989e-01 2.74242729e-01 4.08836365e-01 2.70828873e-01 -2.79112235e-02 -9.32339489e-01 6.30954981e-01 -8.28917682e-01 -3.12317252e-01 7.61169791e-01 1.02481449e+00 5.73526084e-01 5.59714474e-02 6.93624794e-01 -4.72602129e-01 5.58692753e-01 -3.38725358e-01 -5.24323046e-01 3.23691279e-01 -5.16767204e-01 -5.15436642e-02 8.31336856e-01 -3.61122221e-01 -7.24976957e-01 -1.12979457e-01 -2.83206314e-01 -5.04829526e-01 5.01344562e-01 6.01474822e-01 8.02167971e-03 -6.02585852e-01 3.95133555e-01 6.11598253e-01 3.02759055e-02 -3.78138274e-01 4.77519274e-01 7.84010231e-01 4.92854565e-02 -2.90831834e-01 1.08411598e+00 4.55348462e-01 3.76204699e-01 -1.17994499e+00 -9.04949829e-02 -3.39125961e-01 -3.27986568e-01 -2.65218049e-01 6.22127712e-01 -1.13976371e+00 -1.16123891e+00 5.74983895e-01 -8.15747142e-01 5.35667896e-01 1.65959373e-01 1.01232803e+00 -9.90684852e-02 8.60572517e-01 -4.24962491e-01 -1.10745621e+00 -3.16356599e-01 -1.14976895e+00 8.07948053e-01 6.29632294e-01 5.01171947e-01 -4.04294193e-01 -1.80280328e-01 2.26613395e-02 4.39554453e-01 -7.07680881e-02 5.63563168e-01 6.24483973e-02 -4.73499686e-01 -2.39269212e-01 -5.52602649e-01 4.76488560e-01 2.91372985e-01 2.13066548e-01 -3.39673966e-01 -6.14600539e-01 -3.72173846e-01 -2.81821042e-01 5.10829329e-01 -1.93815976e-01 9.31286037e-01 -2.75529921e-01 1.57652706e-01 6.55903220e-01 1.60635722e+00 5.70365965e-01 9.78982925e-01 1.84863746e-01 7.09365547e-01 2.02376813e-01 8.41713130e-01 6.46223962e-01 4.22557712e-01 6.69667661e-01 6.44496307e-02 -1.40697032e-01 -9.16700531e-03 -9.69497766e-03 4.31074381e-01 1.31631029e+00 -5.07396221e-01 2.59141684e-01 -4.27797407e-01 -1.69055313e-02 -1.45221639e+00 -8.95386994e-01 8.79582390e-02 2.27252746e+00 6.30193651e-01 -5.21900654e-01 -3.63548845e-02 4.19855624e-01 6.35216236e-01 2.08149001e-01 -4.91335005e-01 -4.15046662e-01 -3.03225428e-01 6.18987978e-01 4.68301445e-01 6.24793805e-02 -8.94616365e-01 7.66447961e-01 5.41390562e+00 9.33748066e-01 -1.53078711e+00 -1.75511569e-01 3.29170704e-01 5.94665408e-01 -1.04730219e-01 -2.72012144e-01 -5.68040609e-01 3.67189854e-01 4.57785666e-01 4.87412550e-02 4.96038854e-01 8.24793875e-01 -2.48068988e-01 -1.80022672e-01 -5.28266609e-01 2.01598811e+00 6.03499532e-01 -1.16378510e+00 3.51173162e-01 -2.56339461e-01 7.25264668e-01 -4.73293453e-01 2.24244401e-01 3.28354120e-01 -2.38864850e-02 -1.11611378e+00 3.08465004e-01 6.59389436e-01 1.22140121e+00 -9.37431097e-01 6.75030768e-01 -2.27626055e-01 -1.55318666e+00 6.75239339e-02 -9.75788653e-01 4.23006117e-01 -4.92132485e-01 3.85765046e-01 -2.89401025e-01 9.95571911e-01 4.08160686e-01 9.92062807e-01 -6.91669166e-01 1.02876484e+00 -2.38459423e-01 3.81266475e-01 -5.28134294e-02 -3.35647017e-01 -4.37339321e-02 -9.93826807e-01 4.26340640e-01 1.14919496e+00 6.06736839e-01 6.04147017e-01 -2.19722137e-01 4.24247295e-01 -1.77087020e-02 6.63366437e-01 -6.98758438e-02 -1.71381950e-01 4.81187224e-01 1.32799995e+00 -5.89223504e-01 -2.34131604e-01 -3.93233895e-01 1.10573292e+00 -9.15190503e-02 5.14445305e-01 -4.57181752e-01 -8.71881843e-01 8.68712544e-01 -4.89584297e-01 4.08913374e-01 -7.14441001e-01 4.24317986e-01 -1.58092856e+00 -3.51763904e-01 -1.08352041e+00 2.28989989e-01 -7.93585360e-01 -1.19493413e+00 7.73187339e-01 -2.97169417e-01 -1.86570466e+00 4.31752987e-02 -9.18005049e-01 -2.21606225e-01 1.09986842e+00 -1.86874139e+00 -8.73370111e-01 -2.61862904e-01 8.66884112e-01 1.02867290e-01 -5.39517522e-01 9.79925931e-01 4.17458385e-01 -9.36174810e-01 8.87677789e-01 5.17941058e-01 3.55713516e-01 6.95834339e-01 -9.72410738e-01 -3.76390338e-01 7.24380851e-01 2.22303405e-01 8.81756306e-01 2.89077133e-01 -4.46558326e-01 -2.50391221e+00 -9.28231776e-01 2.77556986e-01 3.72233063e-01 6.15703948e-02 2.62748808e-01 -7.23918498e-01 1.12780161e-01 -2.39288643e-01 9.87308323e-02 1.05752647e+00 -1.07473247e-01 -7.84248710e-01 -8.35557461e-01 -1.22712064e+00 6.36570811e-01 5.86292386e-01 -7.29087472e-01 -1.59714505e-01 2.97480911e-01 -1.96270853e-01 3.61257568e-02 -1.33980203e+00 4.23716038e-01 8.11537087e-01 -9.87025201e-01 1.20395744e+00 -1.28805339e-01 -5.52713610e-02 -9.10617590e-01 -5.64404428e-01 -1.44171572e+00 -5.19232690e-01 -2.13561386e-01 6.10822141e-01 8.99319589e-01 1.45219741e-02 -8.93672884e-01 6.75058246e-01 9.82840508e-02 4.21614021e-01 -5.53863585e-01 -1.16992307e+00 -6.57688320e-01 -6.10454828e-02 -1.32206008e-01 4.86294329e-01 8.59611750e-01 9.83293355e-02 5.10712042e-02 -5.55389345e-01 2.02311695e-01 9.40324962e-01 3.97089362e-01 5.27893841e-01 -1.01708567e+00 1.79872483e-01 -3.26464176e-02 -9.90921378e-01 -8.91474247e-01 3.20350751e-02 -5.22156775e-01 -2.43650064e-01 -1.33586836e+00 -2.12463830e-02 -6.20759547e-01 -4.14579898e-01 2.06117764e-01 -2.43250415e-01 9.52161193e-01 5.49680114e-01 3.93425286e-01 -5.96190155e-01 9.28187966e-01 1.47498989e+00 -2.56767988e-01 6.68322742e-02 -3.76551300e-01 -3.49843353e-01 3.17744076e-01 4.05286163e-01 7.62029812e-02 -1.30417347e-01 -1.18491426e-01 3.94836277e-01 2.19754010e-01 -1.04653887e-01 -1.16276336e+00 1.22452319e-01 -1.55467987e-01 1.04654491e+00 -5.53517044e-01 8.63974869e-01 -6.62495494e-01 2.30774805e-01 5.68399906e-01 2.63685316e-01 1.09165743e-01 -6.69252798e-02 4.65165019e-01 -6.21814370e-01 1.46765724e-01 9.34644938e-01 -4.21786234e-02 -8.48949850e-01 3.82483363e-01 -4.75508422e-01 -3.62918466e-01 8.34606469e-01 -4.65604335e-01 -3.23372900e-01 -5.94078898e-01 -2.97956437e-01 -2.72396207e-01 1.38731495e-01 1.90807670e-01 1.23405039e+00 -1.66281366e+00 -8.03041339e-01 7.27198660e-01 3.01725082e-02 -6.82079196e-01 3.76396149e-01 5.40363729e-01 -8.24550569e-01 3.14752996e-01 -7.10839987e-01 -3.61344963e-01 -1.25401986e+00 4.86894883e-02 1.24430783e-01 2.51016319e-01 2.72032060e-02 9.33590412e-01 -3.31325203e-01 -2.53597528e-01 -1.46961093e-01 2.18253762e-01 -6.28197789e-01 1.79336593e-01 5.60964108e-01 6.37786746e-01 7.43633360e-02 -1.48890376e+00 -7.15007305e-01 1.20306289e+00 1.31565884e-01 -1.74445435e-01 1.22753298e+00 1.17588408e-01 -4.71296191e-01 3.26407105e-02 1.61100745e+00 -4.35584001e-02 -7.07792401e-01 -1.10016979e-01 -4.32760119e-01 -1.03663993e+00 1.16089270e-01 -5.97885609e-01 -1.34198344e+00 7.25632608e-01 1.06996548e+00 -3.05501968e-01 1.38444090e+00 -8.28329921e-01 3.30693930e-01 3.78000468e-01 8.39097083e-01 -1.12176394e+00 3.66459191e-01 3.40205908e-01 1.18420160e+00 -1.23219860e+00 6.07913911e-01 -7.51386702e-01 -7.77219892e-01 1.54254091e+00 5.66323340e-01 -2.04852983e-01 9.63556886e-01 -5.18662214e-01 1.42655581e-01 1.82305112e-01 -2.47579917e-01 -3.18171978e-01 4.80056614e-01 5.75316191e-01 5.75126827e-01 1.26341805e-01 -7.47789204e-01 8.73301774e-02 -8.17316324e-02 -2.11665481e-01 3.99407059e-01 1.01150405e+00 -2.23713383e-01 -1.27149343e+00 -8.68948698e-01 3.34673584e-01 -9.96304825e-02 1.21302925e-01 -5.19346595e-02 6.47164941e-01 1.15015641e-01 1.23716664e+00 -4.10448713e-03 -7.78357983e-01 4.17464942e-01 -2.87275404e-01 6.57528818e-01 -2.38189012e-01 -5.49407639e-02 1.24882616e-01 -2.31648162e-01 -6.58776462e-01 -5.85834622e-01 -4.49474454e-01 -1.11513793e+00 -2.24139601e-01 -4.46133792e-01 5.93568146e-01 1.06546080e+00 3.19598347e-01 3.09781522e-01 7.72758434e-03 1.26732790e+00 -7.08895028e-01 -4.23029065e-01 -9.25413847e-01 -1.22500646e+00 4.89431590e-01 -4.26306352e-02 -9.13259923e-01 -2.89556146e-01 -2.51271576e-01]
[10.850279808044434, -1.6529357433319092]
0279606f-099f-493c-b593-dbf57b78069f
combining-hybrid-attention-networks-and-lstm
null
null
https://aclanthology.org/2020.rocling-1.28
https://aclanthology.org/2020.rocling-1.28.pdf
Combining Hybrid Attention Networks and LSTM for Stock Trend Prediction
null
['Jenq-Haur Wang', 'Hsin-Wen Liu']
null
null
null
null
rocling-2020-9
['stock-trend-prediction']
['time-series']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.201622009277344, 3.7338497638702393]
68089783-56f7-41d0-8cc4-e8f762513f2a
a-real-time-and-unsupervised-face-re
1804.03547
null
https://arxiv.org/abs/1804.03547v3
https://arxiv.org/pdf/1804.03547v3.pdf
A real-time and unsupervised face Re-Identification system for Human-Robot Interaction
In the context of Human-Robot Interaction (HRI), face Re-Identification (face Re-ID) aims to verify if certain detected faces have already been observed by robots. The ability of distinguishing between different users is crucial in social robots as it will enable the robot to tailor the interaction strategy toward the users' individual preferences. So far face recognition research has achieved great success, however little attention has been paid to the realistic applications of Face Re-ID in social robots. In this paper, we present an effective and unsupervised face Re-ID system which simultaneously re-identifies multiple faces for HRI. This Re-ID system employs Deep Convolutional Neural Networks to extract features, and an online clustering algorithm to determine the face's ID. Its performance is evaluated on two datasets: the TERESA video dataset collected by the TERESA robot, and the YouTube Face Dataset (YTF Dataset). We demonstrate that the optimised combination of techniques achieves an overall 93.55% accuracy on TERESA dataset and an overall 90.41% accuracy on YTF dataset. We have implemented the proposed method into a software module in the HCI^2 Framework for it to be further integrated into the TERESA robot, and has achieved real-time performance at 10~26 Frames per second.
['Yujiang Wang', 'Jie Shen', 'Maja Pantic', 'Stavros Petridis']
2018-04-10
null
null
null
null
['online-clustering']
['computer-vision']
[-7.14427307e-02 2.57150590e-01 2.74504930e-01 -6.29579484e-01 -7.82651082e-02 -1.46335810e-01 4.63040501e-01 -4.72907364e-01 -3.89354765e-01 2.59316385e-01 -3.06613952e-01 2.34702840e-01 -1.89194158e-01 -4.47086751e-01 -3.63862187e-01 -4.38083649e-01 -3.79158109e-01 7.34378517e-01 -1.73562810e-01 -6.52741194e-02 1.90254614e-01 9.83795643e-01 -2.35862589e+00 1.35017589e-01 2.12914556e-01 1.02818441e+00 4.65870708e-01 7.12372899e-01 2.60696530e-01 5.69869220e-01 -5.39458990e-01 -2.11996749e-01 3.65669340e-01 -1.90988576e-04 -9.38389778e-01 6.35257512e-02 4.21547323e-01 -4.27276373e-01 -2.77847260e-01 7.81277895e-01 5.83603859e-01 6.36509061e-02 7.35538483e-01 -1.93172777e+00 -4.24782842e-01 2.62798518e-01 -5.02529144e-01 -1.53289482e-01 8.21628094e-01 -8.01996365e-02 3.61586034e-01 -1.03075171e+00 5.67337811e-01 1.76652861e+00 7.00141490e-01 8.23604822e-01 -7.68660784e-01 -1.12223887e+00 -4.78600621e-01 4.07181442e-01 -1.80874825e+00 -9.77967441e-01 4.27597791e-01 -3.72516572e-01 9.84700441e-01 -2.20509186e-01 1.40209094e-01 8.89887631e-01 -1.43374624e-02 6.57238305e-01 8.17596555e-01 -4.42367524e-01 4.81294170e-02 1.95056990e-01 -1.31627634e-01 7.73142457e-01 1.15902396e-02 -1.45701542e-01 -4.81513321e-01 2.08149776e-01 7.21771657e-01 -3.30063305e-03 3.09469327e-02 -9.20180827e-02 -8.87723088e-01 5.89819908e-01 3.33027989e-01 3.12897354e-01 -3.70654255e-01 1.05763219e-01 5.36038458e-01 3.88920844e-01 7.23143071e-02 8.29382464e-02 -2.43069172e-01 -2.46429905e-01 -1.91606224e-01 3.17774527e-02 9.21335101e-01 1.35842395e+00 8.64097595e-01 -3.79058599e-01 8.19770843e-02 1.09958446e+00 5.78752995e-01 5.56943357e-01 4.94009793e-01 -1.23399818e+00 -1.20156296e-01 5.71206331e-01 3.48771453e-01 -1.41267335e+00 -6.02972627e-01 5.09061456e-01 -6.25325441e-01 4.04328555e-01 7.34037757e-02 -2.49873117e-01 -7.95655191e-01 1.51358414e+00 4.33047116e-01 1.21291608e-01 2.52239585e-01 8.16258430e-01 1.13516319e+00 3.96381706e-01 -2.16037612e-02 -4.19741869e-02 1.38222075e+00 -7.73692131e-01 -7.32894063e-01 -1.47578297e-02 4.72243577e-01 -7.61549532e-01 7.45655000e-01 4.03314322e-01 -5.73007166e-01 -9.19602275e-01 -1.03558385e+00 1.49395779e-01 -3.63246202e-01 6.92982495e-01 3.15978616e-01 7.99845934e-01 -1.40500152e+00 3.21074396e-01 -3.39179277e-01 -1.18734634e+00 4.39121991e-01 1.19630063e+00 -9.40797627e-01 -1.83216020e-01 -7.55705297e-01 8.31868529e-01 1.99895009e-01 1.92951471e-01 -8.67191732e-01 9.38387439e-02 -8.34537983e-01 -1.38570920e-01 1.61495566e-01 -2.39640430e-01 1.45750141e+00 -1.17833614e+00 -1.99807334e+00 1.06176329e+00 -2.42296472e-01 -1.71998113e-01 1.88415542e-01 -1.22662671e-01 -5.62255621e-01 2.49408588e-01 2.63401598e-01 1.06435144e+00 1.02057552e+00 -1.38010991e+00 -9.88943875e-01 -6.66357279e-01 5.24046123e-02 2.96495587e-01 -2.74261564e-01 3.62045705e-01 -5.12316346e-01 1.97911095e-02 -2.84729034e-01 -1.27515590e+00 3.63318980e-01 -5.43867797e-03 1.46735571e-02 -7.08002627e-01 1.38195062e+00 -2.52912939e-01 4.65683550e-01 -2.29462433e+00 -1.23937510e-01 2.01147899e-01 -1.15890251e-02 5.12908638e-01 -3.02406460e-01 3.93925518e-01 -1.18527062e-01 -2.87161052e-01 2.35403612e-01 -6.31789804e-01 -1.17753081e-01 2.88812965e-01 3.63877267e-01 5.79691887e-01 1.59014255e-01 2.99636602e-01 -7.05603540e-01 -3.95049453e-01 3.83910000e-01 5.53092837e-01 -2.73335755e-01 6.04033887e-01 4.11598563e-01 3.58358830e-01 -7.50674680e-03 6.92183614e-01 8.11952829e-01 2.41366118e-01 2.92895079e-01 -2.81654328e-01 -1.07425690e-01 -3.71712148e-01 -1.14703202e+00 1.09818995e+00 -4.61855263e-01 6.95001543e-01 5.74300289e-01 -8.41249526e-01 1.24211454e+00 3.49992514e-01 5.85351288e-01 -5.60468316e-01 5.01864195e-01 7.98257813e-02 -1.23759501e-01 -9.12955105e-01 5.18849194e-01 2.90229887e-01 1.03192553e-02 3.84429663e-01 3.08560550e-01 2.82590747e-01 -2.78614163e-02 -1.55015782e-01 1.00916600e+00 9.23388824e-02 1.52866080e-01 -1.87232792e-01 6.56175017e-01 -1.78168103e-01 2.11400971e-01 4.61772084e-01 -5.68888485e-01 4.18432862e-01 -5.77860139e-02 -2.65537411e-01 -6.85233414e-01 -4.82643545e-01 1.44023970e-02 1.27472591e+00 3.44804108e-01 -2.58120690e-02 -8.78666699e-01 -7.21253693e-01 5.88258803e-02 3.07806730e-01 -6.93915427e-01 -1.07763171e-01 -3.06495696e-01 -1.85767069e-01 5.54166734e-01 2.82756507e-01 8.45414340e-01 -1.52920663e+00 -7.23215699e-01 -1.04842789e-01 2.49728002e-02 -1.22146559e+00 -2.57298321e-01 1.68771558e-02 -8.89973342e-02 -1.27171862e+00 -5.94345748e-01 -1.35167217e+00 7.44295299e-01 8.90065193e-01 5.79911709e-01 1.16095528e-01 -3.29629242e-01 9.99036670e-01 -7.01272726e-01 -4.52942580e-01 -3.51596981e-01 -4.08581942e-02 6.76574707e-01 2.05143660e-01 7.51488090e-01 -3.82067919e-01 -4.65397149e-01 9.21382725e-01 -3.78440171e-01 -4.51249599e-01 4.28052694e-01 4.97172356e-01 -1.23724952e-01 1.57198176e-01 8.03649664e-01 -5.70266187e-01 5.59415102e-01 -6.26131356e-01 -2.88213253e-01 1.51715130e-01 -3.21837008e-01 -4.11707342e-01 3.33467990e-01 -3.73343766e-01 -1.02037406e+00 6.11474633e-01 -1.50508255e-01 -5.64559758e-01 -5.50807774e-01 1.97863892e-01 -3.32386345e-01 -3.08151811e-01 4.20394063e-01 -1.64966315e-01 5.51412344e-01 -1.45094648e-01 1.11307703e-01 1.71876979e+00 6.23159707e-01 -1.90871701e-01 2.42366835e-01 4.11529183e-01 -3.65704000e-01 -1.26172626e+00 -6.09272309e-02 -8.18289876e-01 -9.22371089e-01 -7.36494005e-01 7.84598768e-01 -9.80000496e-01 -1.48804617e+00 7.98849344e-01 -1.13708282e+00 -2.80567974e-01 3.49921048e-01 2.98930049e-01 -6.94264770e-01 2.14576557e-01 -2.50138074e-01 -1.17452288e+00 -4.41911429e-01 -1.28932679e+00 1.15556705e+00 5.92735767e-01 -4.26722556e-01 -2.86736071e-01 -2.68384635e-01 3.04377615e-01 2.17193380e-01 -1.91609293e-01 2.35793635e-01 -6.43158555e-01 1.13024682e-01 -2.81641632e-01 -4.73278612e-01 8.51613432e-02 5.82440734e-01 1.89945176e-01 -1.31460249e+00 -3.97395045e-01 -1.31958947e-01 -4.07723904e-01 1.73051536e-01 2.84166075e-03 7.45784521e-01 -1.78002790e-01 -5.59989631e-01 1.35442659e-01 9.18500006e-01 8.49030375e-01 7.81177878e-01 2.59763092e-01 6.80320084e-01 1.01154828e+00 8.88533056e-01 5.94956279e-01 5.92561007e-01 7.88701952e-01 5.21361470e-01 2.43720919e-01 9.57578570e-02 1.57328740e-01 5.45110404e-01 3.48166645e-01 -1.03663661e-01 -1.34760186e-01 -8.49045336e-01 5.33811986e-01 -1.94523787e+00 -9.54938769e-01 2.78061815e-02 1.98758411e+00 1.24043219e-01 -4.71219093e-01 4.60525453e-01 1.57856792e-01 1.13124359e+00 -5.92118084e-01 -4.62370783e-01 -5.76246321e-01 3.70312393e-01 -1.91904023e-01 3.84968013e-01 2.53062516e-01 -1.27809811e+00 9.91815329e-01 5.92706203e+00 5.09120166e-01 -1.08210099e+00 -1.16280280e-01 4.90749776e-01 4.24668342e-01 7.47750223e-01 -6.51278555e-01 -8.14774692e-01 2.40367532e-01 1.05055416e+00 7.27600828e-02 7.05086768e-01 1.21152270e+00 1.31053567e-01 -3.46206546e-01 -1.18236947e+00 1.47986579e+00 4.19401854e-01 -6.14905119e-01 -3.61908257e-01 3.27709205e-02 6.88666999e-02 -3.56478132e-02 -5.22174919e-03 4.77909595e-01 2.65363693e-01 -1.17549062e+00 4.98942882e-01 3.64592582e-01 1.07574642e+00 -1.14224696e+00 9.89852369e-01 1.21400796e-01 -1.30899286e+00 -4.34067816e-01 -4.60352689e-01 2.25774944e-02 -2.89163142e-01 -2.27973074e-01 -1.58029747e+00 2.08607435e-01 1.32438850e+00 7.58207619e-01 -4.29921001e-01 6.64965510e-01 2.53181189e-01 -1.03535257e-01 -3.95632893e-01 -1.10366628e-01 -1.95696443e-01 -4.55544703e-02 1.79814413e-01 1.15551567e+00 4.30105865e-01 3.11849624e-01 -6.03410788e-02 1.88501611e-01 -1.04111291e-01 -7.48210400e-02 -8.61852586e-01 1.80535577e-02 6.31974518e-01 1.54749155e+00 -7.69415617e-01 6.26551509e-02 -3.69522452e-01 1.41561294e+00 3.68350923e-01 -4.77224328e-02 -5.56026936e-01 -5.64578533e-01 8.20543289e-01 -4.69714589e-02 3.62035990e-01 -2.31585726e-01 2.61370063e-01 -3.49653244e-01 -3.51503760e-01 -8.98240507e-01 2.10318998e-01 -9.97174621e-01 -1.37588835e+00 9.33290541e-01 -4.08992954e-02 -1.16119945e+00 -4.09276664e-01 -9.93235946e-01 -4.03192073e-01 4.30383503e-01 -9.89840090e-01 -1.30782247e+00 -7.44127393e-01 7.64679253e-01 5.70208251e-01 -5.84252894e-01 9.72795606e-01 5.20813823e-01 -7.95937955e-01 8.47294390e-01 4.31326367e-02 1.23309471e-01 9.12804842e-01 -8.05235028e-01 1.11593492e-01 1.66300878e-01 -3.27865750e-01 5.87819278e-01 5.78346372e-01 -6.43856883e-01 -1.81391931e+00 -1.14048529e+00 5.97818494e-01 -3.70040327e-01 9.86395031e-02 -3.96719366e-01 -5.44442236e-01 7.45205879e-01 2.28329763e-01 -7.34954700e-02 5.79635561e-01 -9.21192169e-02 1.22044347e-01 -2.60669291e-01 -1.75387895e+00 5.01213014e-01 1.10720146e+00 -4.68254775e-01 -1.78847581e-01 7.88695887e-02 2.36839518e-01 -2.94978730e-02 -8.59236240e-01 3.63756686e-01 8.85770500e-01 -1.01814950e+00 7.39471078e-01 1.39107972e-01 1.51556013e-02 -4.11234885e-01 -2.02502042e-01 -9.86755848e-01 -3.46616656e-01 -5.56508422e-01 3.67126524e-01 1.59353793e+00 -1.10778451e-01 -6.38914704e-01 7.50912964e-01 7.33480930e-01 8.01846385e-02 -3.12450588e-01 -9.34154034e-01 -5.01262605e-01 -7.26591527e-01 -2.15140834e-01 6.58959627e-01 7.92550564e-01 2.72883415e-01 3.61664474e-01 -2.96923369e-01 4.02685761e-01 2.49912307e-01 -3.48263830e-01 1.29034245e+00 -1.33630931e+00 4.06074822e-01 -1.66472614e-01 -7.32173920e-01 -6.70713603e-01 4.51894850e-01 -5.49068511e-01 4.78915453e-01 -1.24482787e+00 9.76577997e-02 -5.36796510e-01 2.44028866e-01 5.09997189e-01 3.25511575e-01 4.59003121e-01 1.92805365e-01 3.66449893e-01 -6.99425459e-01 6.55705750e-01 5.06270289e-01 -6.43260479e-02 -2.03930870e-01 -2.54356637e-05 -3.77712309e-01 7.59956777e-01 8.94109964e-01 -2.49330252e-01 -2.06198156e-01 -1.62748352e-01 -3.90087992e-01 -1.39332697e-01 1.72675744e-01 -1.39463365e+00 3.74374181e-01 1.37953654e-01 4.45222050e-01 -4.47428852e-01 4.15291190e-01 -1.22208512e+00 3.47851813e-01 2.03697473e-01 1.33999005e-01 1.74637094e-01 2.68208683e-01 3.98510724e-01 1.05037890e-01 -4.33805823e-01 7.63517439e-01 4.36948501e-02 -1.11650205e+00 1.55761302e-01 -7.52387702e-01 -7.92333186e-01 1.45503712e+00 -5.88191092e-01 2.63330084e-03 -4.11525279e-01 -4.44771945e-01 2.74490356e-01 6.07898891e-01 7.70947039e-01 8.73639405e-01 -1.25068772e+00 -3.81194025e-01 4.41594303e-01 2.86950618e-01 -1.54056415e-01 9.08037350e-02 4.16744560e-01 -6.09862030e-01 1.79203853e-01 -6.13646328e-01 -7.14107454e-01 -1.80123472e+00 5.01082301e-01 2.45561644e-01 6.22965932e-01 -3.29129934e-01 7.58580446e-01 -6.29645064e-02 -6.71545386e-01 5.28417945e-01 2.41075993e-01 -7.71745324e-01 1.96483210e-01 6.67482853e-01 6.26596689e-01 -4.02646475e-02 -1.48548806e+00 -6.36414886e-01 5.83981812e-01 -7.24016801e-02 6.04139790e-02 1.25982177e+00 -6.06137395e-01 -1.35253996e-01 2.03326121e-01 1.33494878e+00 -4.42374349e-01 -1.05468500e+00 1.36114687e-01 1.01366438e-01 -4.02135819e-01 -3.02493483e-01 -5.29772103e-01 -9.60347712e-01 5.60473263e-01 1.22316945e+00 6.18281737e-02 1.01150703e+00 6.42815679e-02 4.84247118e-01 6.81321144e-01 9.13169622e-01 -1.16983104e+00 2.50390679e-01 6.86141968e-01 1.16491675e+00 -1.49144542e+00 -2.79910743e-01 -4.63887960e-01 -8.15669894e-01 1.24061489e+00 9.51449871e-01 -1.31356671e-01 6.81259692e-01 1.45704031e-01 1.03482030e-01 -2.54562706e-01 -1.96003258e-01 -2.47491166e-01 -1.60766199e-01 1.09360790e+00 1.81775734e-01 -6.04638073e-04 3.02544773e-01 4.68883425e-01 -1.96339056e-01 2.30675399e-01 4.14406031e-01 9.95574057e-01 -4.82568651e-01 -7.64186800e-01 -5.70402563e-01 2.11166516e-01 -2.27869600e-01 6.15876198e-01 -5.64576685e-01 7.46359527e-01 3.44877899e-01 1.48323095e+00 2.87327886e-01 -8.78731251e-01 3.30389470e-01 -1.58299565e-01 3.64534795e-01 -6.45645261e-01 -4.56040800e-01 -2.87693590e-01 2.07940623e-01 -5.74675500e-01 -7.91364968e-01 -6.59468949e-01 -1.35778022e+00 -3.51047605e-01 -1.47794500e-01 3.61403264e-02 8.09024632e-01 7.90319264e-01 7.42137611e-01 1.19826663e-02 9.50280786e-01 -1.76704776e+00 1.67397127e-01 -1.19303524e+00 -4.97292101e-01 2.88149595e-01 9.47479233e-02 -1.06331050e+00 -1.49424911e-01 7.36939237e-02]
[13.360966682434082, 0.620349645614624]
5f6549ff-7791-4d6a-bde1-8ffd455bdaeb
hybridgazenet-geometric-model-guided
2111.11691
null
https://arxiv.org/abs/2111.11691v1
https://arxiv.org/pdf/2111.11691v1.pdf
HybridGazeNet: Geometric model guided Convolutional Neural Networks for gaze estimation
As a critical cue for understanding human intention, human gaze provides a key signal for Human-Computer Interaction(HCI) applications. Appearance-based gaze estimation, which directly regresses the gaze vector from eye images, has made great progress recently based on Convolutional Neural Networks(ConvNets) architecture and open-source large-scale gaze datasets. However, encoding model-based knowledge into CNN model to further improve the gaze estimation performance remains a topic that needs to be explored. In this paper, we propose HybridGazeNet(HGN), a unified framework that encodes the geometric eyeball model into the appearance-based CNN architecture explicitly. Composed of a multi-branch network and an uncertainty module, HybridGazeNet is trained using a hyridized strategy. Experiments on multiple challenging gaze datasets shows that HybridGazeNet has better accuracy and generalization ability compared with existing SOTA methods. The code will be released later.
['Xin Wang', 'Rui Wu', 'Zhizhong Su', 'Xiao Jiang', 'Shaobo Guo']
2021-11-23
null
null
null
null
['gaze-estimation']
['computer-vision']
[-6.08892962e-02 6.17030412e-02 -1.64589971e-01 -5.37850738e-01 -5.56740314e-02 4.07965407e-02 2.10771337e-01 -7.05311954e-01 -2.04168275e-01 4.26943004e-01 -3.00015688e-01 -2.74815112e-01 1.10961042e-01 -1.66728824e-01 -7.45728135e-01 -8.37894320e-01 3.90928656e-01 -1.70291230e-01 9.69193801e-02 -2.04931840e-01 5.29905140e-01 4.91059422e-02 -1.99048448e+00 -1.20668866e-01 1.08983493e+00 1.71866107e+00 2.57344022e-02 5.58631480e-01 1.35647714e-01 7.72610724e-01 -2.11350098e-01 -4.11181420e-01 3.49176936e-02 -4.03651237e-01 -4.77982342e-01 -3.23965102e-01 8.42924893e-01 -3.43100965e-01 1.46055698e-01 1.17399371e+00 2.63308525e-01 -9.39475223e-02 5.48838437e-01 -1.72965467e+00 -9.45239842e-01 7.73315430e-02 -9.01611805e-01 2.01649696e-01 2.15837911e-01 3.47873032e-01 9.07849967e-01 -6.82134390e-01 2.29483739e-01 1.24733484e+00 7.52716064e-01 7.62595713e-01 -9.45610285e-01 -1.26189864e+00 2.04746306e-01 5.80436766e-01 -1.40121150e+00 -3.85486573e-01 9.13251102e-01 -4.59507972e-01 8.56610775e-01 2.70038694e-01 6.28812909e-01 1.18486810e+00 4.54944968e-01 1.06700206e+00 1.18794370e+00 -4.89740610e-01 -2.97407091e-01 5.86671988e-03 4.07697618e-01 9.26083863e-01 2.92432636e-01 3.76859367e-01 -7.70909309e-01 4.10042107e-01 5.77123523e-01 1.48446262e-01 -5.03999472e-01 -4.41870779e-01 -6.89107180e-01 6.64992154e-01 1.14408004e+00 -2.31227115e-01 -3.35734010e-01 3.58911812e-01 -1.15787804e-01 -7.92767182e-02 7.19182014e-01 2.46283129e-01 -3.21242839e-01 -8.44206959e-02 -1.05086374e+00 3.03302944e-01 4.91292685e-01 1.02244067e+00 9.46703851e-01 -1.36750946e-02 -2.24899888e-01 4.87827927e-01 8.79911423e-01 9.10718024e-01 3.06178510e-01 -6.16913497e-01 9.52460393e-02 9.89054203e-01 -1.66783646e-01 -1.08999074e+00 -6.15049124e-01 -1.61571994e-01 -7.57722735e-01 5.64351141e-01 3.93100709e-01 -1.73349798e-01 -1.02868617e+00 1.77619064e+00 2.47498497e-01 3.97760004e-01 -5.16024113e-01 1.23349297e+00 9.94300842e-01 3.39083016e-01 1.08868606e-01 -1.38957771e-02 1.46961010e+00 -1.16640460e+00 -9.47432339e-01 -1.97095603e-01 2.91960061e-01 -5.61741590e-01 1.03493607e+00 5.66330194e-01 -1.03460884e+00 -6.40644133e-01 -1.16950428e+00 -5.19375205e-01 -4.35497731e-01 4.25924361e-01 5.29789865e-01 8.88321757e-01 -1.25587869e+00 3.74522619e-02 -7.99740791e-01 -3.48148018e-01 7.75379360e-01 7.48084307e-01 -1.77363798e-01 1.78981021e-01 -1.04584146e+00 7.68232286e-01 1.85574353e-01 6.37940526e-01 -4.25200403e-01 -4.18538421e-01 -1.16589212e+00 1.73603386e-01 4.77133423e-01 -6.39472365e-01 1.52045667e+00 -1.38279283e+00 -1.80987060e+00 6.98863924e-01 -6.86748266e-01 -3.32444191e-01 1.15232222e-01 -4.09425825e-01 -3.95749748e-01 -2.89026529e-01 -3.36530298e-01 1.03674400e+00 1.36364090e+00 -1.09322762e+00 -7.81004369e-01 -6.40528858e-01 6.42831028e-02 -4.23350967e-02 -1.35779604e-01 2.54315794e-01 -6.82068229e-01 -2.48479202e-01 -7.62672871e-02 -1.26862597e+00 4.95462894e-01 1.71040922e-01 -5.95051944e-01 -6.91664517e-01 8.77550542e-01 -4.57653821e-01 1.63893270e+00 -1.82910979e+00 1.59500018e-01 1.14594214e-01 8.91640961e-01 2.87056416e-01 1.57458574e-01 -3.12066495e-01 -1.26470789e-01 -1.08964145e-02 2.03517508e-02 -5.78827918e-01 1.29867136e-01 -3.37993175e-01 9.23305843e-03 3.73619944e-01 4.68524814e-01 1.42065203e+00 -5.73659360e-01 -4.93631184e-01 8.71112198e-02 4.81661022e-01 -5.28532982e-01 3.20377588e-01 -2.22978607e-01 2.74004340e-01 -2.59133130e-01 7.89615154e-01 8.51014495e-01 -6.70448661e-01 -3.09009910e-01 -3.71514112e-01 -1.74507633e-01 -9.86704454e-02 -4.86542404e-01 1.64521742e+00 -2.52020657e-01 1.24233270e+00 -1.79104820e-01 -3.84775579e-01 7.28946865e-01 -4.38982956e-02 4.10355963e-02 -8.55368674e-01 8.45419288e-01 -1.72878206e-01 2.51699030e-01 -7.63665617e-01 5.53977251e-01 2.67342091e-01 2.55860895e-01 5.20256877e-01 2.54944861e-01 4.37113822e-01 -2.01170877e-01 -2.23093182e-01 3.98877800e-01 6.21546984e-01 1.90924570e-01 -2.43037134e-01 6.35348082e-01 -2.12910801e-01 3.81768107e-01 2.41871208e-01 -4.67355877e-01 7.01121986e-01 5.80963969e-01 -5.29091299e-01 -6.98409617e-01 -4.99782354e-01 -9.83507484e-02 1.28640187e+00 4.72391278e-01 -3.69445980e-01 -1.28716910e+00 -7.55851507e-01 -1.05032340e-01 5.86344659e-01 -1.30792451e+00 -2.79646456e-01 -4.10654306e-01 -4.88136202e-01 2.31553376e-01 5.57012796e-01 6.33019865e-01 -1.28967273e+00 -6.88789845e-01 -4.46204513e-01 -5.14364950e-02 -7.38822043e-01 -6.64983690e-01 -1.57202423e-01 -4.10480708e-01 -1.35029972e+00 -7.15050340e-01 -5.98429441e-01 6.17867529e-01 4.13925409e-01 9.85749841e-01 3.44537139e-01 -3.74388732e-02 -4.27078903e-02 -1.67595446e-01 -1.04638052e+00 3.20619434e-01 4.25861716e-01 -7.52829984e-02 2.44892538e-01 1.32745111e+00 -1.10561987e-02 -7.67678678e-01 3.78261924e-01 -4.74490941e-01 2.15770930e-01 4.93532926e-01 8.44089448e-01 1.17979683e-01 -6.39629781e-01 1.52069926e-01 -8.04244995e-01 7.47582376e-01 -4.32653517e-01 -9.42909956e-01 3.52876931e-01 -9.96764779e-01 1.08237870e-01 -2.91972514e-02 -2.58989304e-01 -1.24234402e+00 -2.02191487e-01 1.17148217e-02 -7.43576527e-01 -2.27950677e-01 3.58362585e-01 -9.98580307e-02 -3.11327875e-01 6.48912191e-01 -1.36231676e-01 3.42184633e-01 -2.58873969e-01 1.74382225e-01 1.00797749e+00 3.53256196e-01 -1.37319177e-01 4.90526050e-01 2.19456106e-01 -7.70874396e-02 -5.43664157e-01 -1.13104272e+00 -3.37400913e-01 -6.78611994e-01 -4.22243506e-01 1.18713367e+00 -8.42613935e-01 -1.59731400e+00 9.03244376e-01 -1.16135287e+00 -3.00190717e-01 5.29886961e-01 4.81029212e-01 -3.84934038e-01 -1.53480232e-01 -1.59834102e-01 -8.40578854e-01 -5.72025597e-01 -1.47756922e+00 1.20046246e+00 8.19165885e-01 -1.58455744e-01 -8.95051360e-01 1.15075624e-02 2.74375409e-01 4.32055652e-01 -6.74594566e-02 3.49574834e-01 -3.54096860e-01 -9.65927780e-01 -1.65234789e-01 -8.16946387e-01 2.46626332e-01 6.44102599e-03 2.91947395e-01 -1.50615549e+00 -2.27206647e-01 -9.65472385e-02 -4.01058167e-01 9.38339710e-01 8.57345581e-01 1.30831623e+00 2.47493871e-02 -5.30638754e-01 1.14502132e+00 1.19326711e+00 -3.19582969e-02 7.63166785e-01 4.18381959e-01 1.10327768e+00 5.04882753e-01 4.28769112e-01 1.27321079e-01 7.83479333e-01 5.07321358e-01 8.07022452e-01 -7.29784891e-02 -8.91922042e-02 -6.50990801e-03 6.05247356e-02 4.93276775e-01 -3.85141015e-01 -4.56135333e-01 -9.78338420e-01 7.56937489e-02 -1.85809124e+00 -8.88937533e-01 -3.04129511e-01 1.73763919e+00 5.19177794e-01 1.09296635e-01 2.31011793e-01 -1.88297927e-01 6.56569123e-01 -9.02014598e-02 -8.93649697e-01 -3.17405701e-01 1.91510513e-01 1.30545005e-01 3.65353972e-01 1.33544877e-01 -1.10466838e+00 1.03467953e+00 6.07491064e+00 4.01491016e-01 -1.73892641e+00 7.77410567e-02 4.31867868e-01 -2.40983404e-02 3.69045436e-02 -3.14039677e-01 -1.09280467e+00 5.42493463e-01 8.42732847e-01 3.01785707e-01 4.62024808e-01 8.01191270e-01 4.22274321e-02 -1.75215766e-01 -9.02990758e-01 1.35966098e+00 6.34057462e-01 -1.07349229e+00 -4.10851032e-01 1.61005825e-01 5.58308601e-01 2.81405061e-01 4.96032834e-01 2.69568861e-01 -9.07500386e-02 -1.27005017e+00 7.22410738e-01 1.01260495e+00 1.18674731e+00 -5.26280463e-01 7.77191997e-01 1.23024568e-01 -9.29695189e-01 -2.12375730e-01 -8.68799761e-02 -6.33641034e-02 -5.79487979e-02 -2.11652637e-01 -5.91121078e-01 2.80727684e-01 1.12565279e+00 1.01449847e+00 -9.32761431e-01 1.23490226e+00 -3.44247520e-01 6.69160903e-01 -1.95209295e-01 -2.93209046e-01 1.13968223e-01 -6.33338913e-02 1.83613315e-01 6.55719340e-01 2.09021121e-01 -7.90391490e-02 -4.97386903e-01 1.36098576e+00 -1.64552122e-01 -2.58119702e-01 -3.04059446e-01 2.66407549e-01 1.89069763e-01 1.40495551e+00 -2.27165923e-01 6.03144243e-02 -5.36647618e-01 7.31452227e-01 5.19197524e-01 5.76826096e-01 -9.49184120e-01 -6.02191567e-01 9.10562932e-01 1.16663888e-01 3.23899329e-01 1.12992212e-01 -4.95742023e-01 -1.20131922e+00 -7.49788210e-02 -8.33651364e-01 -9.22161415e-02 -1.42579043e+00 -1.02444601e+00 9.10455346e-01 -1.15021162e-01 -1.26030922e+00 -4.26643699e-01 -9.78751957e-01 -6.29950762e-01 1.28620589e+00 -1.85288918e+00 -1.62029636e+00 -1.02759707e+00 6.34610236e-01 3.80069971e-01 -1.71191081e-01 5.19362807e-01 -1.17459707e-01 -1.18733335e+00 1.02119851e+00 -3.53023410e-01 5.19377142e-02 8.44489336e-01 -1.20859361e+00 4.43505824e-01 7.46569753e-01 -2.49981537e-01 8.44617426e-01 5.97364962e-01 -3.52171302e-01 -1.08442545e+00 -8.36938322e-01 7.69075215e-01 -7.86736250e-01 4.31279510e-01 -2.67275065e-01 -1.00096142e+00 8.56086135e-01 7.34320223e-01 -6.60803467e-02 5.85675657e-01 4.12987709e-01 -2.79668510e-01 -2.74797738e-01 -8.16496491e-01 5.27350545e-01 8.22966397e-01 -6.10825419e-01 -4.19413209e-01 -4.79050547e-01 5.65965712e-01 -7.34887898e-01 -3.94630104e-01 3.54529828e-01 8.79966319e-01 -1.17525876e+00 4.86906558e-01 -4.16638315e-01 2.67843097e-01 -3.21592510e-01 5.07897913e-01 -1.08022451e+00 -2.52268553e-01 -6.73172712e-01 -4.79310036e-01 7.60404944e-01 2.58475780e-01 -6.19542658e-01 8.43524754e-01 8.10458422e-01 -1.05834929e-02 -8.03546071e-01 -2.84829795e-01 -1.74007863e-01 -2.29867324e-01 -5.38582742e-01 9.93341923e-01 5.86876929e-01 -7.69801736e-02 5.91296554e-01 -5.07010400e-01 4.36692387e-02 6.72818005e-01 -1.82831466e-01 1.04874492e+00 -1.75832129e+00 5.59539735e-01 -7.08634257e-01 -4.21693712e-01 -1.22721398e+00 4.34861779e-01 -3.37988585e-01 2.74062902e-01 -1.03840339e+00 1.10228680e-01 -1.55067787e-01 -5.32537818e-01 5.14964759e-01 -4.66661394e-01 5.53128421e-01 -1.22069346e-03 2.33914986e-01 -7.59558916e-01 4.76994842e-01 1.29205859e+00 3.89322144e-04 -1.65183678e-01 2.31032744e-01 -7.87368119e-01 7.20192194e-01 7.76552498e-01 -2.62003839e-01 -4.26607698e-01 -4.13996309e-01 5.51031053e-01 -3.31287861e-01 4.91022497e-01 -8.92407298e-01 5.51797926e-01 -3.75941135e-02 4.61620033e-01 -7.87267685e-01 3.48882705e-01 -8.25282335e-01 -3.44746977e-01 -1.15210526e-02 -3.10965627e-01 1.86520174e-01 3.28496784e-01 5.78058660e-01 -1.99023366e-01 3.93511392e-02 5.67606747e-01 4.81590629e-01 -7.98274040e-01 5.87849677e-01 1.11611769e-01 -3.16656888e-01 9.28203702e-01 -5.68233848e-01 -3.62403095e-01 -2.43994296e-01 -3.61491859e-01 3.67829025e-01 5.28704941e-01 8.02226543e-01 5.86491644e-01 -1.20399213e+00 -2.21283898e-01 7.33064413e-01 4.88564730e-01 3.74993086e-02 1.62581727e-01 1.16347158e+00 -3.98482263e-01 6.64449751e-01 -4.18585896e-01 -9.92085099e-01 -1.27688205e+00 6.69733524e-01 5.05645871e-01 3.67262244e-01 1.94136370e-02 1.17746508e+00 5.59813559e-01 -1.70221269e-01 2.83975959e-01 -7.29116440e-01 -7.28777051e-01 -1.06093705e-01 6.85097277e-01 2.65100509e-01 2.10241433e-02 -9.81798470e-01 -2.31213436e-01 7.07941711e-01 -1.74748704e-01 4.87247437e-01 1.02184260e+00 -5.04449308e-01 -2.42528573e-01 4.56876040e-01 1.20052135e+00 -5.40000200e-01 -1.56449401e+00 -2.34442472e-01 -2.94528939e-02 -3.04897875e-01 5.73677897e-01 -7.81364620e-01 -1.18875766e+00 1.31320846e+00 9.18056846e-01 1.04019359e-01 1.38140619e+00 -2.85335094e-01 6.45646811e-01 2.40618169e-01 -4.29925099e-02 -7.93076217e-01 -5.95550314e-02 4.93552655e-01 8.36149812e-01 -1.88722444e+00 -2.46749759e-01 -1.18891418e-01 -6.48371220e-01 1.09977281e+00 1.20347917e+00 -1.04536302e-01 1.11021638e+00 -2.97730356e-01 4.49293524e-01 -4.92240280e-01 -6.27906024e-01 -5.79391301e-01 1.09566462e+00 5.19411445e-01 4.42571402e-01 -2.01669142e-01 2.30323747e-01 6.94165051e-01 -1.46735817e-01 4.66206282e-01 -3.38259786e-02 4.99323517e-01 -3.04515153e-01 -7.51365006e-01 -2.99603522e-01 4.93397534e-01 -3.68733644e-01 -3.47440243e-01 -1.51618779e-01 7.59718835e-01 2.51249224e-01 7.80181408e-01 4.03982759e-01 -6.52575433e-01 3.96562787e-03 6.78703785e-02 4.83743101e-01 -3.90663803e-01 -4.86533135e-01 4.92402762e-02 -4.71615523e-01 -8.22279871e-01 -7.31922090e-01 -3.85066122e-01 -7.77660728e-01 -5.02564311e-01 -9.53052461e-01 -2.91225463e-01 7.79082119e-01 1.11583292e+00 5.03340065e-01 5.23169994e-01 3.39296639e-01 -1.10068440e+00 -1.25966147e-01 -1.37009692e+00 -3.66047025e-01 1.14709258e-01 7.92964816e-01 -1.11843312e+00 -1.04331478e-01 8.67531449e-02]
[14.122788429260254, 0.06974201649427414]
c09ee9f4-a007-49a4-926a-a593e00067c0
communication-computation-efficient-device
2108.13009
null
https://arxiv.org/abs/2108.13009v2
https://arxiv.org/pdf/2108.13009v2.pdf
Communication-Computation Efficient Device-Edge Co-Inference via AutoML
Device-edge co-inference, which partitions a deep neural network between a resource-constrained mobile device and an edge server, recently emerges as a promising paradigm to support intelligent mobile applications. To accelerate the inference process, on-device model sparsification and intermediate feature compression are regarded as two prominent techniques. However, as the on-device model sparsity level and intermediate feature compression ratio have direct impacts on computation workload and communication overhead respectively, and both of them affect the inference accuracy, finding the optimal values of these hyper-parameters brings a major challenge due to the large search space. In this paper, we endeavor to develop an efficient algorithm to determine these hyper-parameters. By selecting a suitable model split point and a pair of encoder/decoder for the intermediate feature vector, this problem is casted as a sequential decision problem, for which, a novel automated machine learning (AutoML) framework is proposed based on deep reinforcement learning (DRL). Experiment results on an image classification task demonstrate the effectiveness of the proposed framework in achieving a better communication-computation trade-off and significant inference speedup against various baseline schemes.
['Jun Zhang', 'Yuyi Mao', 'Jiawei Shao', 'Xinjie Zhang']
2021-08-30
null
null
null
null
['feature-compression']
['computer-vision']
[ 3.19814861e-01 -3.23991925e-01 -6.62902296e-01 -3.22591454e-01 -6.78118765e-01 -7.25018159e-02 2.88066685e-01 -1.38834774e-01 -3.45883489e-01 6.31084263e-01 -1.47643059e-01 -5.86203635e-01 -2.70614654e-01 -6.70725226e-01 -8.15490186e-01 -7.95376003e-01 2.46841058e-01 3.59551370e-01 -2.24555694e-02 2.85705745e-01 2.18328252e-01 9.17953551e-02 -1.28516340e+00 -6.51696026e-02 9.26077724e-01 1.46293128e+00 5.38690984e-01 4.80025172e-01 -4.92633656e-02 7.05765724e-01 -1.72113970e-01 -6.23873949e-01 2.45000690e-01 -2.18807101e-01 -4.61033314e-01 4.37701307e-02 -1.98823344e-02 -4.35287893e-01 -3.67260784e-01 1.14537954e+00 6.85251474e-01 6.56853290e-03 4.27750915e-01 -1.28855038e+00 -4.56231646e-02 6.36946857e-01 -6.98677003e-01 4.12691414e-01 -1.30552500e-01 -6.54202253e-02 8.03196430e-01 -6.73115313e-01 2.08134189e-01 8.97602558e-01 6.42549098e-01 1.28951430e-01 -9.20843601e-01 -6.89077616e-01 2.57984191e-01 7.09842026e-01 -1.81393039e+00 -6.53066516e-01 7.96021163e-01 5.76064028e-02 9.41473126e-01 1.47487834e-01 7.17093766e-01 8.38663876e-01 3.03907275e-01 7.89928377e-01 7.58705735e-01 -4.26705539e-01 6.52439415e-01 2.22339123e-01 -1.43776774e-01 8.22467327e-01 2.54864961e-01 -3.57136399e-01 -5.94556093e-01 -1.73892379e-01 6.91545427e-01 2.74767011e-01 -1.36230811e-01 -1.47253886e-01 -6.65830612e-01 5.12304485e-01 3.35596323e-01 -5.56138530e-02 -5.47972739e-01 4.11701679e-01 5.30220389e-01 1.71693582e-02 2.07480446e-01 -2.95227587e-01 -5.25207460e-01 -5.89858651e-01 -8.27786028e-01 4.09158356e-02 7.59224057e-01 8.34738553e-01 6.96564019e-01 -1.17231056e-01 -3.67148668e-02 7.82031476e-01 4.09245461e-01 3.88823539e-01 5.74722350e-01 -7.25221753e-01 9.22897935e-01 5.35220921e-01 -6.97001517e-02 -1.26679611e+00 -1.01831950e-01 -7.25788057e-01 -1.07954383e+00 -4.82522726e-01 -1.53261989e-01 -5.27784169e-01 -4.49259251e-01 1.53532064e+00 6.03603661e-01 7.27412641e-01 -9.42207575e-02 8.89690876e-01 2.51830310e-01 7.59248197e-01 -1.38337566e-02 -4.25574064e-01 1.40725720e+00 -1.14291286e+00 -7.44229674e-01 -2.94859141e-01 6.76095128e-01 -5.80405354e-01 8.15713823e-01 3.12561959e-01 -1.20777118e+00 -6.10348284e-01 -1.27292180e+00 -1.11265607e-01 -2.08837539e-02 5.78884542e-01 6.89288855e-01 5.56482077e-01 -7.94642925e-01 4.35511857e-01 -1.10860109e+00 9.97815356e-02 5.40456772e-01 8.88786912e-01 1.54339939e-01 -8.44067559e-02 -1.01592290e+00 3.93027723e-01 5.35765886e-01 5.05398095e-01 -5.24439335e-01 -3.40719223e-01 -3.96231145e-01 3.48838627e-01 5.71082950e-01 -8.66984069e-01 1.16581893e+00 -8.57491434e-01 -1.81957829e+00 3.44195694e-01 -3.04160297e-01 -7.14162469e-01 3.29738021e-01 -5.75380623e-02 -3.74054551e-01 -3.72847356e-03 -2.14115605e-01 4.00023997e-01 1.09912908e+00 -7.65514195e-01 -1.02409375e+00 -6.76781118e-01 -3.37214358e-02 6.77095532e-01 -7.90449023e-01 -3.61396223e-01 -1.04654002e+00 -3.65684688e-01 -2.65601333e-02 -1.09203374e+00 -2.40815803e-01 -2.45921835e-01 -3.77802342e-01 -2.55824447e-01 8.10145915e-01 -8.34213436e-01 1.53356206e+00 -2.08450580e+00 1.24176651e-01 2.76900858e-01 1.40892744e-01 4.03274477e-01 3.62172931e-01 -3.11231619e-04 4.36147928e-01 -1.52563944e-01 6.41428828e-02 -4.39326704e-01 -8.10087770e-02 2.33002469e-01 7.91623145e-02 2.49469772e-01 -2.23913506e-01 8.64487708e-01 -4.22044992e-01 -6.92107618e-01 6.48831502e-02 5.73937416e-01 -7.28402317e-01 2.15943769e-01 -1.29824504e-01 1.26160309e-01 -8.25672090e-01 4.31177646e-01 8.10471416e-01 -5.34457862e-01 3.44731987e-01 -3.34212720e-01 2.14316875e-01 2.32248291e-01 -1.22337699e+00 1.61884201e+00 -9.82135475e-01 1.94650576e-01 1.58952326e-01 -1.17273355e+00 7.56565332e-01 1.70538738e-01 2.78913110e-01 -7.12616384e-01 3.63727838e-01 2.53353536e-01 -4.26832400e-02 -6.49246097e-01 3.35438728e-01 2.59538859e-01 8.58721063e-02 3.11335146e-01 -3.46728981e-01 5.42553186e-01 -9.43384320e-02 -2.01995641e-01 9.76484120e-01 -4.77246568e-02 2.46354222e-01 -1.15429178e-01 7.46188700e-01 -3.61809760e-01 7.57229388e-01 4.75000143e-01 1.73544860e-03 -1.03724577e-01 3.79062295e-01 -4.15599763e-01 -6.40330374e-01 -5.46890855e-01 5.83705045e-02 9.34346318e-01 3.88810903e-01 -4.70204532e-01 -1.02457118e+00 -5.15271246e-01 -2.79610962e-01 3.58141899e-01 -1.82711899e-01 -3.36345226e-01 -5.95114052e-01 -8.28344882e-01 1.91754296e-01 4.57991898e-01 9.15692031e-01 -5.75493753e-01 -7.81577289e-01 1.16778694e-01 -2.41588578e-01 -1.30276036e+00 -5.41855693e-01 3.70645821e-01 -1.03197920e+00 -6.60805941e-01 -4.40623045e-01 -1.00024092e+00 7.44548500e-01 3.85245711e-01 7.28225410e-01 1.42709792e-01 2.75574513e-02 -3.14787120e-01 -1.30746499e-01 -5.14702089e-02 8.13439935e-02 6.48834765e-01 -8.37390199e-02 3.04924667e-01 2.60023475e-01 -7.18757987e-01 -8.93516600e-01 1.69372797e-01 -8.96663964e-01 4.22876358e-01 8.97908151e-01 8.82343113e-01 8.65246832e-01 5.90462029e-01 4.34103966e-01 -9.09915209e-01 6.58218741e-01 -5.60967028e-01 -6.75742567e-01 3.38922322e-01 -6.69667065e-01 1.93405390e-01 1.03540230e+00 -2.86898464e-01 -1.15336156e+00 2.13205904e-01 -1.69366568e-01 -4.59320486e-01 2.53656238e-01 7.22145379e-01 -6.07385933e-01 3.00414786e-02 -3.55643481e-02 4.04882610e-01 -2.42599905e-01 -2.98386604e-01 6.57313094e-02 1.08911633e+00 4.03000563e-01 -4.29443508e-01 3.30080926e-01 3.07423979e-01 2.62538362e-02 -6.52056396e-01 -5.56150258e-01 -2.69876182e-01 -2.03728169e-01 2.07721233e-01 5.09580851e-01 -1.19409847e+00 -9.34284627e-01 2.88517177e-01 -8.70974243e-01 -1.51503339e-01 2.19507247e-01 4.04205501e-01 -3.99220377e-01 1.80108458e-01 -3.07096690e-01 -7.05825686e-01 -6.42803371e-01 -1.54068029e+00 1.13660491e+00 3.97647768e-01 1.98934734e-01 -8.80857348e-01 -3.43428820e-01 6.21241927e-01 3.56295913e-01 -2.63070464e-01 1.00184202e+00 -3.81960958e-01 -6.85818553e-01 -2.48757794e-01 -3.75912964e-01 9.66281965e-02 1.45987598e-02 -4.53063011e-01 -7.25175500e-01 -3.24363738e-01 2.46309385e-01 -1.09688796e-01 4.64226991e-01 4.55056489e-01 1.71368408e+00 -6.40072107e-01 -6.43100739e-01 9.65485811e-01 1.46692216e+00 3.23096514e-01 3.98987144e-01 6.40605092e-02 8.08613598e-01 -1.47890821e-01 4.92205471e-01 7.95778930e-01 5.98910511e-01 9.53964114e-01 3.40705365e-01 1.41405821e-01 8.50573406e-02 -4.12653953e-01 2.09227517e-01 1.21735322e+00 1.30419254e-01 -2.05781728e-01 -6.44092083e-01 1.45634487e-01 -1.99056613e+00 -4.68524516e-01 4.03993607e-01 2.31687427e+00 8.59533191e-01 3.67299795e-01 4.66700038e-03 2.70991713e-01 6.96500838e-01 3.07114106e-02 -1.00620902e+00 -1.95667997e-01 3.59663427e-01 -7.14121982e-02 5.69335938e-01 3.41456562e-01 -8.90793920e-01 7.28021502e-01 4.50727654e+00 1.37301362e+00 -1.43841827e+00 1.96232915e-01 1.10863984e+00 -5.80777265e-02 -8.60329494e-02 -1.03457682e-02 -1.26657987e+00 9.02054489e-01 1.24808896e+00 5.67783136e-03 8.17671180e-01 9.55444694e-01 3.86636615e-01 -7.80753717e-02 -9.13868010e-01 1.55096030e+00 -5.46838194e-02 -1.22837186e+00 3.78550030e-02 1.84064910e-01 7.43371725e-01 -7.68620446e-02 1.28395453e-01 3.28053981e-01 -2.30006233e-01 -6.74101591e-01 3.71343404e-01 2.11631089e-01 7.80256331e-01 -1.05028296e+00 7.55468369e-01 5.74131727e-01 -1.32613528e+00 -2.83036202e-01 -4.07309979e-01 -7.80645013e-02 5.86448424e-02 6.59973562e-01 -7.34767497e-01 2.18724772e-01 6.34967744e-01 5.40020883e-01 -3.75923336e-01 7.40311027e-01 1.16107337e-01 5.00245750e-01 -4.01981473e-01 -3.40682387e-01 2.25679744e-02 -1.92294344e-01 -2.15953141e-02 8.81055951e-01 3.15750867e-01 1.84467547e-02 2.99517483e-01 3.93414438e-01 -5.14980078e-01 1.14257611e-01 -2.92783678e-01 5.91380820e-02 8.33825111e-01 1.32369423e+00 -8.95833194e-01 -2.35075220e-01 -3.49789858e-01 1.11108971e+00 5.06417274e-01 1.92245811e-01 -1.15958428e+00 -2.88755059e-01 4.11640465e-01 5.98097444e-02 5.60086668e-01 -1.59594223e-01 -4.60086405e-01 -1.01217401e+00 2.89485872e-01 -8.39593112e-01 1.14382766e-01 -3.50306571e-01 -6.07794583e-01 4.66842115e-01 -1.81162715e-01 -1.15938377e+00 -2.66237289e-01 -2.14122012e-01 -4.36801374e-01 6.06225908e-01 -1.52997148e+00 -1.02976429e+00 -2.87158191e-01 5.84159493e-01 7.05424368e-01 -1.19036824e-01 5.23870468e-01 5.64295471e-01 -1.08257556e+00 1.02247798e+00 3.48453164e-01 -2.85995901e-01 -4.98176478e-02 -5.69911420e-01 2.87688136e-01 7.03936517e-01 1.48121729e-01 4.09630597e-01 3.68735015e-01 -4.58176881e-01 -2.01491904e+00 -1.05875337e+00 7.47981608e-01 4.42254484e-01 3.74229819e-01 -3.21795136e-01 -4.66633141e-01 4.23880160e-01 1.01236301e-02 1.11589812e-01 6.07761919e-01 -5.88797256e-02 2.68822551e-01 -5.87442636e-01 -7.97303081e-01 7.92250574e-01 1.08784723e+00 -4.11917955e-01 2.84756154e-01 3.53656977e-01 5.89898407e-01 -5.49735069e-01 -7.66566455e-01 3.34524512e-01 5.31436741e-01 -7.60591149e-01 7.78204501e-01 -1.25466660e-01 4.00026768e-01 -1.49008036e-01 -1.15969956e-01 -1.02353358e+00 -5.16914204e-02 -7.86309719e-01 -7.24572599e-01 1.09957385e+00 3.36789548e-01 -3.87902051e-01 1.14272416e+00 5.51769853e-01 1.26389027e-01 -1.39007652e+00 -1.02051270e+00 -3.23275089e-01 -6.62435830e-01 -4.21398133e-01 6.96615696e-01 3.88144791e-01 -2.21642673e-01 7.20058739e-01 -4.76993978e-01 2.37144738e-01 3.42989504e-01 2.09499031e-01 7.94715762e-01 -8.90591860e-01 -7.68572271e-01 -1.86914608e-01 -2.44864061e-01 -1.61115456e+00 1.00030176e-01 -7.22674310e-01 -3.93592380e-02 -1.17944956e+00 1.44474015e-01 -8.11381996e-01 -2.62427688e-01 3.07518274e-01 -1.86568648e-01 -1.31788361e-03 -1.26278952e-01 2.64521569e-01 -9.76147711e-01 7.73978651e-01 9.42659080e-01 -2.44716778e-02 -3.33504707e-01 3.94441158e-01 -7.38125920e-01 6.99295819e-01 8.11721802e-01 -3.35164249e-01 -7.82200933e-01 -8.73238266e-01 3.74788851e-01 5.12570500e-01 -1.30237997e-01 -1.11096954e+00 5.96452951e-01 2.42823362e-02 2.31656894e-01 -4.00809020e-01 5.15527427e-01 -1.17008865e+00 1.38018593e-01 5.95191240e-01 -2.57818997e-01 2.90896088e-01 6.35893196e-02 8.27424884e-01 -1.07894890e-01 -2.12425604e-01 4.87185866e-01 1.75970316e-01 -6.88976228e-01 4.26622391e-01 -1.37979060e-01 -1.39617994e-01 9.80527878e-01 -2.15073526e-01 1.28012985e-01 -2.75300831e-01 -3.53010267e-01 1.36355951e-01 -3.08609474e-02 2.99997568e-01 5.59776127e-01 -1.25707042e+00 -2.82604754e-01 4.59559977e-01 -2.90479064e-01 1.99431837e-01 3.77512753e-01 9.51957285e-01 -4.90186632e-01 5.45183957e-01 4.45015766e-02 -5.51723301e-01 -1.20915484e+00 2.13438660e-01 1.25237226e-01 -5.93398750e-01 -2.83521920e-01 8.88224423e-01 -2.11693376e-01 7.52807707e-02 5.76531410e-01 -2.83063203e-01 6.24049120e-02 -3.31040978e-01 4.66399133e-01 6.20258749e-01 3.91077936e-01 -4.42370355e-01 -1.43852636e-01 4.13371503e-01 -3.63401830e-01 2.41522059e-01 1.13164818e+00 -5.43079197e-01 -6.40083803e-03 6.66877627e-02 1.43213511e+00 -3.89530063e-01 -1.30882788e+00 -3.90872747e-01 -1.75027162e-01 -2.72799075e-01 7.20653057e-01 -3.93254280e-01 -1.23181176e+00 7.86055207e-01 8.40551376e-01 -6.13919161e-02 1.41966474e+00 -3.99865478e-01 1.37740278e+00 5.60684443e-01 4.53817546e-01 -1.27495718e+00 -1.54454380e-01 2.66212612e-01 1.98812291e-01 -1.20740545e+00 8.09241738e-03 -4.50993925e-01 -3.12915742e-01 8.33709717e-01 7.45772004e-01 1.06547259e-01 9.30729151e-01 4.45452571e-01 -4.24112111e-01 1.61092028e-01 -8.18828523e-01 2.53139973e-01 -1.01362858e-02 6.35788068e-02 1.24510117e-01 1.30723655e-01 -4.44722086e-01 7.88361669e-01 -1.64277643e-01 4.68708485e-01 -8.88562128e-02 8.61753523e-01 -1.88577235e-01 -1.11040270e+00 -5.45786358e-02 7.78419912e-01 -4.80531961e-01 3.15485857e-02 2.92233169e-01 1.01957344e-01 3.33250880e-01 8.24174821e-01 1.22973971e-01 -7.46071219e-01 -1.13381162e-01 -2.30839193e-01 2.84777224e-01 -2.28306100e-01 -3.74957860e-01 8.08293968e-02 -1.79805681e-01 -3.31578165e-01 -1.42360389e-01 -3.31177175e-01 -1.18175471e+00 -4.12736416e-01 -6.09996974e-01 2.00478047e-01 9.15191233e-01 1.24423099e+00 8.99960577e-01 5.65883160e-01 8.09821784e-01 -7.90186644e-01 -8.43832850e-01 -6.36690259e-01 -4.08974975e-01 -3.99616873e-03 -4.05019447e-02 -5.40135860e-01 1.08403169e-01 -4.83869500e-02]
[8.494843482971191, 2.9385597705841064]
4cfb36e0-3412-4061-847f-f8531a2db30f
the-acii-2022-affective-vocal-bursts-workshop
2207.03572
null
https://arxiv.org/abs/2207.03572v2
https://arxiv.org/pdf/2207.03572v2.pdf
The ACII 2022 Affective Vocal Bursts Workshop & Competition: Understanding a critically understudied modality of emotional expression
The ACII Affective Vocal Bursts Workshop & Competition is focused on understanding multiple affective dimensions of vocal bursts: laughs, gasps, cries, screams, and many other non-linguistic vocalizations central to the expression of emotion and to human communication more generally. This year's competition comprises four tracks using a large-scale and in-the-wild dataset of 59,299 vocalizations from 1,702 speakers. The first, the A-VB-High task, requires competition participants to perform a multi-label regression on a novel model for emotion, utilizing ten classes of richly annotated emotional expression intensities, including; Awe, Fear, and Surprise. The second, the A-VB-Two task, utilizes the more conventional 2-dimensional model for emotion, arousal, and valence. The third, the A-VB-Culture task, requires participants to explore the cultural aspects of the dataset, training native-country dependent models. Finally, for the fourth task, A-VB-Type, participants should recognize the type of vocal burst (e.g., laughter, cry, grunt) as an 8-class classification. This paper describes the four tracks and baseline systems, which use state-of-the-art machine learning methods. The baseline performance for each track is obtained by utilizing an end-to-end deep learning model and is as follows: for A-VB-High, a mean (over the 10-dimensions) Concordance Correlation Coefficient (CCC) of 0.5687 CCC is obtained; for A-VB-Two, a mean (over the 2-dimensions) CCC of 0.5084 is obtained; for A-VB-Culture, a mean CCC from the four cultures of 0.4401 is obtained; and for A-VB-Type, the baseline Unweighted Average Recall (UAR) from the 8-classes is 0.4172 UAR.
['Alan Cowen', 'Dacher Keltner', 'Anton Batliner', 'Björn Schuller', 'Christopher B. Gregory', 'Jeffrey A. Brooks', 'Panagiotis Tzirakis', 'Alice Baird']
2022-07-07
null
null
null
null
['a-vb-culture', 'a-vb-high', 'a-vb-two']
['speech', 'speech', 'speech']
[-4.02461320e-01 -2.44499251e-01 1.08615562e-01 -4.93345469e-01 -9.11977708e-01 -7.17359662e-01 2.93284029e-01 -1.13655552e-01 -3.72712493e-01 3.38414818e-01 2.07766145e-01 2.51687288e-01 2.36346349e-01 -8.81675705e-02 -2.01123789e-01 -5.88453412e-01 -2.56158084e-01 2.53890842e-01 -5.35269201e-01 -4.21040654e-01 -1.98385909e-01 2.94601202e-01 -1.41735470e+00 5.81173837e-01 1.22506090e-01 1.50149596e+00 -5.08594871e-01 9.16540325e-01 3.08420602e-02 6.37278318e-01 -8.80301237e-01 -6.33471012e-01 -8.38177875e-02 -5.83066225e-01 -8.53158116e-01 -1.46833181e-01 3.38973612e-01 6.25086799e-02 3.90506893e-01 7.06443727e-01 7.51200855e-01 3.42384487e-01 7.41399288e-01 -1.36046159e+00 -3.75407338e-01 4.58708256e-01 -6.37704253e-01 2.91215498e-02 4.68104094e-01 6.19396158e-02 1.26909769e+00 -8.83141816e-01 5.02029777e-01 1.33859551e+00 1.00849366e+00 9.33594823e-01 -1.17046070e+00 -1.25520754e+00 3.40173878e-02 -1.52422681e-01 -1.11787033e+00 -1.18443638e-01 7.21647024e-01 -7.43273258e-01 1.01105034e+00 6.01998329e-01 1.03612542e+00 1.54139018e+00 -9.47318785e-03 5.94296515e-01 1.52680063e+00 -1.45910040e-01 2.15055481e-01 5.99814534e-01 4.45587002e-02 5.35642505e-01 -7.10388184e-01 2.17853427e-01 -7.03019559e-01 -2.15121850e-01 2.83513293e-02 -5.56374729e-01 3.12699527e-02 5.22429466e-01 -7.42711902e-01 9.18857872e-01 3.85626405e-01 4.83939469e-01 -2.39002928e-01 -7.83605874e-03 8.87005627e-01 5.03860593e-01 6.78929985e-01 8.56816649e-01 -5.12480915e-01 -6.68323755e-01 -7.64787853e-01 3.08369815e-01 9.48242664e-01 5.55849969e-01 5.56469202e-01 2.60436445e-01 -5.24683995e-03 1.46432030e+00 8.64172801e-02 3.14571828e-01 7.21136928e-01 -9.06189442e-01 -1.42898113e-01 7.91182220e-02 -9.67868343e-02 -9.03656960e-01 -9.79619980e-01 -5.13191223e-01 -4.43465561e-01 6.42393827e-02 7.31960014e-02 -7.17730522e-01 -5.29617786e-01 2.09030795e+00 1.42481074e-01 -7.60665163e-02 9.85361561e-02 9.64982927e-01 1.23565507e+00 8.31151485e-01 1.22290038e-01 -4.65036184e-01 1.38043582e+00 -1.12322700e+00 -6.07895255e-01 -2.83041447e-01 5.04486680e-01 -8.94666433e-01 1.53012145e+00 7.01902211e-01 -8.76051366e-01 -5.98106802e-01 -8.59529972e-01 3.04975301e-01 -5.64198375e-01 6.23911321e-02 6.33131266e-01 1.01719368e+00 -8.99048150e-01 1.34411767e-01 -1.32551596e-01 -4.67299104e-01 -2.05574274e-01 3.09214592e-01 -4.37359780e-01 5.49225330e-01 -1.22477472e+00 8.93316746e-01 -1.46466911e-01 1.28341287e-01 -7.43209839e-01 -7.66731560e-01 -5.51141858e-01 -1.60136610e-01 -2.57594317e-01 1.11978509e-01 1.52453446e+00 -1.41307700e+00 -1.89770353e+00 1.18801713e+00 2.49033540e-01 1.47570327e-01 1.46370411e-01 -2.23969728e-01 -8.93110871e-01 -1.54406205e-02 -2.56674811e-02 7.01020539e-01 6.22783899e-01 -1.48305392e+00 -2.94839919e-01 -3.13715637e-03 -3.11428159e-01 1.34231061e-01 -4.26202297e-01 5.21753371e-01 -2.99403388e-02 -5.13249934e-01 -4.86462057e-01 -1.34490693e+00 2.43234619e-01 -6.21002734e-01 -3.04669410e-01 -4.07507569e-01 6.31799221e-01 -5.89451969e-01 1.24327219e+00 -2.52134228e+00 1.00179262e-01 2.65875548e-01 8.03736672e-02 -5.22192940e-02 -2.84835786e-01 3.96216780e-01 -3.72531474e-01 2.59768754e-01 3.11469194e-03 -5.61049163e-01 1.52446210e-01 5.27841598e-02 -1.23362161e-01 2.74803251e-01 7.82713220e-02 3.40293258e-01 -8.74253094e-01 -1.58298507e-01 -1.64460197e-01 2.32654974e-01 -7.79017568e-01 6.73698604e-01 1.36942700e-01 4.26099449e-01 5.45816943e-02 4.78337079e-01 3.53451014e-01 4.07647759e-01 2.48353593e-02 -8.85659605e-02 -2.40817577e-01 -1.22540511e-01 -5.63979805e-01 1.39309156e+00 -8.12933207e-01 7.91572630e-01 4.94714886e-01 -3.00639093e-01 1.18064785e+00 4.97127563e-01 6.53260648e-01 -4.11737621e-01 3.83395284e-01 3.62216622e-01 2.81080306e-01 -5.99967778e-01 3.22895527e-01 -8.37670028e-01 -7.33827055e-01 4.17921931e-01 3.10843527e-01 -6.24545991e-01 -9.29446891e-02 6.12303801e-02 8.44426095e-01 -2.38641709e-01 -1.22737527e-01 -1.78107426e-01 1.68422520e-01 -3.62030119e-01 5.40082276e-01 2.97471017e-01 -5.74622095e-01 3.91241252e-01 7.49375224e-01 -2.32057795e-01 -4.50752079e-01 -1.03332531e+00 -9.42846164e-02 1.54283488e+00 -3.63279700e-01 -4.54897106e-01 -7.65780270e-01 -3.92634988e-01 -1.83084413e-01 9.31789756e-01 -9.29947913e-01 -3.86010647e-01 -1.81141526e-01 -6.41615331e-01 9.99329329e-01 3.19528103e-01 1.15610197e-01 -1.29850090e+00 -4.50391442e-01 3.11964341e-02 -5.06818771e-01 -1.08691335e+00 -5.51474452e-01 4.99602735e-01 3.36702503e-02 -4.23282921e-01 -4.30763721e-01 -7.62461722e-01 -1.58320859e-01 -4.95784223e-01 1.08171153e+00 -1.64014861e-01 -3.68492305e-01 5.56095779e-01 -4.88466591e-01 -6.12766802e-01 -3.30487013e-01 -2.94900268e-01 3.47793102e-01 1.85020238e-01 3.22779089e-01 -4.89803284e-01 -2.49545261e-01 2.40885377e-01 -3.14731538e-01 -4.29897934e-01 2.05644652e-01 9.10272419e-01 2.30886176e-01 -5.06783783e-01 8.68153393e-01 -5.82096219e-01 1.04223824e+00 -6.70267344e-01 1.59504458e-01 -9.61568356e-02 -2.26941019e-01 -6.46564662e-01 5.66070557e-01 -8.86234820e-01 -8.10051143e-01 -1.13098785e-01 -5.66337824e-01 -6.48301482e-01 -2.97663301e-01 4.85654771e-01 2.33096674e-01 1.98518187e-01 7.52287865e-01 -3.66663694e-01 -1.34712175e-01 -2.69318819e-01 4.77916479e-01 1.11244023e+00 6.41299427e-01 -7.77132273e-01 2.91211486e-01 -2.27964804e-01 -4.76065695e-01 -1.21527207e+00 -1.01885223e+00 -6.14679098e-01 -3.47527444e-01 -8.50917459e-01 1.17293918e+00 -1.04175103e+00 -1.01936626e+00 5.83950996e-01 -9.41927075e-01 -6.81095183e-01 -1.03840873e-01 6.27469599e-01 -6.37524962e-01 -2.71314979e-01 -9.41896677e-01 -1.10670757e+00 -5.24285316e-01 -1.07599401e+00 9.47563350e-01 2.21891761e-01 -1.05312908e+00 -6.79552495e-01 5.05057335e-01 4.67171997e-01 2.61070877e-01 5.60626149e-01 1.09372842e+00 -8.79477262e-01 8.86288583e-01 -8.70084390e-02 1.26092896e-01 6.51664913e-01 6.39402941e-02 1.66963011e-01 -1.39038515e+00 -1.82748064e-01 7.70613998e-02 -1.24898374e+00 4.00320619e-01 7.99723119e-02 1.12985015e+00 -1.94780812e-01 3.62406760e-01 4.67065573e-01 7.85807133e-01 4.04630244e-01 1.95482209e-01 -1.25158533e-01 3.37742597e-01 7.70362675e-01 6.60762727e-01 4.44752812e-01 2.44840592e-01 7.31027722e-01 3.20613384e-01 -2.24090904e-01 1.85971409e-01 4.11568955e-02 7.33580410e-01 1.36351800e+00 6.99496195e-02 -3.06800269e-02 -6.80702806e-01 3.75024408e-01 -1.20083320e+00 -1.03584111e+00 -5.27726561e-02 1.74614143e+00 9.48511720e-01 -2.39849806e-01 7.03102231e-01 -9.12373960e-02 4.63039219e-01 2.09222823e-01 -3.63108963e-01 -1.62401140e+00 7.85019249e-02 3.55472356e-01 -4.56986338e-01 5.38309455e-01 -9.88625765e-01 9.82401133e-01 6.58510733e+00 6.59660816e-01 -1.59026265e+00 1.03774443e-01 7.96986938e-01 -6.20444179e-01 7.27349594e-02 -6.48669839e-01 -5.04017770e-01 2.92147219e-01 1.24047554e+00 6.41136169e-02 7.39743292e-01 9.45053041e-01 1.10424034e-01 1.00418217e-01 -1.07962465e+00 1.14469039e+00 4.22174603e-01 -3.69824082e-01 -5.84025681e-01 -2.99845010e-01 6.71329856e-01 -6.77971318e-02 3.22693795e-01 9.99870777e-01 1.06889881e-01 -1.09736907e+00 1.07364273e+00 2.00580195e-01 1.12969828e+00 -1.06163967e+00 6.65213823e-01 1.13287501e-01 -9.48046386e-01 -1.54847503e-01 8.95949975e-02 -8.58988911e-02 3.93460989e-02 2.71490932e-01 -7.03648448e-01 7.15439171e-02 1.09073710e+00 4.26311970e-01 -2.14349195e-01 3.94720793e-01 3.79035510e-02 9.73956287e-01 -1.93048358e-01 -5.36997855e-01 4.96541589e-01 -1.60689667e-01 4.90818948e-01 2.03497195e+00 2.49586612e-01 2.21905023e-01 1.40239805e-01 5.73471308e-01 -2.23480031e-01 4.19029146e-01 -3.39578688e-01 -2.78091669e-01 2.78490067e-01 1.82071698e+00 -3.89462620e-01 -5.75096868e-02 -1.75780296e-01 8.68859053e-01 2.99788684e-01 1.48910478e-01 -1.09488750e+00 -4.13320035e-01 8.80735517e-01 -3.57225627e-01 -5.38353622e-02 1.96362630e-01 -1.37156427e-01 -6.54821694e-01 -4.29342449e-01 -1.30792773e+00 2.90545017e-01 -9.61994946e-01 -1.65224075e+00 1.04038715e+00 -1.06303394e-01 -9.68767822e-01 -4.65065271e-01 -5.94465613e-01 -7.18334675e-01 8.77479970e-01 -7.48239756e-01 -1.07318425e+00 -3.17714393e-01 5.78545690e-01 4.46145505e-01 -2.28032261e-01 1.20845926e+00 3.64413649e-01 -5.72652102e-01 8.85349870e-01 -3.04115802e-01 1.66318029e-01 1.05052209e+00 -1.24728227e+00 -3.48459810e-01 3.16331945e-02 -9.84096914e-05 4.25934464e-01 6.94532812e-01 -1.47532433e-01 -9.51456904e-01 -9.89016294e-01 7.22227216e-01 -1.93422273e-01 8.26248884e-01 -6.04305506e-01 -5.98414242e-01 4.31861490e-01 3.74536127e-01 -1.96288288e-01 1.45252204e+00 7.59152949e-01 -7.12438107e-01 -1.78500265e-01 -1.20661783e+00 4.37906802e-01 4.44402933e-01 -7.62408853e-01 -2.63442069e-01 3.24533343e-01 6.25936389e-01 -3.61792743e-01 -1.16932857e+00 1.28697649e-01 1.08082831e+00 -1.14051831e+00 5.80120504e-01 -8.08210015e-01 5.75879872e-01 2.42928982e-01 -4.48262006e-01 -1.77773523e+00 -2.92241275e-01 -6.29014552e-01 4.90699947e-01 1.42539895e+00 5.81692517e-01 -3.01644415e-01 2.37502083e-01 3.76180917e-01 -4.45372850e-01 -9.63426471e-01 -9.96185660e-01 -6.65994585e-01 4.10488755e-01 -6.10482872e-01 2.39226341e-01 1.24160933e+00 3.60136151e-01 7.25139558e-01 -5.95509171e-01 -3.58119905e-01 -8.73346254e-03 5.68147562e-02 7.81555712e-01 -1.04908073e+00 -3.91316682e-01 -7.74516165e-01 -1.66367278e-01 -4.09234673e-01 3.93441826e-01 -8.23335588e-01 4.34261262e-01 -9.25418615e-01 1.25881121e-01 -2.34562933e-01 -3.90082955e-01 5.36181450e-01 1.12457491e-01 2.71510273e-01 3.60092700e-01 -2.24997848e-01 -2.39953399e-01 5.13593853e-01 8.60138893e-01 7.92422146e-02 -3.78305644e-01 5.16948989e-04 -7.13268638e-01 8.82613122e-01 7.88660526e-01 -3.70704174e-01 -2.33633518e-01 -3.04054492e-03 1.20614238e-01 1.76728159e-01 9.63643044e-02 -9.02492464e-01 -3.93170655e-01 -1.15449324e-01 2.45388806e-01 -2.76992798e-01 1.07529163e+00 -4.52832282e-01 6.37493059e-02 2.38105625e-01 -5.25731742e-01 2.13563994e-01 5.06570756e-01 1.91981480e-01 -3.13385546e-01 -2.92939302e-02 1.25426221e+00 -3.59954759e-02 -2.96756864e-01 -1.64615542e-01 -7.10197866e-01 3.68299454e-01 8.28267932e-01 1.75203055e-01 -2.33266801e-01 -7.72447944e-01 -9.33273077e-01 1.94799248e-02 -8.25716704e-02 7.38198698e-01 3.94681603e-01 -1.39952636e+00 -6.98662877e-01 2.66315304e-02 1.90497696e-01 -6.93840861e-01 3.78581017e-01 8.84796143e-01 -9.00652334e-02 -1.46362379e-01 -2.98971087e-01 -5.26648462e-01 -1.54300559e+00 1.41903192e-01 5.54438353e-01 -1.77229285e-01 1.01125360e-01 1.36551940e+00 1.44266143e-01 -6.58439517e-01 1.73268348e-01 -1.40449196e-01 -1.23330317e-01 6.26052558e-01 3.54761034e-01 3.73558939e-01 -1.06947217e-02 -1.02291763e+00 -4.92165536e-01 4.71366614e-01 1.68050423e-01 -5.86393774e-01 1.28108001e+00 2.70833224e-01 -2.33183458e-01 1.22419667e+00 1.71973026e+00 3.33277792e-01 -5.49813151e-01 3.84035826e-01 -3.02937359e-01 -3.06813908e-03 -6.26606643e-02 -1.31564486e+00 -9.52996671e-01 1.09953809e+00 7.37540483e-01 2.82591462e-01 1.24158216e+00 1.07204057e-02 7.03738034e-01 8.63552466e-02 -1.27304986e-01 -1.27320921e+00 4.84494656e-01 7.11715698e-01 1.38654065e+00 -9.37285841e-01 -3.73924017e-01 -8.72707665e-02 -1.40719700e+00 9.27039742e-01 9.90665972e-01 2.61026800e-01 6.10774457e-01 2.92417258e-01 7.00941086e-01 -2.82956362e-01 -9.77003753e-01 3.63324545e-02 2.97722161e-01 3.19454432e-01 7.73129642e-01 5.42216063e-01 -1.31725684e-01 1.15519488e+00 -8.79250169e-01 -5.48372447e-01 1.85252666e-01 2.55487502e-01 -1.87194452e-01 -6.27655327e-01 -1.95185125e-01 4.66151744e-01 -5.56684792e-01 1.13071479e-01 -1.09904134e+00 7.29464352e-01 2.41606846e-01 1.37232637e+00 1.76513091e-01 -1.04006183e+00 5.94567537e-01 3.00036192e-01 2.79003948e-01 -6.06730759e-01 -1.49932742e+00 4.18132037e-01 6.03684306e-01 -3.96214902e-01 -3.09253186e-01 -8.13756883e-01 -1.25779808e+00 -1.89120069e-01 -1.94144309e-01 4.37327504e-01 7.54580736e-01 5.72447240e-01 -1.00953504e-01 3.39994550e-01 1.03759599e+00 -8.41475487e-01 -3.76720667e-01 -1.37224567e+00 -8.39516580e-01 6.03279769e-01 3.21775049e-01 -6.41549230e-01 -6.94097817e-01 -2.88419962e-01]
[13.528290748596191, 5.661630153656006]
75f755a0-3eab-4218-9cfc-85cb65a15522
a-comparison-of-architectures-and-pretraining
1912.10169
null
https://arxiv.org/abs/1912.10169v1
https://arxiv.org/pdf/1912.10169v1.pdf
A Comparison of Architectures and Pretraining Methods for Contextualized Multilingual Word Embeddings
The lack of annotated data in many languages is a well-known challenge within the field of multilingual natural language processing (NLP). Therefore, many recent studies focus on zero-shot transfer learning and joint training across languages to overcome data scarcity for low-resource languages. In this work we (i) perform a comprehensive comparison of state-ofthe-art multilingual word and sentence encoders on the tasks of named entity recognition (NER) and part of speech (POS) tagging; and (ii) propose a new method for creating multilingual contextualized word embeddings, compare it to multiple baselines and show that it performs at or above state-of-theart level in zero-shot transfer settings. Finally, we show that our method allows for better knowledge sharing across languages in a joint training setting.
['Ekaterina Shutova', 'Samira Abnar', 'Niels van der Heijden']
2019-12-15
null
null
null
null
['multilingual-word-embeddings']
['methodology']
[-2.64542878e-01 -3.48900035e-02 -4.84844595e-01 -3.54302108e-01 -1.35112357e+00 -7.32841253e-01 6.29092932e-01 4.10673946e-01 -1.20345175e+00 9.90518808e-01 6.23945177e-01 -4.78626519e-01 4.37635303e-01 -5.39159656e-01 -7.29333580e-01 -2.28991792e-01 -5.33615164e-02 4.39429909e-01 1.04553938e-01 -2.89357692e-01 -2.80459464e-01 2.71804035e-01 -7.30490863e-01 -2.89495029e-02 1.01168537e+00 9.19770300e-02 3.59299511e-01 4.24147129e-01 -3.66934389e-01 5.29774070e-01 -2.75190353e-01 -7.57957578e-01 2.68756673e-02 -6.77487254e-02 -9.75597799e-01 -4.44563419e-01 4.01019603e-01 4.03930359e-02 -1.14789695e-01 1.16347337e+00 8.52289021e-01 3.17933917e-01 4.87867504e-01 -8.80952954e-01 -1.30576885e+00 7.31329322e-01 -2.25718603e-01 3.41672629e-01 1.89333379e-01 1.81344580e-02 1.29435468e+00 -1.24572051e+00 1.13071287e+00 1.06776345e+00 7.44556606e-01 6.67501450e-01 -1.09138489e+00 -6.06471002e-01 8.20457786e-02 1.57568678e-01 -1.45013511e+00 -4.96213049e-01 3.62703949e-01 -4.15209591e-01 1.49552703e+00 -5.69871724e-01 1.98510975e-01 1.29284430e+00 1.52865872e-01 7.48531759e-01 9.19749141e-01 -8.96057785e-01 1.21665403e-01 2.73246497e-01 2.27799341e-01 7.40156353e-01 2.74672478e-01 -1.79879069e-01 -5.28390765e-01 -1.22850016e-02 4.34117347e-01 -3.87364566e-01 -5.83983101e-02 -2.95277774e-01 -1.38508344e+00 1.09458566e+00 9.93443429e-02 9.49940264e-01 -3.52415800e-01 -4.68349010e-02 6.17443919e-01 2.20336422e-01 9.29658294e-01 5.99424243e-01 -9.49271083e-01 -1.73699051e-01 -6.39720917e-01 -2.38271073e-01 8.96338105e-01 9.79442418e-01 8.71176779e-01 8.65691900e-02 -3.29525173e-01 1.07830703e+00 3.20165902e-02 6.16277218e-01 5.89914441e-01 -2.51629263e-01 6.30314887e-01 -3.39156203e-02 6.21865951e-02 -2.75625408e-01 -1.34003848e-01 6.07285351e-02 -3.65922481e-01 -2.67511487e-01 3.22861880e-01 -6.34402215e-01 -7.42124915e-01 1.98498559e+00 1.87515184e-01 2.45158345e-01 5.00530660e-01 4.99563962e-01 5.14284849e-01 8.08350921e-01 7.80880630e-01 9.73054692e-02 1.72919881e+00 -1.15444446e+00 -9.28862810e-01 -4.37570274e-01 8.68937373e-01 -8.67905617e-01 1.15835679e+00 -2.26523399e-01 -7.37173259e-01 -5.41390300e-01 -8.23235512e-01 -5.85113227e-01 -9.54306662e-01 7.57133737e-02 6.60086393e-01 5.84906280e-01 -1.02520061e+00 2.89087087e-01 -9.14171040e-01 -7.32307553e-01 1.54089049e-01 -1.37970671e-01 -7.95897901e-01 -4.60494161e-01 -1.76693487e+00 1.43641627e+00 3.83532405e-01 -4.70765322e-01 -7.78635681e-01 -1.11969638e+00 -1.50577426e+00 6.77981530e-04 5.45863658e-02 -3.14581752e-01 1.23143685e+00 -4.07194287e-01 -1.27558815e+00 1.02061820e+00 -2.61482388e-01 -3.95310372e-01 -2.28582993e-02 -6.82264268e-01 -5.15113831e-01 -1.59690902e-01 5.17518818e-01 8.28712523e-01 9.35960263e-02 -6.94163382e-01 -5.60860753e-01 -1.62506521e-01 -1.84865445e-01 2.17266664e-01 -6.93098664e-01 6.11629307e-01 -3.13125253e-01 -6.10606790e-01 -9.48732495e-01 -6.01275921e-01 -4.19371396e-01 -2.99411714e-01 1.21140935e-01 -6.26425803e-01 2.92452842e-01 -1.11530018e+00 1.05893826e+00 -2.15962720e+00 -3.10969409e-02 -5.26062191e-01 -5.04171133e-01 6.73802078e-01 -5.54496169e-01 7.95493543e-01 -2.13844309e-04 2.65074521e-01 -1.64950997e-01 -5.67023873e-01 1.01745166e-01 4.37593400e-01 -2.41114438e-01 3.98414701e-01 5.87500274e-01 1.09895897e+00 -1.35743701e+00 -6.03218079e-01 2.75542498e-01 8.80427241e-01 -3.88986260e-01 2.93628156e-01 1.77560985e-01 4.28197235e-01 -1.22802377e-01 4.17774320e-01 4.26283777e-01 3.19424391e-01 4.09056097e-01 2.50133406e-02 -5.23114383e-01 5.37187874e-01 -7.24353790e-01 2.17522311e+00 -1.05233562e+00 5.28689802e-01 -1.23368233e-01 -7.99582124e-01 7.43911028e-01 8.84527564e-01 2.31998608e-01 -6.18302763e-01 -7.73115316e-03 2.86717027e-01 -1.71148300e-01 -4.44573879e-01 6.51376665e-01 -4.34267819e-01 -6.00516915e-01 3.79141241e-01 1.01902711e+00 1.19012408e-01 3.07818323e-01 -4.76429835e-02 9.70265269e-01 2.79376924e-01 8.99090946e-01 -4.67881650e-01 2.85774261e-01 -7.57898251e-03 6.68610990e-01 4.90927666e-01 -5.79205930e-01 1.46010965e-01 1.14422506e-02 -2.88229138e-01 -1.31747878e+00 -1.18313658e+00 -2.80203283e-01 1.64575088e+00 -3.91529053e-01 -3.47141251e-02 -6.21863484e-01 -8.44290614e-01 -1.55935347e-01 9.93285418e-01 -4.94774044e-01 2.59349167e-01 -6.55919969e-01 -6.08610034e-01 9.27437365e-01 6.11650288e-01 1.34012237e-01 -1.25321567e+00 -3.69636640e-02 3.90580893e-01 -1.40167266e-01 -1.60381520e+00 -7.04298556e-01 4.40647990e-01 -3.15025210e-01 -7.96621263e-01 -1.07022500e+00 -1.32732522e+00 3.19146693e-01 1.09472610e-01 1.21742022e+00 -6.65310979e-01 -1.52921483e-01 3.18239808e-01 -5.38238823e-01 -4.85287786e-01 -3.39749515e-01 5.01514852e-01 2.65038580e-01 -2.37575486e-01 8.23258817e-01 -3.02162439e-01 4.02708873e-02 -3.27109933e-01 -7.82945275e-01 -4.80347693e-01 5.46775401e-01 7.96169043e-01 4.61023122e-01 -6.45437062e-01 7.77833164e-01 -1.03804982e+00 7.83857882e-01 -6.09786987e-01 -5.54685831e-01 7.01408565e-01 -2.55918026e-01 2.63577998e-01 6.53444409e-01 -2.96294391e-01 -1.34861279e+00 -2.22375616e-02 -3.74621391e-01 -7.01157302e-02 -4.49640185e-01 4.94570464e-01 -2.07512364e-01 5.01217805e-02 4.81622964e-01 -3.21226195e-03 -5.74522972e-01 -7.47722328e-01 1.01164389e+00 7.64579833e-01 4.35080469e-01 -6.69573724e-01 5.37959218e-01 1.32328585e-01 -6.64353609e-01 -1.02670848e+00 -1.04203856e+00 -7.81139553e-01 -1.15728152e+00 2.08106816e-01 1.44654894e+00 -1.43387640e+00 1.43590197e-01 6.69516101e-02 -1.47084761e+00 -4.29202229e-01 -3.91231835e-01 8.85542095e-01 -2.11637318e-01 2.63076812e-01 -1.02254331e+00 -5.98631680e-01 -5.14357746e-01 -8.69272888e-01 1.09219265e+00 1.08599164e-01 -1.65397406e-01 -1.64655018e+00 8.21979344e-01 6.30994886e-02 3.22866321e-01 -2.35960066e-01 8.83274972e-01 -1.10268092e+00 2.72093043e-02 -1.21209063e-02 -2.01113805e-01 4.89808708e-01 1.13800570e-01 -3.65032345e-01 -8.86881113e-01 -3.50394845e-01 -5.48517346e-01 -5.62007189e-01 6.38524771e-01 7.93374702e-02 9.94499549e-02 -7.68021643e-02 -2.01212004e-01 3.22341025e-01 1.61334562e+00 -9.78333429e-02 3.91510010e-01 2.77506292e-01 6.21189117e-01 6.42172754e-01 5.30628800e-01 3.26970071e-02 6.32879615e-01 4.35812652e-01 -3.40053022e-01 -2.66360939e-01 -3.81414324e-01 -4.80342716e-01 5.04184783e-01 1.56861091e+00 2.45415196e-01 -1.03460737e-01 -1.20235312e+00 1.35594690e+00 -1.61087072e+00 -7.04902887e-01 2.93418825e-01 1.97110927e+00 1.23930645e+00 -4.35834050e-01 -2.94805259e-01 -7.98855126e-01 9.58686948e-01 3.06881756e-01 1.66139696e-02 -7.46268392e-01 -1.63707301e-01 9.31323588e-01 7.72737086e-01 7.29000390e-01 -1.34321356e+00 1.77664316e+00 6.35441065e+00 7.79423594e-01 -8.22166026e-01 8.95679832e-01 1.01504967e-01 4.83082443e-01 -1.49486572e-01 5.90826906e-02 -1.14530218e+00 5.11805676e-02 1.20365083e+00 -3.82629156e-01 1.38316825e-01 1.02648389e+00 -2.58146048e-01 2.56567478e-01 -9.61519659e-01 4.63771671e-01 3.39925915e-01 -1.01640213e+00 -1.09509014e-01 -1.06883794e-01 1.12685883e+00 7.39952445e-01 -2.86794484e-01 8.60064983e-01 9.65817750e-01 -7.37480640e-01 2.47587681e-01 1.48403704e-01 9.32478726e-01 -8.44398260e-01 9.15849268e-01 2.45403588e-01 -1.29737008e+00 3.94072473e-01 -5.20246267e-01 8.15359429e-02 6.74995542e-01 5.26287854e-01 -8.09230387e-01 7.20252335e-01 4.17723209e-01 6.29740000e-01 -4.17693526e-01 8.59962404e-01 -7.23157465e-01 6.40619099e-01 4.52063456e-02 -1.75150618e-01 4.36069429e-01 8.53929743e-02 3.19037199e-01 1.90807605e+00 3.42703372e-01 -3.65164518e-01 4.18473870e-01 3.92319918e-01 -4.23770547e-01 6.67056262e-01 -8.68674338e-01 -3.74455154e-01 4.82899249e-01 1.20610297e+00 -2.30231494e-01 -4.94527400e-01 -9.59511757e-01 1.08483386e+00 1.00831735e+00 2.83654243e-01 -4.83831495e-01 -7.11206675e-01 9.03001606e-01 -4.81833339e-01 5.33010781e-01 -6.83527648e-01 5.09366058e-02 -1.52814662e+00 -3.21379513e-01 -4.17861104e-01 4.36625481e-01 -4.84516084e-01 -1.82787347e+00 6.16963625e-01 -2.23103866e-01 -8.21738482e-01 -1.80231199e-01 -7.87123442e-01 -4.44689929e-01 1.05382633e+00 -2.03982091e+00 -1.35022235e+00 4.74026084e-01 4.51055735e-01 5.78168809e-01 -1.91342711e-01 1.41397679e+00 6.81866944e-01 -3.47661883e-01 5.49140811e-01 2.00622320e-01 7.64024913e-01 1.22136772e+00 -1.24626958e+00 7.32469857e-01 1.13544667e+00 5.92630923e-01 6.30765676e-01 3.21882755e-01 -7.02231765e-01 -1.06070459e+00 -1.35134268e+00 1.94815397e+00 -4.84501600e-01 1.22722673e+00 -6.37928843e-01 -9.89206970e-01 9.76763070e-01 8.87310505e-01 2.27489904e-01 1.17501962e+00 5.74951351e-01 -7.04718471e-01 2.16891438e-01 -9.82443988e-01 5.47373533e-01 6.96743488e-01 -9.37676251e-01 -1.09681356e+00 4.08360511e-01 1.14926386e+00 6.15623705e-02 -9.90365326e-01 1.26365587e-01 2.16336921e-01 -9.95393246e-02 6.66938365e-01 -1.02984250e+00 2.00085074e-01 5.31515367e-02 -2.60115385e-01 -1.73727989e+00 -3.09919089e-01 -3.24707508e-01 4.10286725e-01 1.61170983e+00 5.80696642e-01 -5.56738377e-01 -1.71397242e-03 1.57553926e-01 -3.04914117e-01 -2.53129810e-01 -1.22432911e+00 -1.02770722e+00 6.73753798e-01 -4.63983148e-01 1.51798144e-01 1.55386007e+00 3.25913042e-01 7.42119789e-01 -5.38009226e-01 1.44534305e-01 4.66356039e-01 -4.77022588e-01 4.26292002e-01 -9.33729410e-01 -5.09084538e-02 1.15055665e-02 -2.65886545e-01 -6.13373458e-01 7.76053965e-01 -1.24043751e+00 4.57194477e-01 -1.72919536e+00 1.54815421e-01 -1.59561023e-01 -6.28718078e-01 8.03362548e-01 -4.03870136e-01 1.97763249e-01 1.52350008e-01 -2.40488991e-01 -7.54235387e-01 5.43324113e-01 7.95843899e-01 6.81232363e-02 1.39784664e-01 -7.84593523e-01 -4.36324835e-01 4.65988219e-01 6.12057328e-01 -6.82898164e-01 1.77373588e-01 -8.91333997e-01 4.22599129e-02 -1.42209068e-01 -1.74149096e-01 -8.17366898e-01 6.38271719e-02 -1.80943441e-02 3.24201249e-02 -8.66455287e-02 2.22882070e-03 -4.61117864e-01 -7.44084239e-01 3.20218295e-01 -3.18044573e-01 2.35560358e-01 2.99994230e-01 3.77001315e-01 -4.44658786e-01 -3.81151497e-01 7.06093431e-01 -2.65266061e-01 -1.15972495e+00 4.24120069e-01 -4.43740666e-01 6.10175967e-01 8.63048732e-01 6.85700119e-01 -3.15421671e-01 2.71337684e-02 -5.02098203e-01 2.15927497e-01 2.55969852e-01 8.44257593e-01 2.85096075e-02 -1.39685321e+00 -8.70077252e-01 -5.05358949e-02 5.03175974e-01 -5.61651707e-01 1.62356541e-01 5.05148292e-01 -3.30332130e-01 7.70268857e-01 -2.57545024e-01 4.72256392e-02 -9.11599278e-01 5.48151195e-01 -1.00563422e-01 -7.51310587e-01 -2.52776742e-01 8.29477191e-01 -6.06784001e-02 -1.11853135e+00 -5.41221611e-02 1.47125781e-01 -2.35452756e-01 1.77720889e-01 6.26031280e-01 8.60421546e-03 4.55604941e-02 -8.85806024e-01 -5.87971747e-01 3.82787675e-01 -1.04080848e-01 -5.05089879e-01 1.54077995e+00 -3.52735296e-02 2.72172466e-02 7.33894527e-01 1.40789652e+00 3.27315956e-01 -8.49392772e-01 -5.89505911e-01 3.85684520e-01 -4.30300459e-02 -2.21132100e-01 -5.54857016e-01 -4.24237013e-01 1.28914821e+00 3.90011609e-01 -3.55133712e-01 3.94733042e-01 2.37653375e-01 1.11467636e+00 5.47236323e-01 5.65683365e-01 -1.39738607e+00 -2.70826250e-01 1.14822412e+00 2.65625477e-01 -1.31316400e+00 -3.40161502e-01 -1.60697132e-01 -8.13076854e-01 8.21836352e-01 4.01172906e-01 -2.36633420e-01 6.98581159e-01 4.01714861e-01 4.10307825e-01 2.14904830e-01 -6.27645433e-01 -7.09026992e-01 2.18764797e-01 9.30367768e-01 1.03040004e+00 1.94907680e-01 -5.98188758e-01 6.37776792e-01 1.27477441e-02 -3.36869396e-02 1.61154151e-01 9.62277114e-01 -3.86213899e-01 -1.49943936e+00 1.87908933e-01 -3.20769101e-01 -8.47249508e-01 -8.52851808e-01 -1.00202672e-02 8.65095317e-01 2.43445039e-01 7.35318244e-01 1.39206471e-02 1.94747895e-01 2.24528462e-01 6.82127833e-01 4.15626347e-01 -1.13857853e+00 -5.85221350e-01 -1.48450449e-01 3.64169687e-01 -1.86703071e-01 -4.34944153e-01 -6.21238470e-01 -1.01816893e+00 2.45812222e-01 -4.17795777e-02 3.67466748e-01 9.34073329e-01 1.03708529e+00 4.46646661e-01 5.15482187e-01 1.05829895e-01 -5.36792338e-01 -3.07210267e-01 -1.19318497e+00 -3.86898786e-01 4.74322706e-01 -7.53757134e-02 -3.98298353e-01 -2.20026374e-02 1.98802069e-01]
[10.660219192504883, 9.789510726928711]
9e6eda32-349e-4b70-ab34-c3718e799321
daformer-improving-network-architectures-and
2111.14887
null
https://arxiv.org/abs/2111.14887v2
https://arxiv.org/pdf/2111.14887v2.pdf
DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation
As acquiring pixel-wise annotations of real-world images for semantic segmentation is a costly process, a model can instead be trained with more accessible synthetic data and adapted to real images without requiring their annotations. This process is studied in unsupervised domain adaptation (UDA). Even though a large number of methods propose new adaptation strategies, they are mostly based on outdated network architectures. As the influence of recent network architectures has not been systematically studied, we first benchmark different network architectures for UDA and newly reveal the potential of Transformers for UDA semantic segmentation. Based on the findings, we propose a novel UDA method, DAFormer. The network architecture of DAFormer consists of a Transformer encoder and a multi-level context-aware feature fusion decoder. It is enabled by three simple but crucial training strategies to stabilize the training and to avoid overfitting to the source domain: While (1) Rare Class Sampling on the source domain improves the quality of the pseudo-labels by mitigating the confirmation bias of self-training toward common classes, (2) a Thing-Class ImageNet Feature Distance and (3) a learning rate warmup promote feature transfer from ImageNet pretraining. DAFormer represents a major advance in UDA. It improves the state of the art by 10.8 mIoU for GTA-to-Cityscapes and 5.4 mIoU for Synthia-to-Cityscapes and enables learning even difficult classes such as train, bus, and truck well. The implementation is available at https://github.com/lhoyer/DAFormer.
['Luc van Gool', 'Dengxin Dai', 'Lukas Hoyer']
2021-11-29
null
http://openaccess.thecvf.com//content/CVPR2022/html/Hoyer_DAFormer_Improving_Network_Architectures_and_Training_Strategies_for_Domain-Adaptive_Semantic_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Hoyer_DAFormer_Improving_Network_Architectures_and_Training_Strategies_for_Domain-Adaptive_Semantic_CVPR_2022_paper.pdf
cvpr-2022-1
['synthetic-to-real-translation']
['computer-vision']
[ 4.00949657e-01 4.20444608e-01 -4.59288396e-02 -5.59685826e-01 -7.20063448e-01 -5.46421707e-01 7.36174583e-01 -1.24807425e-01 -6.41796887e-01 6.77727163e-01 -2.29922652e-01 -2.87886471e-01 8.18341896e-02 -8.79376054e-01 -1.00063550e+00 -6.41828120e-01 3.46780539e-01 8.14810395e-01 5.94371319e-01 -2.39367008e-01 -2.54240513e-01 3.51361781e-01 -1.48511624e+00 2.84539878e-01 1.02861452e+00 1.00499523e+00 4.49310690e-01 3.65689009e-01 -6.42622858e-02 5.67289948e-01 -5.79205275e-01 -4.75296617e-01 4.60585743e-01 -4.65739578e-01 -1.14170420e+00 2.96457589e-01 4.07017320e-01 -1.67831749e-01 3.29652405e-03 9.36910808e-01 4.79040116e-01 7.79146999e-02 6.14300728e-01 -1.23006618e+00 -4.74709183e-01 6.34099364e-01 -2.20214516e-01 -6.23823926e-02 -6.38554543e-02 4.14088458e-01 7.55669057e-01 -8.47348452e-01 7.51551270e-01 1.02327490e+00 8.65219235e-01 7.13590264e-01 -1.21107078e+00 -6.51656151e-01 1.84045911e-01 2.12483868e-01 -1.39820719e+00 -3.90585333e-01 5.54631293e-01 -4.09605533e-01 8.19651723e-01 -8.42169765e-03 6.93588674e-01 1.30526924e+00 -3.19445223e-01 8.42994034e-01 1.25230384e+00 -4.79788363e-01 3.51301432e-01 3.72783124e-01 -1.04923047e-01 6.13023221e-01 1.45076782e-01 -1.80543344e-02 -2.02788979e-01 3.65058571e-01 6.14585936e-01 -2.68910080e-01 2.81526819e-02 -5.04642129e-01 -1.00854754e+00 8.22301090e-01 6.12734616e-01 3.40707093e-01 -2.43251398e-01 -3.08228061e-02 3.55904549e-01 3.05821210e-01 4.06948596e-01 5.75711846e-01 -6.74549460e-01 9.52649117e-02 -1.01078904e+00 5.85650764e-02 5.37034512e-01 1.03213573e+00 1.01082444e+00 2.05022693e-01 1.11311041e-02 9.84223723e-01 1.62336826e-01 4.44853008e-01 5.94805896e-01 -8.40191126e-01 2.77362615e-01 6.07427776e-01 -1.78030565e-01 -3.93635690e-01 -4.15622681e-01 -5.04933000e-01 -5.24382889e-01 3.02780718e-01 8.73600006e-01 -1.08917981e-01 -1.36962354e+00 1.68632293e+00 3.66029143e-01 6.98212683e-02 1.16473697e-01 8.14849198e-01 5.84150791e-01 4.32276368e-01 2.97181070e-01 3.61215562e-01 1.17669582e+00 -1.11329544e+00 -1.82388812e-01 -4.34224814e-01 7.61403203e-01 -6.74483895e-01 1.33591497e+00 3.07196409e-01 -1.03547227e+00 -8.25155199e-01 -1.06287575e+00 -3.53298821e-02 -8.06286752e-01 1.78012505e-01 5.19784510e-01 6.64533138e-01 -1.10763013e+00 5.29950678e-01 -8.29494298e-01 -7.67177641e-01 8.18231344e-01 3.70442420e-01 -3.10647815e-01 -9.14350152e-02 -1.16460526e+00 8.64701033e-01 7.70091236e-01 -1.05619662e-01 -1.02709329e+00 -6.09959066e-01 -9.17822957e-01 -1.39731809e-01 4.20572639e-01 -7.59898067e-01 1.36018324e+00 -1.58415735e+00 -1.71060336e+00 1.14154661e+00 1.93699315e-01 -7.57320881e-01 7.32133865e-01 -5.04518561e-02 -3.76126885e-01 7.28871971e-02 2.99965709e-01 1.19748926e+00 8.67353261e-01 -1.23183584e+00 -7.89955437e-01 -7.20348209e-02 -2.48370469e-02 1.10010132e-01 -1.63999155e-01 -3.67303938e-01 -3.54476392e-01 -6.12786829e-01 -9.90308002e-02 -8.76094699e-01 -2.32563093e-01 -1.85212448e-01 -3.08594495e-01 -1.11716919e-01 7.93160617e-01 -3.61641973e-01 6.91767812e-01 -2.10073447e+00 -9.88072604e-02 2.60702819e-01 -5.99636883e-02 5.19619286e-01 -2.66182095e-01 1.12208471e-01 -1.66220814e-01 -9.32490826e-02 -6.22585595e-01 -2.04321951e-01 8.00256059e-02 3.28613698e-01 -8.43971446e-02 3.21019262e-01 3.88020694e-01 9.76614952e-01 -1.02056050e+00 -3.24086934e-01 3.63441855e-01 3.24447125e-01 -5.73065877e-01 4.76717278e-02 -4.50511396e-01 7.30292618e-01 -2.77792037e-01 6.04230344e-01 6.89624071e-01 -3.24751496e-01 9.58044082e-02 -1.65119469e-01 -5.91889136e-02 3.60876769e-01 -1.02146649e+00 1.94070327e+00 -5.22777140e-01 5.16470730e-01 -2.10773826e-01 -1.27634346e+00 9.13769484e-01 6.76115230e-02 3.28209698e-01 -9.90368187e-01 2.19208702e-01 5.31605363e-01 5.91393523e-02 -3.82719040e-01 4.17267233e-01 -7.23650604e-02 -7.39859641e-02 1.08101010e-01 5.29011965e-01 -3.42028260e-01 3.55998933e-01 1.01952245e-02 9.53757167e-01 4.75911081e-01 1.67534221e-02 -3.53962868e-01 4.56728369e-01 3.11420441e-01 6.42504454e-01 6.29963100e-01 -2.21592143e-01 7.66471624e-01 2.31748059e-01 -4.08184379e-01 -1.23218727e+00 -1.09365952e+00 -2.18031153e-01 1.04399049e+00 7.72867426e-02 -1.96307585e-01 -1.06138039e+00 -1.05983186e+00 -1.62498713e-01 1.01472044e+00 -5.86429238e-01 -2.99193859e-01 -4.32635933e-01 -6.83545947e-01 5.66975653e-01 6.02555215e-01 1.01732135e+00 -1.02583992e+00 -5.89791119e-01 1.59859940e-01 -9.40419883e-02 -1.35492504e+00 -2.99450248e-01 4.03285444e-01 -7.47260988e-01 -9.99916494e-01 -8.59132051e-01 -8.64407122e-01 7.13796198e-01 -1.00614026e-01 1.25606322e+00 -2.92602032e-01 -3.73419151e-02 2.68784165e-01 -3.39045256e-01 -4.69008803e-01 -6.19128585e-01 4.01644826e-01 -1.41862452e-01 9.27248523e-02 4.32993114e-01 -4.92871284e-01 -6.69068515e-01 5.30389249e-01 -8.54345322e-01 2.63175279e-01 6.01096690e-01 7.74846435e-01 7.40518212e-01 -1.08194076e-01 7.46881127e-01 -1.29791498e+00 1.22551151e-01 -4.40739483e-01 -7.43849635e-01 5.40832281e-02 -7.70732284e-01 -7.95819759e-02 6.03237867e-01 -4.68296528e-01 -1.16974533e+00 4.69079494e-01 -4.14787829e-01 -4.15078580e-01 -5.44426143e-01 1.43884659e-01 -2.46695474e-01 6.91436529e-02 9.28310752e-01 1.23274006e-01 -2.22139731e-01 -3.54117543e-01 5.35660565e-01 5.66675067e-01 5.51561654e-01 -6.25444233e-01 7.77779937e-01 5.75291693e-01 -3.36365789e-01 -7.12588131e-01 -8.27641487e-01 -3.50483924e-01 -7.68317103e-01 -2.16670468e-01 1.01590335e+00 -9.85565186e-01 -5.16500659e-02 6.44504070e-01 -7.70856261e-01 -9.79893982e-01 -7.64421105e-01 3.79052550e-01 -6.06950641e-01 -7.67671317e-02 -3.17102432e-01 -2.72850811e-01 -3.46634127e-02 -1.29397607e+00 8.71272087e-01 3.79752725e-01 -2.85195827e-01 -9.85089064e-01 -8.85905027e-02 4.17426139e-01 4.14201736e-01 1.76367536e-01 6.48646176e-01 -8.44769895e-01 -5.43289244e-01 -2.79155467e-02 -2.40551293e-01 7.01590002e-01 7.26940259e-02 -1.82470754e-01 -1.26562178e+00 -1.48546338e-01 -3.16243440e-01 -4.43493545e-01 8.20469379e-01 3.68487120e-01 1.12328649e+00 -4.01752591e-02 -2.38486469e-01 6.83435619e-01 1.20291758e+00 2.82390386e-01 6.33589089e-01 6.17674470e-01 6.69964969e-01 4.66376394e-01 4.96884465e-01 -6.37545483e-03 5.86878419e-01 5.70265055e-01 4.36116308e-01 -3.29659432e-01 -6.55530274e-01 -3.67008448e-01 2.65930653e-01 7.58370936e-01 7.30846869e-03 -1.04482494e-01 -1.03407371e+00 9.23904300e-01 -1.63771212e+00 -5.18508315e-01 -1.70912534e-01 2.20346212e+00 8.10247481e-01 4.31358606e-01 3.35651726e-01 7.30194896e-02 5.27173936e-01 -1.52533606e-01 -6.27382636e-01 -3.49800915e-01 -2.69477248e-01 4.31117356e-01 7.15973675e-01 3.24785709e-01 -1.13478351e+00 1.28839862e+00 5.47213554e+00 1.09802294e+00 -1.20334184e+00 4.32978868e-01 7.69290924e-01 2.25556254e-01 -1.15803152e-01 4.18594070e-02 -6.22983873e-01 5.10689318e-01 9.55425501e-01 2.63383687e-01 2.47815281e-01 1.01963556e+00 -8.73508155e-02 -1.46222368e-01 -9.67778921e-01 7.71846175e-01 -8.70653689e-02 -1.27515304e+00 1.58984661e-02 -1.33568689e-01 9.33659256e-01 3.64067644e-01 2.57083215e-02 5.02033949e-01 4.54379529e-01 -8.16300631e-01 9.70991671e-01 2.56805807e-01 8.48316789e-01 -6.21610701e-01 6.25965953e-01 1.19272016e-01 -8.69813144e-01 -3.25206257e-02 -2.70885587e-01 1.83711544e-01 1.82357114e-02 5.35831928e-01 -9.63908434e-01 5.96298814e-01 8.28273475e-01 6.80296540e-01 -7.30108023e-01 9.06696081e-01 -4.33568925e-01 8.96699786e-01 -4.52421963e-01 3.88708919e-01 6.06883168e-01 -1.53391048e-01 3.11433882e-01 1.19539011e+00 2.58674860e-01 -2.54925519e-01 5.67312390e-02 8.65056396e-01 -1.95339814e-01 -9.11888853e-02 -5.64353824e-01 1.79923579e-01 2.83787727e-01 1.20206869e+00 -1.02881742e+00 -3.74176085e-01 -4.21521187e-01 1.13118505e+00 2.08161891e-01 4.59680408e-01 -9.69053805e-01 -2.93624312e-01 3.43721211e-01 3.28939885e-01 4.45321172e-01 -2.66647991e-03 -3.68945181e-01 -9.61084366e-01 -1.39124900e-01 -8.49996388e-01 4.59766626e-01 -8.28607321e-01 -1.06763327e+00 7.99854457e-01 6.39171451e-02 -1.24464071e+00 -2.07174644e-01 -6.54456079e-01 -5.05556941e-01 6.42998517e-01 -1.60935152e+00 -1.23266971e+00 -3.08892757e-01 7.48863220e-01 8.18762183e-01 -2.64297426e-01 7.02922761e-01 5.39942920e-01 -5.58454573e-01 6.45026743e-01 -1.65797733e-02 1.13604471e-01 6.75535738e-01 -1.29472399e+00 5.38084567e-01 8.53994370e-01 2.00042531e-01 -2.35501882e-02 4.90939349e-01 -3.54537696e-01 -6.22938156e-01 -1.32883394e+00 7.30606318e-01 -5.21981597e-01 5.72097182e-01 -5.60074449e-01 -9.98181522e-01 7.52736628e-01 2.38914654e-01 7.38339946e-02 2.11423427e-01 -4.90746908e-02 -3.42582643e-01 -3.34716111e-01 -1.20848346e+00 4.87687469e-01 9.92423594e-01 -4.31103081e-01 -3.72233212e-01 3.67196798e-01 4.72436696e-01 -4.69781637e-01 -7.16425180e-01 4.76813167e-01 2.09248960e-01 -9.76109028e-01 8.76542687e-01 -2.67062306e-01 2.89439768e-01 -3.89517635e-01 -1.55370282e-02 -1.33550370e+00 -9.38319266e-02 -4.57215101e-01 3.12365294e-01 1.26480854e+00 6.39573216e-01 -8.43010485e-01 8.04966569e-01 3.80777538e-01 -4.16278422e-01 -5.79979300e-01 -9.64569211e-01 -1.00757134e+00 2.45500728e-01 -4.81693923e-01 6.63944185e-01 1.15327966e+00 -5.25472760e-01 4.11880970e-01 5.79776168e-02 9.77875218e-02 4.03488249e-01 -1.74339101e-01 7.78985620e-01 -1.14113152e+00 -2.00160131e-01 -5.08647680e-01 -3.13979208e-01 -8.22554350e-01 7.02379122e-02 -1.16064084e+00 2.37369779e-02 -1.48017752e+00 -1.19733177e-01 -6.66061461e-01 -2.07151309e-01 7.20600903e-01 1.26602113e-01 4.14575279e-01 6.92890808e-02 1.61384463e-01 -5.82903743e-01 4.39542621e-01 1.24146819e+00 -1.86980024e-01 -1.08402833e-01 -1.63731687e-02 -3.54869455e-01 9.07649159e-01 1.10377097e+00 -6.11519992e-01 -6.30780399e-01 -5.42277515e-01 7.07898214e-02 -5.67187548e-01 6.00273132e-01 -1.25075960e+00 -1.06306829e-01 1.27838001e-01 3.94385844e-01 -2.40608141e-01 1.45285815e-01 -8.84056926e-01 1.58091411e-01 3.41261029e-01 -9.34304148e-02 -2.19606593e-01 4.15599763e-01 2.66773671e-01 -1.70909807e-01 -3.20816457e-01 1.09784591e+00 -2.44810164e-01 -1.10168433e+00 1.12530440e-01 -3.36057663e-01 2.81542569e-01 9.60217118e-01 -5.88889420e-01 -1.87636182e-01 -2.95188632e-02 -8.08638573e-01 1.45056814e-01 5.61639071e-01 3.27933162e-01 2.24193230e-01 -1.09395957e+00 -5.83225191e-01 3.69293928e-01 1.66469291e-01 3.79407585e-01 2.04030782e-01 8.10993373e-01 -5.69027245e-01 1.80211052e-01 -3.59218955e-01 -7.97851026e-01 -7.33542681e-01 3.37372214e-01 5.28319776e-01 -2.21720934e-01 -6.23813272e-01 9.58200455e-01 3.28495443e-01 -6.80568278e-01 2.39715949e-02 -3.60187948e-01 5.23328520e-02 1.24789076e-02 6.81979060e-02 1.48586705e-01 3.70955735e-01 -6.42789841e-01 -2.95112669e-01 3.70803028e-01 -1.34021625e-01 2.16540117e-02 1.23965859e+00 -1.85713887e-01 3.40166956e-01 4.03182477e-01 9.00902629e-01 -2.89281219e-01 -1.69831574e+00 -1.75508767e-01 1.02616712e-01 -1.72146529e-01 -4.93286438e-02 -1.14671504e+00 -1.23597360e+00 7.94476688e-01 8.18220615e-01 5.58371246e-02 1.28425622e+00 1.20579243e-01 7.47740924e-01 1.62351310e-01 3.22008550e-01 -1.46059155e+00 6.98802397e-02 5.31714201e-01 6.37087107e-01 -1.35285008e+00 -3.55897307e-01 -2.28800222e-01 -7.74353266e-01 7.46427774e-01 6.66772664e-01 -1.09942377e-01 4.87195671e-01 5.50316088e-02 3.27941954e-01 -1.30066991e-01 -3.27450305e-01 -6.48408949e-01 1.65583298e-01 8.08952749e-01 1.73469305e-01 -1.76140182e-02 3.83884348e-02 6.36426568e-01 -2.13437095e-01 8.65772739e-02 2.65700579e-01 7.07448363e-01 -1.12469241e-01 -1.12645090e+00 -1.26944497e-01 2.40544170e-01 -2.87838250e-01 -2.19172258e-02 -1.71591729e-01 1.15521502e+00 5.02329469e-01 6.68665171e-01 7.58106932e-02 -2.65886217e-01 6.67762399e-01 2.76465923e-01 4.29422170e-01 -5.62348366e-01 -5.96767128e-01 -2.39356626e-02 1.53243825e-01 -4.90069926e-01 -6.21012092e-01 -7.59302974e-01 -1.25513887e+00 4.44800267e-03 -2.62583375e-01 -2.07099598e-02 6.56241059e-01 9.65866208e-01 3.47453415e-01 7.35385060e-01 3.24629366e-01 -7.35765934e-01 -1.28453732e-01 -9.04557467e-01 -4.08018202e-01 4.46030349e-01 -2.63690483e-02 -7.33420014e-01 -7.20827729e-02 3.22905183e-01]
[9.747227668762207, 1.305688500404358]
3c581156-0267-4c4a-96ff-d456933516c2
searching-for-effective-neural-extractive
1907.03491
null
https://arxiv.org/abs/1907.03491v1
https://arxiv.org/pdf/1907.03491v1.pdf
Searching for Effective Neural Extractive Summarization: What Works and What's Next
The recent years have seen remarkable success in the use of deep neural networks on text summarization. However, there is no clear understanding of \textit{why} they perform so well, or \textit{how} they might be improved. In this paper, we seek to better understand how neural extractive summarization systems could benefit from different types of model architectures, transferable knowledge and learning schemas. Additionally, we find an effective way to improve current frameworks and achieve the state-of-the-art result on CNN/DailyMail by a large margin based on our observations and analyses. Hopefully, our work could provide more clues for future research on extractive summarization.
['PengFei Liu', 'Xuanjing Huang', 'Xipeng Qiu', 'Danqing Wang', 'Ming Zhong']
2019-07-08
searching-for-effective-neural-extractive-1
https://aclanthology.org/P19-1100
https://aclanthology.org/P19-1100.pdf
acl-2019-7
['extractive-document-summarization']
['natural-language-processing']
[ 2.28673011e-01 4.06086087e-01 -3.15501750e-01 -3.52856219e-01 -6.19235873e-01 -5.06878674e-01 5.46940148e-01 4.65253025e-01 -4.27692145e-01 6.93275332e-01 1.00087404e+00 -4.98509437e-01 -1.97779685e-01 -8.12119961e-01 -7.23733187e-01 -1.80487409e-01 2.92935461e-01 3.70333105e-01 2.13389727e-03 -6.19748175e-01 5.86243629e-01 2.43162915e-01 -1.11744559e+00 5.17456949e-01 9.65894222e-01 5.78189671e-01 -6.94361404e-02 6.58577919e-01 -6.29946515e-02 1.03507555e+00 -7.34342098e-01 -8.13062906e-01 4.22003195e-02 -5.40223598e-01 -1.20366478e+00 -3.24091524e-01 8.44915032e-01 -6.34179592e-01 -5.10899603e-01 9.13864017e-01 4.90280598e-01 1.64670378e-01 6.61316872e-01 -6.53109550e-01 -1.15632248e+00 1.12156630e+00 -3.05519283e-01 6.53830171e-01 1.36795118e-01 3.45399708e-01 1.11165118e+00 -3.03873837e-01 6.83772802e-01 1.03251839e+00 6.24764562e-01 7.33342171e-01 -7.87102818e-01 -6.46661401e-01 1.61527932e-01 2.20455930e-01 -6.66121542e-01 -6.87119842e-01 6.03522480e-01 -4.76564355e-02 1.31103277e+00 3.39756370e-01 5.83146691e-01 1.17263770e+00 3.25089604e-01 1.08762729e+00 6.32606387e-01 -2.19926551e-01 -1.52408570e-01 -6.33440390e-02 4.69669133e-01 7.42916286e-01 7.95965910e-01 -3.44773293e-01 -7.57363141e-01 1.19038366e-01 4.75056499e-01 -1.77198395e-01 -3.20544749e-01 1.31167904e-01 -1.09128129e+00 9.11173284e-01 5.25072396e-01 4.38686281e-01 -2.96374559e-01 3.62159997e-01 7.20647871e-01 3.84358734e-01 6.36992931e-01 1.15841305e+00 -4.70700771e-01 -3.85413408e-01 -1.44538426e+00 3.89749229e-01 9.48261678e-01 8.05108130e-01 4.96150523e-01 3.48648578e-01 -1.95136264e-01 5.68083167e-01 -1.10471591e-01 2.45980889e-01 5.52221894e-01 -1.11298418e+00 8.36008787e-01 5.19990742e-01 -8.02158490e-02 -9.35351431e-01 -5.53172767e-01 -5.70202410e-01 -9.50566113e-01 -2.12483466e-01 3.71239573e-01 -4.47439849e-01 -7.01256037e-01 1.44757307e+00 -6.32617295e-01 -7.10073709e-02 1.02710925e-01 5.48762977e-01 1.05318427e+00 6.10734105e-01 2.41042282e-02 9.28609446e-02 1.06471717e+00 -9.32320297e-01 -6.65917337e-01 -6.53241813e-01 7.60818481e-01 -6.91619694e-01 8.76662254e-01 2.84294277e-01 -1.42919612e+00 -3.25309694e-01 -1.27661705e+00 -3.63321036e-01 -2.19103947e-01 1.64743781e-01 9.01682615e-01 6.76034093e-01 -1.21306562e+00 9.44752395e-01 -9.84793961e-01 -8.84975791e-01 6.53183222e-01 3.92538279e-01 -2.00893089e-01 2.70069391e-01 -1.09161901e+00 1.12923813e+00 6.36432528e-01 7.89443105e-02 -4.81134027e-01 -4.57122952e-01 -7.06602454e-01 3.93905371e-01 4.33974236e-01 -1.26935339e+00 1.51998198e+00 -9.54062998e-01 -1.39848232e+00 7.61550963e-01 -8.75856280e-02 -1.02862275e+00 1.84313804e-01 -4.14386243e-01 -3.90640721e-02 1.90080211e-01 -1.76903903e-01 5.96193671e-01 3.83562565e-01 -8.67316246e-01 -5.06028175e-01 -4.87484753e-01 1.53809622e-01 2.50105351e-01 -6.60784304e-01 6.99974597e-02 -8.60928148e-02 -7.44391501e-01 -2.18311638e-01 -8.33109498e-01 5.94111495e-02 -5.86470425e-01 -4.70244557e-01 -3.15332413e-01 4.91743356e-01 -1.03964448e+00 1.28854966e+00 -1.71294248e+00 1.92670196e-01 -3.75298798e-01 3.38282675e-01 6.54609263e-01 -1.34463012e-01 7.97841072e-01 2.11472675e-01 5.91794014e-01 -2.96808034e-02 -1.59556255e-01 -1.26652857e-02 -1.92915902e-01 -6.99549973e-01 1.45356387e-01 3.91815424e-01 1.27191460e+00 -8.51960421e-01 -2.56813526e-01 -9.57308635e-02 1.18150547e-01 -5.55234313e-01 -1.94317132e-01 -3.25467825e-01 4.15271036e-02 -7.37164438e-01 1.43653318e-01 2.05866680e-01 -1.37430698e-01 4.91079427e-02 -1.21839687e-01 -1.13945417e-01 7.15272963e-01 -5.00542581e-01 1.65876770e+00 -1.78950697e-01 1.36303699e+00 -1.27792776e-01 -1.20847547e+00 7.63440311e-01 1.63752913e-01 2.11184219e-01 -7.37178624e-01 2.88071841e-01 4.11677286e-02 2.42244914e-01 -5.36140442e-01 1.23079765e+00 -5.95956258e-02 1.09795369e-01 7.21296608e-01 1.49183989e-01 -1.26184821e-01 2.57974952e-01 3.36076260e-01 1.23312736e+00 -1.62789509e-01 1.52737096e-01 -2.01850697e-01 9.04261693e-02 1.73149675e-01 1.78256527e-01 1.06197548e+00 -2.21897587e-01 6.36985302e-01 6.83271646e-01 -3.30602378e-01 -1.24695086e+00 -7.36164629e-01 2.59569257e-01 1.22962999e+00 -1.85254753e-01 -5.74029922e-01 -1.02147281e+00 -6.36096001e-01 -2.75999844e-01 1.19788349e+00 -4.87064898e-01 -3.09654146e-01 -8.26322973e-01 -7.04338372e-01 1.08508563e+00 8.27552676e-01 7.20179975e-01 -1.40473199e+00 -5.95081449e-01 9.50980112e-02 -3.00515324e-01 -7.24596977e-01 -2.99397528e-01 5.56086898e-02 -1.26786792e+00 -6.54928744e-01 -6.47713304e-01 -7.28069603e-01 2.02072412e-01 5.29544413e-01 1.27543068e+00 1.59708977e-01 2.95561612e-01 2.75139362e-01 -3.94017398e-01 -8.37095559e-01 -5.45646310e-01 9.15309131e-01 -2.72750407e-01 -6.27715766e-01 4.11462337e-01 -5.64233601e-01 -7.17363834e-01 -2.55796045e-01 -1.08685720e+00 9.65868682e-02 7.77369678e-01 5.22542000e-01 -1.70502588e-01 -1.47814780e-01 1.04010165e+00 -1.12804067e+00 1.32233250e+00 -4.65297759e-01 1.81663945e-01 2.26357862e-01 -6.92964613e-01 3.12421352e-01 6.27610147e-01 5.86714149e-02 -1.22386038e+00 -7.15925395e-01 -3.24158698e-01 1.32274985e-01 -7.17853159e-02 8.23548198e-01 3.37785512e-01 5.33078432e-01 9.22774732e-01 2.07866922e-01 -2.74646372e-01 -2.84342647e-01 5.35210431e-01 8.36187065e-01 6.33172035e-01 -4.80639756e-01 4.65863585e-01 5.96356094e-01 -4.82821494e-01 -9.29456770e-01 -1.12116849e+00 -2.33328685e-01 -4.08235967e-01 1.69042304e-01 1.07061815e+00 -7.42286801e-01 -3.98848146e-01 3.70211631e-01 -1.26525378e+00 -4.05154914e-01 -2.33642772e-01 1.28678352e-01 -5.89020848e-01 6.16479158e-01 -7.58969069e-01 -5.26639760e-01 -1.07896459e+00 -8.26586068e-01 9.28048074e-01 5.90705216e-01 -7.16279149e-01 -1.04541969e+00 -1.16280854e-01 7.90988386e-01 6.68175638e-01 -1.45563051e-01 1.05540490e+00 -1.09104872e+00 -5.43032646e-01 -1.30119249e-01 -2.15755582e-01 6.00130439e-01 2.82300375e-02 1.72632694e-01 -7.34940886e-01 -2.38182724e-01 -4.13157865e-02 -3.60267729e-01 1.34717894e+00 5.37254095e-01 1.22229111e+00 -5.20894408e-01 -1.66494012e-01 2.77480990e-01 1.01207817e+00 -1.20396577e-01 8.40296805e-01 4.82836425e-01 5.49699306e-01 5.16800284e-01 2.06137031e-01 1.74073875e-01 5.76968074e-01 1.52454942e-01 2.67542034e-01 1.32722974e-01 -3.36732388e-01 -3.05522531e-01 4.09849554e-01 1.03768408e+00 -2.22060814e-01 -5.92276454e-01 -8.97546530e-01 6.35706902e-01 -2.04887938e+00 -1.36642325e+00 -4.75406200e-02 1.91583121e+00 8.51626337e-01 3.84910434e-01 4.21566814e-02 -2.43873045e-01 3.93947005e-01 5.41522980e-01 -6.29678547e-01 -8.82560313e-01 -2.39472389e-01 2.96961427e-01 4.41549867e-01 6.71429634e-02 -8.85437012e-01 1.05470753e+00 6.84614992e+00 5.18481791e-01 -1.17976499e+00 -2.11531878e-01 7.43740380e-01 -2.83396542e-01 -4.16450560e-01 -5.15389033e-02 -7.91780412e-01 1.96503505e-01 1.09427643e+00 -4.22518939e-01 4.20291483e-01 5.28394103e-01 1.72924429e-01 -9.35477689e-02 -1.22799623e+00 5.68550050e-01 5.27447522e-01 -1.76447630e+00 2.72732973e-01 -2.45965179e-02 9.70058143e-01 3.43782037e-01 5.76162077e-02 6.32205486e-01 4.44320798e-01 -1.13220263e+00 4.94520783e-01 7.14149833e-01 2.36383840e-01 -5.47271788e-01 8.37007105e-01 6.96481049e-01 -4.13135380e-01 2.85310224e-02 -4.34921384e-01 -2.47738153e-01 -5.09615950e-02 2.05706105e-01 -9.38701749e-01 6.71993494e-01 4.47349370e-01 7.97371209e-01 -8.65475476e-01 9.71194446e-01 -2.74922848e-01 9.55590367e-01 7.24751279e-02 -4.32456762e-01 2.81683177e-01 -1.02756973e-02 5.83513200e-01 1.39048445e+00 2.06528261e-01 3.54316225e-03 -1.76127732e-01 6.23391926e-01 -5.30536413e-01 1.14223927e-01 -7.12071478e-01 -4.30100530e-01 2.44951904e-01 9.32012618e-01 -7.70317078e-01 -4.77727741e-01 -3.92535329e-01 8.51838589e-01 5.60353935e-01 4.13067937e-01 -6.49795115e-01 -4.17116374e-01 3.99931818e-01 1.54563248e-01 2.43329734e-01 -2.99292982e-01 -6.92369938e-01 -1.37652314e+00 -7.49058425e-02 -9.49266136e-01 3.82662982e-01 -9.30511057e-01 -1.22205746e+00 3.48978996e-01 -7.11202621e-02 -4.84403461e-01 -2.78692544e-01 -4.52894539e-01 -9.48486567e-01 5.82247078e-01 -1.26778471e+00 -9.39525843e-01 -6.30127639e-02 9.18884482e-03 9.10356700e-01 -1.77898169e-01 7.16210723e-01 -1.40053287e-01 -6.34734213e-01 5.30683160e-01 3.02822948e-01 2.44439259e-01 8.66407394e-01 -1.16522801e+00 6.68550551e-01 7.58537352e-01 2.92138189e-01 7.53314793e-01 9.50978637e-01 -6.58459127e-01 -1.28693080e+00 -9.67457771e-01 7.97397792e-01 -8.59806240e-01 6.20135486e-01 7.48017207e-02 -8.50739479e-01 1.17967391e+00 1.06957817e+00 -1.01678526e+00 6.26649261e-01 4.97357935e-01 -1.65980980e-01 2.48624180e-02 -7.30359852e-01 8.36306751e-01 9.43126798e-01 -3.42216223e-01 -1.23565328e+00 2.95473099e-01 7.90680885e-01 -2.77982563e-01 -7.51235783e-01 2.50221282e-01 5.42714715e-01 -1.14848149e+00 6.43035293e-01 -8.55689883e-01 1.03018379e+00 2.26183563e-01 1.56151265e-01 -1.55191684e+00 -2.79859394e-01 -4.90080595e-01 -1.85900688e-01 1.44518483e+00 5.58361292e-01 -5.98483264e-01 9.40799415e-01 5.50547004e-01 -4.72525358e-01 -8.51291120e-01 -5.54741085e-01 -5.72856903e-01 7.41532683e-01 -3.04150105e-01 4.70699102e-01 8.11713278e-01 1.61902025e-01 7.77274489e-01 -2.11351469e-01 -3.36738825e-01 1.23837799e-01 -3.93443108e-02 8.47821593e-01 -1.09312320e+00 -1.63919367e-02 -9.76660609e-01 -9.74902213e-02 -1.14660287e+00 4.57007378e-01 -1.03898621e+00 -3.51983607e-01 -2.25557828e+00 5.12637556e-01 2.16917589e-01 -1.49291575e-01 5.62103689e-01 -3.01061958e-01 6.88472986e-02 5.07982731e-01 7.67934844e-02 -8.84257734e-01 5.94133973e-01 1.15979755e+00 -4.08003271e-01 -1.49018347e-01 -3.55835557e-02 -1.55248058e+00 6.47877395e-01 1.18095303e+00 -3.03311139e-01 -3.36945623e-01 -9.48676825e-01 5.24373353e-01 -7.97846913e-02 7.19007030e-02 -9.50494051e-01 3.43905687e-01 1.55827831e-02 3.30538899e-01 -7.12793708e-01 7.91540816e-02 -2.28729710e-01 -3.91136348e-01 3.27447981e-01 -6.81088030e-01 1.72105983e-01 4.07909155e-01 3.45020384e-01 -1.65625185e-01 -6.74656153e-01 3.12981546e-01 -3.49633485e-01 -4.22564268e-01 -6.82196617e-02 -5.82933187e-01 3.00836354e-01 4.75843638e-01 -1.50044128e-01 -9.01402950e-01 -8.64635885e-01 -2.57168531e-01 2.70773858e-01 5.23072541e-01 5.92869103e-01 3.85668844e-01 -8.39291334e-01 -1.08231723e+00 -3.57217282e-01 -1.35791451e-01 -1.51523516e-01 1.55510381e-01 6.40955746e-01 -5.66192269e-01 8.24793756e-01 -2.48342872e-01 -9.56798419e-02 -1.04710972e+00 2.71742523e-01 2.15059131e-01 -4.34616864e-01 -5.75012505e-01 5.87238073e-01 -1.15508258e-01 -2.80961066e-01 1.28127933e-01 -4.04181987e-01 -1.85288385e-01 2.14666143e-01 3.64477426e-01 6.18787110e-01 2.21535802e-01 -1.93488836e-01 2.00208910e-02 2.09649965e-01 -3.89002115e-01 -2.94394698e-02 1.54405606e+00 6.29128963e-02 -8.30577612e-02 3.02703798e-01 8.41399550e-01 5.36454506e-02 -8.90609980e-01 2.20827758e-02 1.29984438e-01 6.22356348e-02 -1.52508304e-01 -9.49369788e-01 -8.93317461e-01 1.03338921e+00 -1.76373180e-02 4.81758058e-01 1.10416758e+00 1.79157592e-03 1.03833187e+00 8.12716424e-01 -1.63570642e-01 -1.09792197e+00 1.78424761e-01 8.02838326e-01 1.03999209e+00 -1.09750617e+00 3.92265946e-01 1.35272771e-01 -7.76586592e-01 1.30213368e+00 4.89797443e-01 -3.55758548e-01 1.64573476e-01 -1.74058110e-01 -2.37703219e-01 -3.65259707e-01 -9.59472418e-01 -9.64762419e-02 1.88066721e-01 4.39822227e-01 8.74776781e-01 1.80037115e-02 -4.41951305e-01 6.87559724e-01 -7.67556250e-01 -4.74510565e-02 8.46430659e-01 6.07245624e-01 -8.12683284e-01 -9.60065544e-01 -1.01665303e-01 9.27258968e-01 -6.38946831e-01 -2.95816332e-01 -9.28431571e-01 8.42649281e-01 -3.20939004e-01 1.11612129e+00 -3.59501056e-02 -2.89989084e-01 4.71518159e-01 1.87925696e-01 6.40493333e-01 -8.60067844e-01 -9.23970878e-01 -3.24162662e-01 5.63855827e-01 -5.49424142e-02 -1.73374191e-01 -7.32215405e-01 -1.24783671e+00 -6.33268654e-01 -2.46489793e-01 -6.57595471e-02 4.98630643e-01 9.93236840e-01 5.17859519e-01 6.23376667e-01 1.25590160e-01 -6.55931890e-01 -7.75684655e-01 -1.29920506e+00 -1.82038337e-01 2.17348337e-01 1.22025326e-01 2.06709430e-01 1.19487084e-02 6.28515780e-02]
[12.371536254882812, 9.403623580932617]
167969aa-a8ba-45a5-a036-50e574e9d2b4
volumetric-and-multi-view-cnns-for-object
1604.03265
null
http://arxiv.org/abs/1604.03265v2
http://arxiv.org/pdf/1604.03265v2.pdf
Volumetric and Multi-View CNNs for Object Classification on 3D Data
3D shape models are becoming widely available and easier to capture, making available 3D information crucial for progress in object classification. Current state-of-the-art methods rely on CNNs to address this problem. Recently, we witness two types of CNNs being developed: CNNs based upon volumetric representations versus CNNs based upon multi-view representations. Empirical results from these two types of CNNs exhibit a large gap, indicating that existing volumetric CNN architectures and approaches are unable to fully exploit the power of 3D representations. In this paper, we aim to improve both volumetric CNNs and multi-view CNNs according to extensive analysis of existing approaches. To this end, we introduce two distinct network architectures of volumetric CNNs. In addition, we examine multi-view CNNs, where we introduce multi-resolution filtering in 3D. Overall, we are able to outperform current state-of-the-art methods for both volumetric CNNs and multi-view CNNs. We provide extensive experiments designed to evaluate underlying design choices, thus providing a better understanding of the space of methods available for object classification on 3D data.
['Leonidas J. Guibas', 'Mengyuan Yan', 'Matthias Niessner', 'Hao Su', 'Angela Dai', 'Charles R. Qi']
2016-04-12
volumetric-and-multi-view-cnns-for-object-1
http://openaccess.thecvf.com/content_cvpr_2016/html/Qi_Volumetric_and_Multi-View_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Qi_Volumetric_and_Multi-View_CVPR_2016_paper.pdf
cvpr-2016-6
['3d-object-recognition']
['computer-vision']
[-0.21142255 -0.08810913 -0.24320441 -0.3570324 -0.49209148 -0.6327943 0.73459584 0.20711856 -0.14142933 0.25213274 0.19162375 -0.20867431 -0.05932335 -1.1671256 -0.6558174 -0.19594818 -0.1649012 0.42538887 0.28459382 -0.21965572 0.18313166 1.1681058 -1.6775457 0.6792851 0.08140536 1.6572212 -0.1457059 0.47520688 -0.37873074 0.43182588 -0.3505378 -0.30270824 0.37747133 0.21355964 -0.9128774 0.06876624 1.0301327 -0.45344725 -0.48859426 0.654727 0.56152105 -0.03804395 0.8289933 -1.0170296 -1.1620831 0.23449707 -0.5409863 0.5840959 0.06297404 0.1905721 1.2237386 -1.2145989 0.67227435 1.5371423 1.0687296 0.571418 -1.0972043 -0.35710093 0.48726374 -0.10375696 -1.1857804 -0.2937971 0.8876086 -0.59377664 1.4091141 -0.06959279 1.0382664 1.1094155 0.17368971 0.99208146 1.0531799 -0.04849696 0.00984119 -0.05168679 0.03749869 0.787889 0.45077097 0.15078294 -0.6017582 0.01931405 1.1641527 0.1532328 -0.16056064 -0.69840395 -0.9309625 0.9929323 0.73927164 0.37170282 -0.32663158 0.35786963 0.50040805 0.05449319 1.1755676 0.24493922 -0.43431064 0.17864469 -0.98570085 0.53623873 0.52829796 0.9691574 0.58528835 0.23597975 -0.17064466 0.8491834 0.38413638 0.2308981 0.18881226 -0.582651 0.635413 1.0038787 -0.36410445 -0.89818424 -0.6546207 -0.78739864 -0.80567175 0.36408827 0.3829218 0.39597765 -1.1521145 1.2955093 0.15754668 -0.04355574 -0.19320966 0.8166069 1.2908454 0.11094075 -0.19374175 0.39837787 1.3347961 -0.8251003 -0.21065374 -0.16195552 0.6476633 -0.5024641 0.8294352 0.2449831 -1.3783412 -0.68696034 -1.0794315 -0.35001653 -0.690153 -0.05839307 0.702442 0.83094424 -1.1880075 0.89096725 -0.82385325 -0.3395859 1.1115468 0.24184732 -0.5332133 -0.32243538 -0.7992187 0.957102 0.1546673 0.15439767 -0.9537708 -1.1159275 -0.9616557 -0.06743652 -0.11336159 -0.9021299 1.1714864 -0.4292452 -0.91398877 1.2668214 0.14503287 -0.40648305 0.50266963 -0.1123765 0.138848 0.2253931 -0.23685777 0.684602 0.60935664 -1.5140014 -0.46225804 -0.787885 0.43809915 0.00949265 -0.34885827 -0.3420359 -0.3402475 -0.5651718 0.20975451 -0.68043256 -0.12998545 0.59775454 -0.2884957 -0.38491085 1.099213 -0.18088931 0.5729363 -1.8733284 0.31074676 -0.18706341 0.76883066 0.30150744 -0.2754301 0.18277232 0.01630722 0.51883036 0.01615359 -0.8963145 -0.11300097 0.22447923 -0.09334211 0.69434524 0.5675733 1.203793 -0.6689274 -0.28750932 0.33643484 0.82135314 -0.66214985 0.14283004 -0.18314719 0.3202872 -0.4816166 0.9956038 0.7961388 -0.46895164 -0.18500051 -0.43563807 -0.08118971 0.27980307 -0.6709025 1.8079269 -0.62320364 0.71344334 0.07284268 -1.2112049 0.9846023 0.24012728 0.6767927 -0.670673 0.34005964 0.19783065 -0.31181198 -0.08512174 0.6107083 -0.24314886 0.1850115 0.36710706 0.5066026 -0.42238215 -0.10881077 -0.04446179 0.7825895 0.18575576 0.22850028 -0.2609127 0.18763539 -0.1383987 0.01103953 0.6559841 -0.27445784 0.9322596 0.35736278 -0.997508 -1.1490337 -0.99327826 -0.29927638 0.7388626 -0.08706944 -0.3154794 -0.38305843 -0.71907204 0.3926039 0.06203257 -1.0900159 0.01770163 -0.5931359 -0.5352164 0.6808159 1.0990667 0.47833017 -0.7517075 -0.7427059 0.11043791 0.20893374 -1.4146748 0.15052886 0.28918734 -1.4444602 -1.1412343 -1.0423518 -0.4507399 0.36173663 0.63704705 1.4819809 0.21417789 -0.2981914 0.77822006 -0.44536448 -0.67653114 -0.06125484 0.5195636 -0.14447907 -0.3033077 0.19379817 -0.88110363 -0.7600855 0.03687305 -0.90105164 -0.01853439 0.5078728 0.5356725 0.74069744 -0.5048815 0.30636242 -0.83333004 0.45110267 -0.5656974 -0.38733405 -0.01580174 -0.4043073 -0.0893746 0.4880181 -0.2748264 -0.5572451 -0.16567883 -0.4247413 -1.0627201 -0.3846386 0.45797947 0.19746262 -0.4182301 0.56586856 0.1972134 -0.02983534 -0.6808643 0.6590277 0.28126472 -0.00939102 -0.63531214 0.70189905 0.89221525 0.1514549 -0.71023023 -1.1296487 -0.5213112 -0.9786186 -0.3973511 0.8936474 -0.8748429 -0.5670922 0.5456958 -1.2443216 -0.27889398 -0.36880407 0.08223154 -0.71629566 0.08109377 -0.629474 -0.7474334 -0.29248238 -1.2863336 1.5379679 -0.05947033 0.16461329 -1.2890356 -0.11185523 0.36484206 0.72760785 0.445847 0.9738096 -0.6326315 -0.7214314 -0.16346957 -0.6723597 0.2501329 -0.01929394 -0.27606392 -1.3853538 -0.4268164 -0.23576911 -0.46921694 1.2388446 0.618135 1.4036665 0.08632002 -0.34897894 0.9088337 1.5055075 -0.26613045 0.5060477 0.2852296 0.9277644 0.4969675 0.0078036 0.32975957 0.42454457 0.6768574 1.0555085 -0.21206982 -0.51434594 -0.0989848 -0.3321681 0.7819604 -0.48838514 -0.25777018 -0.8948339 0.78189874 -1.4325712 -0.7344778 -0.06919515 1.722303 0.24652472 0.32103214 0.1296326 0.11829634 0.28176433 0.5515369 -0.5985997 -0.33202574 0.04330034 0.48602864 0.47799352 0.01436636 -1.3392888 0.7156652 6.8821144 0.60727906 -1.2731632 0.09530621 0.72244865 -0.2251943 -0.39714876 -0.5731809 -1.0085322 -0.25314036 0.54989713 0.35655546 0.01700535 1.0722661 -0.13471766 0.26279017 -1.3321466 1.0455266 0.29169244 -1.8944081 0.47863683 0.39834067 0.74939024 0.36540422 0.29513857 0.3052949 0.02135636 -1.3275688 1.0438992 0.31161386 0.8780998 -0.504361 0.6498182 0.20953004 -1.7354739 -0.01369949 -0.4969289 -0.0656306 0.01345623 0.57806635 -0.26933292 0.8015741 0.89405125 1.098384 -0.65243113 1.018672 0.35656154 0.14433151 -0.18340486 0.04198585 0.5045415 0.26445308 0.37364626 1.2621802 0.2986461 -0.0627852 0.25178042 1.2644295 -0.41441384 0.0103442 -1.0205102 0.04005359 0.11141048 1.0612237 -0.96097744 -0.17034133 -0.7832037 0.4756969 0.7537324 0.19339886 -0.6458826 0.14511095 0.8408348 0.27549085 0.49006486 -0.56355166 -0.5200477 -1.0321935 -0.12314879 -0.4503052 0.16743942 -0.6657196 -1.6126477 0.5969453 0.1542075 -1.1197332 0.19701543 -1.1666611 -0.28535056 0.8541427 -1.9664543 -1.5222847 -0.5530425 0.5320118 0.8218798 -0.17598842 0.8126728 0.30530983 -0.02813495 0.39509964 -0.31518826 0.27066055 0.23919135 -1.0681086 0.701796 0.32183757 0.20378116 0.5476531 0.06061205 -0.3792443 -1.568036 -1.2812446 0.43152192 -0.6231074 0.31979907 -0.45374668 -0.7752437 0.7702941 -0.17807087 0.68802583 0.60138714 0.41226214 -0.88103664 0.27863443 -1.0021963 0.18016192 1.3579135 -0.6768357 -0.3853902 0.11360269 0.62742925 -0.59039295 -1.2269161 0.69342864 0.7236659 -1.3165938 1.4244155 -0.7711815 0.8415838 0.22955516 -0.43251687 -1.288855 -0.40703383 0.32181326 -0.39864054 0.7341733 0.2935501 -0.49637362 0.98046917 0.03832643 -0.5411888 -1.4543468 -1.0718317 -0.8944793 0.7020098 -0.6873241 0.47902754 0.79414254 -0.55982894 0.20267758 0.06976326 -0.07227957 0.51219267 0.2707927 0.6054531 -1.5011117 0.00872638 -0.96508414 -0.720162 -1.2184727 0.00967241 -1.0730208 -0.45061007 -1.996302 0.3615475 -0.4426656 -0.20078416 0.4215392 0.20518647 0.83011585 0.36036488 0.13048369 -0.3933493 0.60516477 1.7278895 -0.42302737 0.09927564 -0.09689885 -0.5672447 0.76308775 0.7441297 -0.04302175 -0.36088696 -0.6887792 0.11867784 -0.09190401 0.7994528 -1.0396752 -0.14586093 0.16946103 0.7738408 -0.9611158 0.68796414 -0.57997143 -0.45564282 0.14540346 -0.20242718 0.08921912 0.50438803 0.6311084 -0.20030837 0.10313508 0.82091177 -0.69863415 -0.49917734 0.8925044 -0.2743385 0.12138536 0.68398356 -0.55439806 -0.3901566 -0.28397676 -0.93289614 -0.19334938 0.3069079 0.6256367 0.87306523 -1.4915106 -0.5698712 0.08235177 0.29488614 0.17161924 0.33185077 0.7183209 -0.49437377 0.63649946 -0.34281385 -0.83871454 -1.0444736 0.47749156 0.6129687 -0.38497677 -0.74827266 0.9980411 0.35261744 -0.6192627 0.37129545 -0.7091098 -0.32659003 0.15849489 0.27188134 0.08744916 0.46028876 -0.6377475 -0.31834847 0.8185972 0.0809931 0.26428115 1.6200755 0.19294594 0.13455817 0.5892335 1.3986064 -0.5125028 -1.1671945 -0.14653845 -0.33358636 -0.45320135 0.24049753 -0.3593361 -1.4800829 1.4842283 0.57951677 0.47235402 0.7615979 0.18168882 0.8151316 0.11885254 0.48291305 -0.5865362 0.36620244 0.7656286 1.0131794 -1.1993005 0.15522441 -0.5689093 0.03375729 1.399942 0.6432859 -0.45031947 1.1062706 0.1121919 -0.21772334 -0.76127386 -0.55196524 -0.4462765 0.58006287 0.8550194 0.6933708 -0.14084171 0.33159766 0.3343735 -0.11216824 -0.26747274 0.24021004 0.89415264 -0.29394123 -0.8029753 -0.19522472 0.5975034 -0.49734706 -0.06018246 -0.51200426 1.1146275 0.15676495 0.49143812 0.26607755 -0.28401884 0.5319675 -0.05146475 0.92389494 -0.8060712 -0.63618743 -0.11071005 -0.09759074 -0.7247439 -0.70450646 -0.46628797 -0.5607748 -0.35088354 -0.33894676 -0.63329357 0.6416299 1.0093687 0.1936951 0.713913 0.2044478 -1.6916996 -0.42261076 -0.7620654 -0.49243718 0.15436278 0.27073738 -1.0645217 -0.06423502 -0.25472987]
[8.173046112060547, -3.6383273601531982]
d5fd3fef-9716-4587-9a9e-ef60116190f6
scale-recovery-for-monocular-visual-odometry
null
null
http://openaccess.thecvf.com/content_iccv_2017/html/Yin_Scale_Recovery_for_ICCV_2017_paper.html
http://openaccess.thecvf.com/content_ICCV_2017/papers/Yin_Scale_Recovery_for_ICCV_2017_paper.pdf
Scale Recovery for Monocular Visual Odometry Using Depth Estimated With Deep Convolutional Neural Fields
Scale recovery is one of the central problems for monocular visual odometry. Normally, road plane and camera height are specified as reference to recover the scale. The performances of these methods depend on the plane recognition and height measurement of camera. In this work, we propose a novel method to recover the scale by incorporating the depths estimated from images using deep convolutional neural fields. Our method considers the whole environmental structure as reference rather than a specified plane. The accuracy of depth estimation contributes to the scale recovery. We improve the performance of depth estimation by considering two consecutive frames and egomotion of camera into our networks. The depth refinement and scale recovery are obtained iteratively. In this way, our method can eliminate the scale drift and improve the depth estimation simultaneously. The effectiveness of our method is verified on the KITTI dataset for both visual odometry and depth estimation tasks.
['Xiangwei Wang', 'Qijun Chen', 'Xiaochuan Yin', 'Xiaoguo Du']
2017-10-01
null
null
null
iccv-2017-10
['monocular-visual-odometry']
['robots']
[ 5.12919538e-02 -1.99735105e-01 -1.26090169e-01 -4.68980908e-01 -1.70436874e-01 -4.36625808e-01 4.23844814e-01 -4.12961870e-01 -6.12788796e-01 5.60894608e-01 -1.21127758e-02 1.68176919e-01 2.16909409e-01 -1.02932048e+00 -8.18265676e-01 -6.04743123e-01 6.39833212e-01 2.74058849e-01 7.14528441e-01 -8.79440382e-02 5.40611506e-01 6.41976535e-01 -1.53737068e+00 -3.14789563e-01 8.62304509e-01 1.08642423e+00 4.17407811e-01 7.81516850e-01 -5.07917516e-02 5.94868302e-01 -3.56759161e-01 3.33460458e-02 5.05134165e-01 9.14627537e-02 -3.96043122e-01 4.27500248e-01 7.15135396e-01 -9.13069785e-01 -5.62837541e-01 1.29431593e+00 5.40724397e-01 -7.38105774e-02 4.73842323e-01 -9.27091181e-01 -4.01919112e-02 6.01225570e-02 -8.28781307e-01 9.71252769e-02 2.37964854e-01 -2.43143868e-02 5.51411808e-01 -8.09567750e-01 6.22666001e-01 1.17011774e+00 6.35056794e-01 8.66265595e-02 -7.53797412e-01 -6.71218514e-01 1.77141994e-01 3.27125549e-01 -1.50704074e+00 -5.72589815e-01 9.82366323e-01 -5.15783072e-01 6.37362778e-01 -3.95366758e-01 6.16889954e-01 4.81404454e-01 1.07586227e-01 4.16418523e-01 9.00419354e-01 -2.17721939e-01 1.62542149e-01 -1.50110304e-01 -2.20908970e-01 5.98913133e-01 6.19824708e-01 1.30325407e-01 -4.01180923e-01 4.03785616e-01 1.36811113e+00 2.32960552e-01 -4.80563849e-01 -6.24340653e-01 -1.13406992e+00 5.29371679e-01 6.01422071e-01 -2.09529430e-01 -1.36771560e-01 2.39583030e-01 1.90324113e-01 7.56376535e-02 4.42245781e-01 2.46497411e-02 -3.48736137e-01 -8.30662772e-02 -8.42487931e-01 6.99545518e-02 3.81722748e-01 1.04688251e+00 1.29816020e+00 1.46348625e-01 4.66954291e-01 5.83635390e-01 3.91521633e-01 9.41815495e-01 5.03920913e-01 -1.12603688e+00 6.35115445e-01 7.52013624e-01 4.25167352e-01 -1.18892264e+00 -7.06771374e-01 -2.87035286e-01 -7.17202008e-01 4.54770386e-01 6.19324386e-01 -1.70567214e-01 -1.13435137e+00 1.42876077e+00 5.91359496e-01 3.09542000e-01 3.19751315e-02 1.20806468e+00 7.52873778e-01 3.26773345e-01 -6.67090595e-01 5.78374835e-03 1.17100120e+00 -8.72107327e-01 -7.18437254e-01 -7.09854245e-01 4.48266238e-01 -7.93383420e-01 5.83125830e-01 5.22473037e-01 -8.39753091e-01 -7.10271835e-01 -1.44307518e+00 -4.44219977e-01 1.42506296e-02 5.93296289e-01 3.14797878e-01 3.54836494e-01 -7.91665196e-01 4.52125430e-01 -1.13610268e+00 -3.47397506e-01 -1.47766739e-01 4.72813040e-01 -3.02872896e-01 -1.39431089e-01 -1.16106772e+00 7.52053022e-01 3.36610734e-01 2.90828258e-01 -4.09750462e-01 -2.19740063e-01 -9.96969998e-01 -1.67094901e-01 2.06926212e-01 -7.41613984e-01 1.20150292e+00 -8.74049366e-01 -1.69943917e+00 9.14374888e-01 -3.32655936e-01 -3.01218688e-01 8.19516301e-01 -3.99478644e-01 -3.85134555e-02 3.70762259e-01 1.13031141e-01 5.01933634e-01 7.24970102e-01 -9.72874761e-01 -9.79458272e-01 -6.53252006e-01 2.08309978e-01 6.17845833e-01 1.05948269e-01 -6.80850923e-01 -7.65905797e-01 2.84353793e-02 9.21923280e-01 -9.73303318e-01 -1.93235390e-02 2.49355540e-01 -1.68098152e-01 3.13636154e-01 5.75247824e-01 -7.93164611e-01 8.73495400e-01 -1.94421029e+00 8.75012130e-02 -3.33779640e-02 2.56797284e-01 -1.03848213e-02 2.29851216e-01 -1.17000982e-01 3.57101619e-01 -4.75633919e-01 1.77532248e-02 -3.37208897e-01 -3.60197991e-01 1.50048539e-01 -7.12671801e-02 8.82053971e-01 -3.08540054e-02 6.12711370e-01 -5.95377624e-01 -4.28515583e-01 5.11974096e-01 6.50603831e-01 -4.45480883e-01 2.16885149e-01 7.38381445e-02 5.94604433e-01 -3.36631328e-01 4.33926791e-01 1.21939290e+00 -9.08885375e-02 2.13634390e-02 -4.59767908e-01 -3.75001878e-01 4.28042024e-01 -1.66441274e+00 1.76234615e+00 -4.59161639e-01 6.19902849e-01 1.29459500e-01 -5.36684453e-01 1.17494285e+00 -6.10796064e-02 2.37632111e-01 -8.40755641e-01 1.91685915e-01 4.21894878e-01 -9.73306596e-02 -3.87624532e-01 6.51549757e-01 9.30166803e-03 3.56398523e-01 -1.22522034e-01 -3.58993709e-01 -5.01769423e-01 -1.59106985e-01 -1.95940927e-01 7.23462105e-01 3.58182460e-01 6.00691319e-01 5.72980940e-02 7.65925586e-01 -2.01193303e-01 9.21461046e-01 1.99025840e-01 -3.40067625e-01 8.07601035e-01 2.17775881e-01 -5.28330445e-01 -1.33545339e+00 -9.01798248e-01 -8.29582885e-02 1.90743953e-01 7.96308041e-01 7.76830912e-02 -4.76211011e-01 -2.30650753e-01 2.17053499e-02 -6.87709600e-02 -5.10148048e-01 -2.55534034e-02 -7.56028652e-01 -4.27635282e-01 2.62716293e-01 6.76894724e-01 8.97623658e-01 -4.56979096e-01 -6.61922991e-01 1.12644203e-01 -4.19500142e-01 -1.51570284e+00 -4.82486993e-01 -3.21760327e-02 -1.15471685e+00 -1.13528180e+00 -7.11099565e-01 -7.08455026e-01 6.34102643e-01 7.13324726e-01 6.40390337e-01 -1.67593703e-01 2.63960958e-01 -2.90039062e-01 7.28425011e-02 -1.27325788e-01 8.80133137e-02 3.61976266e-01 1.19060680e-01 -9.38671678e-02 1.41593486e-01 -8.78669322e-01 -1.05524898e+00 6.12816036e-01 -6.70945823e-01 2.55100876e-01 4.12494779e-01 4.75456446e-01 5.57589352e-01 -1.51904076e-02 -5.93934283e-02 -6.00045025e-01 -1.49879605e-01 -1.94866378e-02 -1.23367536e+00 -2.51370817e-01 -5.56336105e-01 1.88349396e-01 3.27855349e-01 -4.01993871e-01 -1.10812950e+00 4.40413535e-01 -6.54108450e-02 -4.09881234e-01 -2.67664306e-02 1.52340770e-01 -3.74530584e-01 -2.12344572e-01 6.34666085e-01 1.51521280e-01 -6.50416315e-02 -3.31208527e-01 1.01704651e-03 6.12063110e-01 6.55171037e-01 -2.62048066e-01 9.51715529e-01 1.06944084e+00 3.12722713e-01 -8.55787158e-01 -7.88546205e-01 -6.61788046e-01 -1.01707375e+00 -4.86936234e-02 8.08496356e-01 -1.54177535e+00 -7.53552735e-01 8.80341649e-01 -1.30620492e+00 -1.20862849e-01 3.39662343e-01 7.75392056e-01 -4.09756869e-01 7.01154947e-01 -6.41889274e-01 -6.70601964e-01 -1.92404240e-01 -1.32175195e+00 1.29314160e+00 4.44408506e-01 3.86386216e-01 -9.00830448e-01 2.92288005e-01 1.70906261e-01 9.56964716e-02 1.54666901e-01 1.26038969e-01 3.29586565e-01 -1.00217891e+00 -1.30748898e-01 -5.52624345e-01 1.01000495e-01 2.81537771e-01 6.26702979e-02 -1.17985690e+00 -2.83232719e-01 1.06047809e-01 4.34562415e-02 8.41805577e-01 5.42155087e-01 3.92902285e-01 8.07800367e-02 -1.28202349e-01 1.30470014e+00 1.62391078e+00 2.90492654e-01 8.36892307e-01 8.96292210e-01 1.13321972e+00 5.79071224e-01 8.17550838e-01 4.95091736e-01 7.43720591e-01 8.46440077e-01 6.36823773e-01 -5.58578745e-02 -9.63446945e-02 -3.21171194e-01 2.69140124e-01 8.99282217e-01 -2.10365251e-01 1.52834311e-01 -9.32791054e-01 5.41180432e-01 -1.83562827e+00 -7.07077265e-01 -3.62625778e-01 2.33714151e+00 6.42703831e-01 2.59826869e-01 -2.42446721e-01 9.37793106e-02 8.08254957e-01 1.88947484e-01 -8.14090729e-01 -2.40391381e-02 -1.26568839e-01 -2.46671140e-01 1.00517523e+00 8.17710698e-01 -1.02440798e+00 9.50857460e-01 5.30336189e+00 3.40007454e-01 -1.50744748e+00 -2.93510020e-01 1.21333547e-01 8.75607505e-02 -6.63973093e-02 1.03801697e-01 -1.30007267e+00 2.98522502e-01 4.83728319e-01 9.40224826e-02 4.23641831e-01 9.36751604e-01 3.71221244e-01 -3.39349210e-01 -7.78605521e-01 1.25343597e+00 -1.67710427e-02 -9.04577732e-01 -3.25614184e-01 1.28169522e-01 9.26954567e-01 2.77081549e-01 -2.14671463e-01 -1.59063801e-01 9.57488269e-02 -5.97340763e-01 7.85516798e-01 3.86419386e-01 9.55298126e-01 -7.79811859e-01 8.10864747e-01 4.80949968e-01 -1.54819417e+00 1.32853016e-01 -6.97684467e-01 -4.55389947e-01 2.08968744e-01 7.71443903e-01 -7.70972967e-01 6.73159301e-01 5.37841499e-01 1.12080717e+00 -5.56010306e-01 1.07081175e+00 -5.53435802e-01 2.92000026e-02 -5.07163048e-01 5.14042318e-01 -1.00372069e-01 -3.64918858e-01 3.30985397e-01 7.09044933e-01 4.85624105e-01 -9.93425921e-02 9.72704813e-02 6.00739598e-01 5.18526137e-02 -6.56399503e-02 -5.97489119e-01 4.71551150e-01 5.30360043e-01 1.10234642e+00 -5.91561139e-01 -1.61834002e-01 -4.78055924e-01 9.16949868e-01 3.10506910e-01 3.11898619e-01 -7.70757437e-01 -5.24959564e-01 7.64678597e-01 3.10219467e-01 3.96631211e-01 -5.15416324e-01 -4.40421700e-01 -1.60466146e+00 1.73975453e-01 -4.58400905e-01 -1.10508777e-01 -9.40785050e-01 -6.51488364e-01 4.66960013e-01 -3.29597950e-01 -1.66403115e+00 -1.50034815e-01 -5.97451925e-01 -3.57943922e-01 8.44810307e-01 -2.05139041e+00 -8.59709263e-01 -1.05907524e+00 4.66756642e-01 4.12418842e-01 3.90218616e-01 1.33260667e-01 4.97497380e-01 -3.54227304e-01 2.35520080e-01 -1.33948354e-02 2.00015575e-01 1.01839602e+00 -1.02088547e+00 6.31658852e-01 1.07440937e+00 -3.51907849e-01 4.26603585e-01 6.46851838e-01 -8.01563621e-01 -1.00938809e+00 -8.51229906e-01 6.01356685e-01 -1.11708455e-01 5.25902748e-01 -9.30714160e-02 -8.71873915e-01 7.46100485e-01 -3.31018597e-01 3.66326831e-02 -3.56598228e-01 -3.10712636e-01 -2.89593309e-01 -4.73043412e-01 -9.77961183e-01 3.37949306e-01 1.06332231e+00 -6.55833781e-01 -2.92046517e-01 -2.85685748e-01 7.85431445e-01 -1.09255099e+00 -6.01959050e-01 5.78623235e-01 8.17166507e-01 -1.10833108e+00 9.99372721e-01 3.23420763e-01 5.31209528e-01 -7.76842058e-01 -1.50763318e-01 -9.42282736e-01 -1.23345152e-01 -1.61963105e-01 -3.79547924e-02 1.00082946e+00 -8.27209279e-02 -7.73514688e-01 1.18570638e+00 3.28779042e-01 1.07106287e-02 -2.04379275e-01 -8.85960042e-01 -4.16300386e-01 -2.16424450e-01 -3.21968108e-01 6.40105426e-01 7.44346976e-01 -4.54674602e-01 4.31998402e-01 -5.94889462e-01 6.53205037e-01 7.97831178e-01 9.47014913e-02 1.29824984e+00 -1.41871309e+00 -1.82923257e-01 4.04511727e-02 -7.60537267e-01 -1.69028246e+00 -1.82614058e-01 -2.51442224e-01 1.07186049e-01 -1.57661557e+00 -8.43154639e-02 -4.67372872e-02 1.29544407e-01 -1.47693112e-01 -1.06239088e-01 1.66448742e-01 -1.37308970e-01 4.41492468e-01 -1.30443215e-01 5.10427833e-01 1.50712776e+00 2.03844070e-01 -3.85796219e-01 5.75019494e-02 -1.54679388e-01 1.12394929e+00 9.73722756e-01 -2.53531218e-01 -5.69163322e-01 -7.87496865e-01 4.90960419e-01 3.53210419e-01 1.87878057e-01 -1.14001083e+00 3.83176416e-01 -2.32977197e-01 5.59103608e-01 -9.42104757e-01 4.86228198e-01 -8.22119892e-01 4.82108034e-02 4.74672854e-01 1.01351961e-01 8.92369002e-02 -9.29435715e-03 5.78455448e-01 -2.97759533e-01 -6.83728382e-02 1.12452626e+00 -1.21305689e-01 -9.01205182e-01 3.77552927e-01 1.96305037e-01 -1.12441376e-01 5.98839581e-01 -5.65074503e-01 -5.95958710e-01 -4.56903428e-01 -3.41822207e-01 2.18728825e-01 1.11603725e+00 2.12610006e-01 8.07351410e-01 -1.27206028e+00 -3.64804298e-01 3.95618975e-01 2.63449103e-01 6.49892688e-01 2.79375881e-01 1.05260539e+00 -1.12355757e+00 3.40796351e-01 -2.31961027e-01 -9.08299565e-01 -1.10873008e+00 3.22443485e-01 8.09229732e-01 -1.75330207e-01 -6.78747296e-01 4.37195808e-01 6.23342752e-01 -4.99016970e-01 1.21457055e-01 -7.55201459e-01 -4.04280543e-01 -1.79516241e-01 5.58251023e-01 5.45137525e-01 -2.07423326e-02 -7.34097064e-01 -3.19100142e-01 1.35271204e+00 -3.19721475e-02 -1.93771169e-01 1.03085935e+00 -7.97121644e-01 4.32988703e-02 4.44852173e-01 1.33002245e+00 1.63515419e-01 -1.73622453e+00 -3.02377582e-01 -2.63932645e-01 -5.88558137e-01 1.27194852e-01 -1.03205614e-01 -1.16359103e+00 1.06399345e+00 6.88780129e-01 -5.28558969e-01 1.02769327e+00 -5.63105702e-01 8.50293815e-01 3.79411995e-01 5.00151455e-01 -1.11889803e+00 -1.18283369e-01 6.31094754e-01 4.03785318e-01 -1.61485660e+00 2.85644650e-01 -6.44831479e-01 -3.08879286e-01 1.36097562e+00 8.15995574e-01 -1.92175493e-01 4.28486228e-01 1.45453855e-01 2.55025059e-01 1.22873664e-01 -2.12539449e-01 -3.39096993e-01 2.61386484e-01 4.25949007e-01 1.34120941e-01 -2.32811332e-01 -2.19087094e-01 -1.36299767e-02 -1.36581168e-01 1.03102751e-01 8.44334245e-01 7.39857495e-01 -9.02369320e-01 -7.49603450e-01 -5.86971521e-01 -2.69078463e-01 -1.32249236e-01 8.99008512e-02 -9.98439491e-02 6.47515655e-01 1.27431139e-01 7.77162075e-01 1.36091501e-01 -4.85805243e-01 3.71658564e-01 -3.59911740e-01 3.85292858e-01 -4.58867580e-01 1.13002181e-01 2.00776741e-01 -3.02250907e-02 -5.67128301e-01 -5.33100069e-01 -3.66216183e-01 -1.55320442e+00 -4.44686890e-01 -3.36189657e-01 -2.75000393e-01 8.52441847e-01 7.22306609e-01 9.77472067e-02 2.19657183e-01 8.32164347e-01 -1.02600312e+00 -2.35188201e-01 -8.15648258e-01 -7.49526739e-01 2.54377991e-01 7.30024934e-01 -7.91644871e-01 -7.31250167e-01 -1.86637100e-02]
[8.066484451293945, -2.2246460914611816]
50b44127-8010-44e7-bc73-f14a0e02a770
deep-neural-networks-based-invisible
2102.09173
null
https://arxiv.org/abs/2102.09173v1
https://arxiv.org/pdf/2102.09173v1.pdf
Deep Neural Networks based Invisible Steganography for Audio-into-Image Algorithm
In the last few years, steganography has attracted increasing attention from a large number of researchers since its applications are expanding further than just the field of information security. The most traditional method is based on digital signal processing, such as least significant bit encoding. Recently, there have been some new approaches employing deep learning to address the problem of steganography. However, most of the existing approaches are designed for image-in-image steganography. In this paper, the use of deep learning techniques to hide secret audio into the digital images is proposed. We employ a joint deep neural network architecture consisting of two sub-models: the first network hides the secret audio into an image, and the second one is responsible for decoding the image to obtain the original audio. Extensive experiments are conducted with a set of 24K images and the VIVOS Corpus audio dataset. Through experimental results, it can be seen that our method is more effective than traditional approaches. The integrity of both image and audio is well preserved, while the maximum length of the hidden audio is significantly improved.
['Thanh Ta Minh', 'Toan Pham Van', 'Ngoc N. Tran', 'Thoi Hoang Dinh', 'Quang Pham Huu']
2021-02-18
null
null
null
null
['image-steganography']
['computer-vision']
[ 6.94131017e-01 9.57227424e-02 1.95779633e-02 3.14355530e-02 -5.36339879e-01 -7.61316270e-02 2.83941120e-01 -2.80027986e-01 -4.50894684e-01 3.52687567e-01 6.36394620e-02 -3.61634612e-01 2.96626985e-01 -7.28146076e-01 -6.46705210e-01 -1.12471366e+00 -3.07564139e-01 -2.57884085e-01 4.79324877e-01 -2.32122153e-01 3.82198393e-01 8.76595303e-02 -1.21755505e+00 2.70421088e-01 4.50874478e-01 1.13485014e+00 -1.22849410e-02 5.72729945e-01 2.11082354e-01 8.37642252e-01 -8.85786176e-01 -2.26182863e-01 2.13336468e-01 -8.17231119e-01 -5.79599917e-01 2.12217346e-01 -3.02440912e-01 -4.57215548e-01 -8.69669616e-01 1.17330623e+00 7.86463678e-01 -2.94703931e-01 1.14071243e-01 -1.17254031e+00 -5.12388289e-01 7.89026916e-01 -6.90615535e-01 4.37077954e-02 3.78026552e-02 1.48905054e-01 6.59762204e-01 -3.21353316e-01 1.94289312e-01 1.34222436e+00 4.66026932e-01 4.00205761e-01 -7.97911525e-01 -1.26187789e+00 -2.84248590e-01 7.20090806e-01 -1.50326478e+00 -5.81864119e-01 1.12762082e+00 -1.29532740e-01 5.59340060e-01 5.34085222e-02 7.74042606e-01 7.95321584e-01 5.43141782e-01 8.39748383e-01 1.11314178e+00 -7.56067574e-01 -1.71178058e-02 9.73379612e-02 -4.58299667e-01 4.87811685e-01 1.15245923e-01 3.84489894e-01 -3.13099325e-01 1.01967573e-01 5.67275822e-01 3.81658152e-02 -5.89198589e-01 -3.11027318e-01 -1.14482677e+00 1.01563299e+00 3.72497052e-01 6.27139688e-01 2.63782293e-02 3.54952633e-01 5.26062310e-01 6.29307091e-01 3.62539083e-01 2.81262188e-03 1.09011039e-01 1.10330738e-01 -9.97691274e-01 2.47178599e-02 8.40537250e-01 3.19766313e-01 4.23961908e-01 4.25377399e-01 4.82189208e-01 5.67616642e-01 5.86976588e-01 3.62289250e-01 7.94692218e-01 -6.65312827e-01 4.06589746e-01 1.66626140e-01 -3.99497777e-01 -1.64017117e+00 7.08483346e-03 -2.45765373e-01 -1.33440709e+00 4.01286393e-01 9.44352001e-02 -2.13456914e-01 -8.52222919e-01 1.34975290e+00 1.73681751e-02 4.31273013e-01 2.57893294e-01 7.19167352e-01 6.51792884e-01 1.01700008e+00 -4.00544703e-01 -1.37587696e-01 1.07216227e+00 -9.04555559e-01 -9.96150970e-01 -2.81141609e-01 3.65889788e-01 -8.54816079e-01 2.42038906e-01 5.41598618e-01 -8.94716024e-01 -5.91265798e-01 -1.52078259e+00 2.45286763e-01 -2.12339386e-01 -3.42383623e-01 2.17941463e-01 9.33884203e-01 -9.13456619e-01 4.89328265e-01 -5.40321231e-01 9.33952928e-02 4.42202389e-01 6.76222086e-01 -2.76610643e-01 5.86456293e-03 -1.63232219e+00 3.88306379e-01 9.29022908e-01 2.06706718e-01 -1.06692195e+00 1.42601565e-01 -8.99820268e-01 2.26926103e-01 3.48558426e-01 -1.36448830e-01 9.77318704e-01 -1.23077142e+00 -1.58891881e+00 8.11932147e-01 2.88503319e-01 -7.14041531e-01 4.50533688e-01 2.12993994e-01 -8.02627802e-01 3.62119943e-01 -2.72839308e-01 5.55317879e-01 1.50574470e+00 -1.28376150e+00 -6.67861044e-01 -2.06138119e-01 -1.50522545e-01 -1.97274879e-01 -5.87753177e-01 3.39365527e-02 -4.79757369e-01 -7.73186743e-01 2.25987270e-01 -1.03304648e+00 -9.75929871e-02 -1.80902004e-01 -4.12281066e-01 1.67714193e-01 1.19180810e+00 -8.91021192e-01 1.47513509e+00 -2.47872353e+00 1.57629281e-01 2.97411025e-01 3.05969507e-01 7.30615973e-01 -5.64925745e-02 5.44974208e-01 -3.96103412e-01 2.16130123e-01 -3.09286624e-01 -1.58633798e-01 -2.62080044e-01 1.37618527e-01 -4.65131491e-01 6.05896652e-01 -3.16048741e-01 7.06262529e-01 -5.23780584e-01 -5.77684820e-01 1.12585917e-01 6.88751101e-01 -3.87003273e-01 1.05408035e-01 1.68597892e-01 6.48594022e-01 -2.13552818e-01 2.93814838e-01 8.76870573e-01 -1.13923758e-01 3.22838008e-01 1.24638125e-01 1.06158391e-01 8.78852606e-03 -1.04381168e+00 1.41742992e+00 -7.41558746e-02 9.10797536e-01 1.43885091e-01 -1.28588974e+00 8.45158756e-01 7.32311487e-01 5.09866238e-01 -7.52037704e-01 5.30738175e-01 3.22412878e-01 3.85898173e-01 -7.79411614e-01 3.24809760e-01 -2.93283254e-01 -1.52210608e-01 5.49175680e-01 -3.35606694e-01 9.55625400e-02 -1.50977001e-01 -2.12944783e-02 8.74857426e-01 -4.30216640e-01 2.06134588e-01 4.60974872e-01 8.87358367e-01 -4.28293407e-01 4.88709420e-01 4.16330189e-01 -1.82594895e-01 3.36142182e-01 5.09068370e-01 -2.82946229e-01 -1.07315648e+00 -4.04517621e-01 2.77665734e-01 5.66187024e-01 4.65615243e-01 -2.74297953e-01 -8.29517782e-01 -4.29062158e-01 -1.73714042e-01 2.15993464e-01 -2.55283922e-01 -5.78032553e-01 -7.91794658e-01 -3.55908036e-01 1.06106901e+00 -2.24670917e-02 1.22362566e+00 -1.48015344e+00 -5.32663643e-01 2.75799304e-01 -4.13757533e-01 -1.13632679e+00 -3.53306651e-01 -2.60695457e-01 -8.63024175e-01 -8.56402278e-01 -9.43280101e-01 -1.17755032e+00 4.17244583e-01 5.10600328e-01 3.98050487e-01 4.50834215e-01 -1.25086084e-01 -1.86603159e-01 -5.17097354e-01 -5.33541024e-01 -9.20741737e-01 -7.61781260e-03 -1.34928808e-01 3.81478727e-01 1.84123427e-01 -6.25507236e-01 -5.09071112e-01 1.99345335e-01 -1.30193305e+00 -2.06783619e-02 9.85263169e-01 8.21742594e-01 2.43614584e-01 9.23791409e-01 3.39748174e-01 -7.16146886e-01 3.97657007e-01 -2.93242037e-01 -4.12725598e-01 -1.87977985e-01 -5.01794040e-01 -2.09275946e-01 5.07048190e-01 -4.74732786e-01 -6.64246440e-01 -2.35660762e-01 -4.25685704e-01 -2.80183733e-01 -7.82752112e-02 6.34639502e-01 -4.33695287e-01 -3.93811464e-01 -3.09902821e-02 6.57949328e-01 4.16202992e-01 -4.43844974e-01 -2.09255427e-01 1.19601059e+00 3.70903939e-01 3.00043136e-01 9.12486732e-01 6.22214854e-01 -1.69374973e-01 -1.02511001e+00 -3.23008299e-01 -2.43244529e-01 -3.29492360e-01 -1.01493537e-01 6.70970023e-01 -9.00331497e-01 -8.15650225e-01 1.11120498e+00 -1.10556257e+00 -8.72789603e-03 2.23348796e-01 4.89632785e-01 -4.07287747e-01 9.09429848e-01 -5.62089920e-01 -6.73180878e-01 -2.07548887e-01 -1.29433584e+00 7.28133321e-01 -1.69598963e-02 1.93626121e-01 -9.94350433e-01 3.21686193e-02 4.25036967e-01 1.44630298e-01 3.95370215e-01 9.31139648e-01 -4.26183701e-01 -7.54607618e-01 -6.96959257e-01 -1.10570900e-01 6.73827708e-01 1.41257450e-01 -5.54886758e-01 -8.44106019e-01 -7.22566128e-01 4.57121968e-01 -2.11053386e-01 1.13647604e+00 1.58666044e-01 1.32311487e+00 -5.84882319e-01 -3.06244701e-01 8.26337874e-01 1.45225549e+00 6.93409443e-01 1.26580703e+00 6.28409982e-01 6.93523407e-01 4.69508201e-01 1.87414035e-01 2.34087631e-01 1.13019146e-01 5.53895295e-01 7.16961145e-01 -1.58450410e-01 5.38870730e-02 -1.51328906e-01 5.77425778e-01 1.19106007e+00 9.77639258e-02 -5.95066965e-01 -6.69108629e-01 4.48122829e-01 -1.64651084e+00 -1.21118522e+00 2.02341750e-01 1.96783698e+00 7.06169605e-01 3.30938935e-01 -1.95486799e-01 9.39276636e-01 8.72534692e-01 6.87245131e-01 -3.43540967e-01 -1.71628669e-01 -4.75039072e-02 2.02660430e-02 4.62135106e-01 3.14691395e-01 -1.22133660e+00 7.63896525e-01 6.11378098e+00 1.01812065e+00 -1.50391150e+00 2.15246193e-02 3.30383033e-01 2.90359765e-01 8.63865539e-02 1.84616596e-02 -3.90951365e-01 7.94163048e-01 9.12559092e-01 4.61601429e-02 3.75830770e-01 3.33868444e-01 2.68147737e-01 9.04157236e-02 -4.66219097e-01 1.13418221e+00 3.50024551e-01 -1.14760804e+00 2.14472339e-02 4.33772027e-01 5.86706996e-01 -5.27556658e-01 3.49132389e-01 -6.82116859e-03 -2.46781617e-01 -1.08745396e+00 6.06012285e-01 2.81370491e-01 9.54893112e-01 -1.01116383e+00 9.57384586e-01 5.71027696e-01 -9.12743986e-01 -3.05504471e-01 -2.76418507e-01 -9.74969491e-02 1.62058577e-01 4.11595464e-01 -6.57437742e-01 4.87250149e-01 7.22696662e-01 7.83325493e-01 -1.88952193e-01 1.05602014e+00 -3.90018404e-01 9.87533867e-01 1.01622276e-01 5.80920652e-02 3.53925526e-01 1.15212612e-01 7.29669929e-01 1.02448642e+00 3.42234105e-01 -1.61261439e-01 1.22387677e-01 3.03317815e-01 -1.93264484e-01 -6.21065646e-02 -7.44653463e-01 -3.03648412e-01 1.51354313e-01 9.10232306e-01 -6.99310601e-01 -2.93694168e-01 -2.02250257e-01 1.01005888e+00 -4.42532212e-01 1.85675204e-01 -7.71766126e-01 -1.06938207e+00 1.46061718e-01 1.22581452e-01 8.06680620e-01 -2.21195400e-01 2.12457404e-02 -8.50169003e-01 -2.22195029e-01 -1.35741758e+00 1.94233820e-01 -4.44415420e-01 -5.31719089e-01 4.07728612e-01 -2.47401491e-01 -1.34153759e+00 -3.69775444e-01 -2.76726097e-01 -5.21544099e-01 6.06481850e-01 -1.78162277e+00 -1.08346498e+00 -1.86659560e-01 5.32249093e-01 4.50885266e-01 -5.24475157e-01 7.13385701e-01 4.82625037e-01 -4.71359193e-01 6.14572167e-01 4.32304353e-01 5.69281280e-01 5.50111711e-01 -6.34652615e-01 3.25871825e-01 8.96236539e-01 -1.60026066e-02 3.39237571e-01 7.63890982e-01 -5.33770084e-01 -1.07187021e+00 -8.31658244e-01 7.81362891e-01 4.32734519e-01 4.69621599e-01 -2.80375808e-01 -8.41918826e-01 6.71201289e-01 6.78817093e-01 -4.00525630e-01 6.61864042e-01 -8.29564273e-01 -3.16260666e-01 -2.17882022e-01 -1.04834270e+00 3.41329783e-01 4.03824776e-01 -3.83198500e-01 -4.09600943e-01 -5.75400470e-03 7.66244113e-01 -3.69535595e-01 -4.90407139e-01 2.07871035e-01 7.06048191e-01 -1.09796631e+00 9.13267374e-01 -7.72313923e-02 5.09942532e-01 -4.29095834e-01 -4.36674915e-02 -1.20007432e+00 -1.75344855e-01 -9.06222463e-01 -2.55440742e-01 9.20732081e-01 -1.76210195e-01 -6.56841278e-01 7.99408734e-01 -4.25671041e-01 1.23950623e-01 -4.39213634e-01 -1.02149451e+00 -7.24976838e-01 -1.78066090e-01 -2.33878046e-01 6.11154675e-01 7.75108337e-01 -1.35645747e-01 2.49913871e-01 -1.01414466e+00 1.75234020e-01 7.74822533e-01 9.42806154e-03 8.17811131e-01 -1.00669622e+00 -3.78880262e-01 -1.58616930e-01 -7.88384497e-01 -1.18310440e+00 1.41794473e-01 -6.96528673e-01 1.67933881e-01 -1.20697582e+00 -7.15283304e-02 -6.73547462e-02 -3.61367762e-01 3.75889003e-01 2.53497124e-01 6.57212734e-01 9.89072993e-02 3.34940732e-01 -4.45659637e-01 4.88384038e-01 1.31977332e+00 -4.74454850e-01 2.17116810e-02 1.54845923e-01 -7.78403044e-01 8.60137939e-01 9.43095744e-01 -7.02383757e-01 -2.78318435e-01 -4.52376485e-01 2.27725673e-02 2.06218749e-01 4.71760571e-01 -1.32531631e+00 3.04594308e-01 3.11538130e-01 2.07804859e-01 -4.09980416e-01 3.62111211e-01 -1.12982047e+00 6.72643557e-02 1.07722342e+00 -2.35679060e-01 -3.79587889e-01 6.68301526e-03 8.11661482e-01 -6.09955966e-01 -2.34068453e-01 9.66134369e-01 -2.36598805e-01 -9.04770613e-01 1.53976336e-01 -6.03894889e-01 -3.91113043e-01 1.09740984e+00 -4.75380212e-01 2.19159439e-01 -8.23777258e-01 -5.80994010e-01 -7.28447661e-02 1.55091003e-01 3.74116063e-01 1.08351159e+00 -1.24899399e+00 -5.93376279e-01 5.11148214e-01 -1.04999915e-01 -2.32678100e-01 3.62839460e-01 6.60635650e-01 -7.35347867e-01 5.42209148e-01 -1.73314065e-01 -4.98206586e-01 -1.71564567e+00 6.30849719e-01 1.42535970e-01 -3.93137813e-01 -7.16861665e-01 6.37467325e-01 6.76357448e-02 1.00431353e-01 4.68629479e-01 1.67718709e-01 -3.99872750e-01 -1.05919980e-01 8.37374091e-01 2.98659056e-01 -3.15292150e-01 -9.16898966e-01 -2.87998300e-02 6.46303535e-01 -3.32640111e-01 -1.53612912e-01 1.42869210e+00 -4.64655340e-01 -3.25164735e-01 2.79802084e-01 1.47400367e+00 -1.23178018e-02 -9.14481699e-01 -3.11487168e-01 -2.76095599e-01 -6.14276767e-01 2.31506035e-01 -3.40880215e-01 -1.46226883e+00 1.23968291e+00 8.18453193e-01 5.92531323e-01 1.35158944e+00 -4.97477561e-01 1.49840117e+00 2.27351680e-01 4.69572067e-01 -8.92094672e-01 4.29093421e-01 4.76231337e-01 5.50969005e-01 -1.13348949e+00 -1.08659200e-01 -1.94422752e-01 -3.50332052e-01 1.10019732e+00 7.22224638e-02 -1.92823380e-01 7.72026002e-01 5.94586283e-02 1.66193679e-01 -2.74370108e-02 -3.08748692e-01 1.91747800e-01 -7.49749243e-02 5.92999697e-01 6.62134141e-02 -2.95738280e-01 -1.56783924e-01 2.22096652e-01 -2.17956290e-01 1.22354880e-01 6.46358252e-01 1.07795012e+00 -7.28460252e-01 -1.21606040e+00 -7.42175639e-01 -3.69135812e-02 -9.95223284e-01 6.32971227e-02 -1.82693779e-01 7.18633950e-01 3.42786014e-01 1.11555433e+00 -7.26526454e-02 -8.01713526e-01 -8.46921578e-02 -2.53085107e-01 2.43607059e-01 -3.17665309e-01 -2.55644977e-01 3.23710322e-01 -4.42101508e-01 -2.32019156e-01 -6.06090844e-01 -2.57534444e-01 -1.11339509e+00 -6.80383444e-01 -5.39078295e-01 3.32990766e-01 5.83518684e-01 9.44048345e-01 1.58271715e-01 7.09837019e-01 1.00082827e+00 -7.60591149e-01 -3.70564252e-01 -7.42156327e-01 -8.96735311e-01 1.33063689e-01 8.30868602e-01 -1.62455201e-01 -5.39065659e-01 2.33931363e-01]
[4.297200679779053, 8.060617446899414]
853d05d5-fbb7-41cc-b086-8ed5d1395753
cptam-constituency-parse-tree-aggregation
2201.07905
null
https://arxiv.org/abs/2201.07905v2
https://arxiv.org/pdf/2201.07905v2.pdf
CPTAM: Constituency Parse Tree Aggregation Method
Diverse Natural Language Processing tasks employ constituency parsing to understand the syntactic structure of a sentence according to a phrase structure grammar. Many state-of-the-art constituency parsers are proposed, but they may provide different results for the same sentences, especially for corpora outside their training domains. This paper adopts the truth discovery idea to aggregate constituency parse trees from different parsers by estimating their reliability in the absence of ground truth. Our goal is to consistently obtain high-quality aggregated constituency parse trees. We formulate the constituency parse tree aggregation problem in two steps, structure aggregation and constituent label aggregation. Specifically, we propose the first truth discovery solution for tree structures by minimizing the weighted sum of Robinson-Foulds (RF) distances, a classic symmetric distance metric between two trees. Extensive experiments are conducted on benchmark datasets in different languages and domains. The experimental results show that our method, CPTAM, outperforms the state-of-the-art aggregation baselines. We also demonstrate that the weights estimated by CPTAM can adequately evaluate constituency parsers in the absence of ground truth.
['Qi Li', 'Oliver Eulenstein', 'Alexey Markin', 'Nasim Sabetpour', 'Adithya Kulkarni']
2022-01-19
null
null
null
null
['constituency-parsing']
['natural-language-processing']
[ 1.92861184e-01 6.51081204e-01 -2.71187257e-02 -8.29396129e-01 -1.70232022e+00 -1.04451680e+00 3.06936920e-01 5.08942246e-01 -8.51860195e-02 8.69974017e-01 5.49932182e-01 -4.75305051e-01 3.79539460e-01 -8.81654263e-01 -6.72327101e-01 -3.21860045e-01 2.45879158e-01 4.36046332e-01 3.22993547e-01 -1.66315049e-01 1.95316166e-01 -1.08773351e-01 -1.37142086e+00 6.66527510e-01 9.76508737e-01 8.98313344e-01 2.84628421e-01 6.51624262e-01 -8.53792489e-01 8.98865461e-01 -6.79743707e-01 -9.67354059e-01 1.09571308e-01 -1.48678944e-01 -1.25963390e+00 -7.18639269e-02 7.68998146e-01 1.23905517e-01 4.87326920e-01 1.34895837e+00 5.87461004e-03 -4.00820762e-01 2.86482006e-01 -8.32796752e-01 -5.67486405e-01 1.05382180e+00 -3.67927909e-01 4.16628808e-01 5.36268592e-01 -2.39857614e-01 1.97328460e+00 -8.39540660e-01 5.63191593e-01 1.63909006e+00 5.62778056e-01 3.92662764e-01 -1.32606483e+00 -4.81141806e-01 6.47571683e-01 -3.29415172e-01 -7.70215750e-01 -3.79728019e-01 6.80023551e-01 -3.52781922e-01 1.14546096e+00 2.05238298e-01 -5.50674349e-02 9.35016870e-01 4.66947019e-01 8.81951630e-01 1.33190691e+00 -6.44873977e-01 2.13020638e-01 -2.57130265e-01 9.59422648e-01 8.51464868e-01 5.84856749e-01 -1.94455415e-01 -4.07049328e-01 -3.99791837e-01 8.08520690e-02 -8.33127201e-01 1.81215450e-01 2.25815654e-01 -1.10090292e+00 1.01007283e+00 -4.78383526e-03 5.31982064e-01 -3.24174017e-01 -2.31809113e-02 4.43670809e-01 1.54082045e-01 8.97301078e-01 4.83197391e-01 -8.11922848e-01 9.56641138e-02 -6.17373765e-01 4.54043388e-01 9.03462112e-01 1.18513596e+00 7.10034430e-01 -4.24021930e-01 -3.58004332e-01 4.55624759e-01 3.96232337e-01 2.96172678e-01 8.32207873e-02 -9.69155073e-01 1.27557814e+00 7.01695502e-01 -6.32968172e-02 -7.65062153e-01 -4.48032558e-01 -2.57842660e-01 -4.34931487e-01 -2.51342833e-01 4.82841551e-01 -2.85281301e-01 -6.75042033e-01 2.04085135e+00 6.74694657e-01 -4.81043458e-02 4.63021427e-01 5.20460844e-01 1.15650809e+00 7.37753570e-01 6.72476232e-01 -3.85444552e-01 1.81674707e+00 -7.13301420e-01 -8.05935085e-01 -6.30620360e-01 9.35526431e-01 -9.40355659e-01 1.15287662e+00 5.97090051e-02 -1.17612922e+00 -4.26151425e-01 -9.42460954e-01 -4.31256145e-01 4.27881256e-02 1.33086473e-01 8.05397391e-01 7.02253819e-01 -7.80190706e-01 5.70697606e-01 -8.28381658e-01 1.06658213e-01 3.50958370e-02 1.69327721e-01 -4.36389863e-01 1.29923001e-01 -1.18996930e+00 8.91147792e-01 5.76079905e-01 -7.45009482e-02 -3.55078727e-01 -6.08645797e-01 -1.12441313e+00 2.04836816e-01 4.42292839e-01 -6.39927506e-01 1.66145229e+00 -5.12991905e-01 -1.10714340e+00 1.15600514e+00 -5.19927025e-01 -2.57912070e-01 -1.09683350e-01 -3.18598896e-01 -1.73285544e-01 -1.93421245e-01 9.04880762e-01 3.79980028e-01 1.46277443e-01 -1.30140817e+00 -1.33988798e+00 -3.56763810e-01 3.41077298e-01 3.54936980e-02 2.42488399e-01 2.75202364e-01 1.15598775e-01 -4.76662606e-01 6.61387265e-01 -5.33499777e-01 -1.96237519e-01 -8.36833239e-01 -6.46110058e-01 -8.89670551e-01 3.49504381e-01 -6.98281705e-01 1.28727603e+00 -1.76646900e+00 -8.78014863e-02 -1.03319801e-01 3.73827636e-01 -1.29376560e-01 -3.32564920e-01 1.25887394e-01 2.60175150e-02 4.20992196e-01 -5.10787785e-01 -6.93253100e-01 1.01175524e-01 4.58340943e-01 -5.76445043e-01 1.61809325e-02 6.32975876e-01 7.03986645e-01 -1.23995185e+00 -1.01714110e+00 -2.03342482e-01 -1.88124120e-01 -2.63054788e-01 3.01720381e-01 -3.51224184e-01 3.70536804e-01 -6.80958509e-01 6.40079439e-01 7.47220814e-01 -1.27522901e-01 6.94216371e-01 -2.80462027e-01 -2.69789040e-01 1.21014559e+00 -1.04601610e+00 1.67084002e+00 -5.77916741e-01 2.15600923e-01 1.08984210e-01 -8.48660886e-01 1.02170682e+00 3.66153479e-01 9.52164978e-02 -4.56784725e-01 1.62061110e-01 4.42255497e-01 2.76062526e-02 -2.19132379e-01 5.93823314e-01 -4.29881722e-01 -8.61759245e-01 4.23591852e-01 2.61722296e-01 -3.58313113e-01 7.43259788e-01 3.30862403e-01 1.34544885e+00 -4.21836898e-02 5.55863202e-01 -7.42221892e-01 4.89305079e-01 1.90429866e-01 1.16374516e+00 7.43065178e-01 -1.86974347e-01 3.91755283e-01 7.45330453e-01 -5.76962471e-01 -9.05458331e-01 -1.29815865e+00 -3.37528169e-01 1.25886643e+00 7.05293193e-02 -6.29615486e-01 -1.01108599e+00 -1.19244766e+00 -5.29746473e-01 9.94423151e-01 -4.20137614e-01 5.25342286e-01 -1.13710988e+00 -8.80215108e-01 4.75964099e-01 5.84546208e-01 4.37732935e-01 -1.17252290e+00 -4.84381974e-01 6.49333537e-01 -8.23724687e-01 -1.65650487e+00 -2.26709321e-01 2.96023518e-01 -9.04946148e-01 -1.25951242e+00 6.97022155e-02 -1.03186369e+00 4.45452839e-01 -3.12196203e-02 1.89794838e+00 5.61484098e-02 2.27031067e-01 2.66803559e-02 -6.23894334e-01 -3.64676476e-01 -1.01516664e+00 1.75825968e-01 -3.53603549e-02 -5.03933668e-01 7.85141706e-01 -4.01850760e-01 4.35564518e-02 -1.73565581e-01 -5.87825716e-01 1.09794833e-01 5.15120029e-01 6.98595226e-01 9.58833873e-01 -8.22518207e-03 4.08925921e-01 -1.36636865e+00 9.15531099e-01 -3.73536050e-01 -6.62265301e-01 7.88793564e-01 -3.22610468e-01 5.85185587e-01 6.87820613e-01 1.67579874e-01 -1.35134208e+00 1.65432274e-01 -3.41197789e-01 6.76395774e-01 -2.41462082e-01 4.16759968e-01 -5.59759676e-01 5.28279126e-01 5.72637677e-01 -2.25104436e-01 -8.61706257e-01 -4.67266202e-01 5.03198385e-01 4.15053070e-01 7.52189279e-01 -1.35801911e+00 4.23558742e-01 6.10136539e-02 5.58287837e-02 -2.60855973e-01 -1.81703854e+00 -3.83304507e-01 -8.18349063e-01 3.62539858e-01 9.62541759e-01 -8.20412457e-01 -2.29911834e-01 -1.15122952e-01 -1.93175423e+00 2.70320356e-01 -1.78903982e-01 1.50852978e-01 -4.29328948e-01 6.59826577e-01 -7.29495525e-01 -7.96591103e-01 -7.58114696e-01 -1.13806808e+00 1.53709221e+00 6.37986511e-02 -3.22197556e-01 -8.55984569e-01 4.55658615e-01 2.76530772e-01 -3.27585459e-01 4.48476732e-01 1.24452448e+00 -4.92729992e-01 -3.75790894e-01 -3.89852636e-02 -2.19644681e-01 2.60556996e-01 3.43658090e-01 -4.61898223e-02 -1.00905180e+00 2.35997200e-01 2.81348556e-01 -2.17571720e-01 8.53537917e-01 5.00176549e-01 6.21375382e-01 -5.05852461e-01 -1.73857152e-01 1.96874719e-02 1.20142758e+00 -1.37472734e-01 4.44746882e-01 2.37157941e-01 4.73670930e-01 1.00537837e+00 9.02066052e-01 -1.06139392e-01 7.31799066e-01 2.86541432e-01 3.32254410e-01 1.79848596e-01 -4.38711559e-03 -4.15213823e-01 3.75773311e-01 1.17094100e+00 1.07999593e-01 -3.09840262e-01 -8.56164277e-01 7.66731083e-01 -1.85877562e+00 -6.80777431e-01 -5.77363491e-01 1.74282634e+00 1.12416148e+00 4.36364502e-01 -2.24066511e-01 1.72408015e-01 9.63931262e-01 1.02967739e-01 -1.13108251e-02 -6.90410435e-01 -3.91729027e-01 4.62871492e-01 1.43658206e-01 7.49900877e-01 -1.38808966e+00 1.29800379e+00 6.25016260e+00 5.45042872e-01 -3.44016731e-01 3.59310567e-01 7.50335276e-01 5.17524362e-01 -4.92342561e-01 4.63730276e-01 -1.24935412e+00 2.31747806e-01 9.14045572e-01 -8.49645287e-02 -3.14905167e-01 9.41785872e-01 -1.58911034e-01 -1.02825858e-01 -1.25105309e+00 4.27571923e-01 -4.01546150e-01 -1.30559421e+00 -1.77168146e-01 -2.26757169e-01 5.89731395e-01 4.54517603e-02 -3.78606617e-01 3.27150404e-01 9.22330976e-01 -6.37563467e-01 9.01280403e-01 -1.24214776e-01 6.01440132e-01 -3.29436302e-01 8.82964373e-01 3.85189831e-01 -1.75464916e+00 1.56521156e-01 -5.30621350e-01 -1.43293411e-01 7.18878746e-01 1.06193590e+00 -5.82316041e-01 5.49540222e-01 6.03760600e-01 1.14025153e-01 -3.18307936e-01 3.69513005e-01 -8.62756789e-01 8.09613168e-01 -3.47788870e-01 -1.36011809e-01 3.14496726e-01 -6.86882958e-02 6.35558546e-01 1.39877963e+00 1.06168702e-01 4.69353974e-01 4.02726948e-01 9.68138635e-01 -2.03771397e-01 2.98258960e-01 -3.55708897e-01 1.39875650e-01 6.99961483e-01 1.36539245e+00 -7.02477157e-01 -5.71450293e-01 -4.74415958e-01 4.66750830e-01 7.60399342e-01 -2.15524539e-01 -5.65798402e-01 -1.71884429e-02 6.61879718e-01 -1.47697002e-01 1.44625846e-02 -1.82941064e-01 -6.99976325e-01 -1.16856027e+00 6.00432217e-01 -7.96174467e-01 6.29204810e-01 -2.65496463e-01 -1.60499930e+00 1.05394351e+00 1.15727678e-01 -8.40787470e-01 -2.43309155e-01 -6.97610855e-01 -7.67276883e-01 8.66756678e-01 -1.55872786e+00 -1.13530207e+00 1.44872844e-01 -1.11284219e-01 8.52938890e-01 2.86120828e-02 1.14692879e+00 -1.05943128e-01 -4.58382398e-01 3.69270146e-01 -4.89532679e-01 3.02233219e-01 1.71099111e-01 -1.64730489e+00 1.20277238e+00 1.32369232e+00 4.76879686e-01 5.72831511e-01 8.42713773e-01 -9.09909129e-01 -8.49211454e-01 -1.19138527e+00 1.72292471e+00 -8.57689559e-01 6.90233111e-01 -5.43446243e-01 -9.20568526e-01 6.70226693e-01 8.54584873e-02 -5.46147451e-02 6.51122570e-01 6.77845478e-01 -5.79049945e-01 2.45978206e-01 -1.19593143e+00 1.68030903e-01 1.07946479e+00 -1.85056865e-01 -1.22962832e+00 5.34703434e-01 1.12829041e+00 -4.97511357e-01 -8.32422256e-01 5.38071334e-01 1.80478334e-01 -7.95710564e-01 5.43449700e-01 -8.66992652e-01 7.48621345e-01 -2.43959114e-01 -5.65998137e-01 -1.10883939e+00 -4.92219687e-01 -3.50500852e-01 -7.34877288e-02 1.63374400e+00 9.06558335e-01 -3.20121378e-01 6.96824849e-01 9.01246250e-01 -3.73084605e-01 -4.94740456e-01 -1.21320140e+00 -7.26039946e-01 4.58316982e-01 -6.51695430e-01 8.53414416e-01 6.24606669e-01 4.67051342e-02 1.03403735e+00 1.72575966e-01 4.33100581e-01 9.32493329e-01 4.85421777e-01 4.09639180e-01 -1.51373851e+00 -2.78370619e-01 -8.99302736e-02 -2.87261251e-02 -9.43208098e-01 7.35654831e-01 -8.88679624e-01 4.65908080e-01 -1.73834085e+00 2.27634206e-01 -6.89930558e-01 1.16436422e-01 5.91610849e-01 -7.23449409e-01 -2.87257612e-01 1.70295104e-01 -1.26321157e-02 -6.00537658e-01 3.00226569e-01 9.94202375e-01 -2.35261336e-01 -4.39381227e-02 -1.13621668e-03 -1.10976338e+00 9.79499519e-01 8.45103085e-01 -9.57004607e-01 -1.96430132e-01 -7.94828534e-01 3.29632521e-01 1.29411882e-02 -1.32892027e-01 -7.27094233e-01 -7.04574287e-02 -1.92715585e-01 -2.39193425e-01 -6.02303684e-01 -1.31259924e-02 -4.44672108e-01 -3.33426774e-01 1.47371933e-01 -2.75134325e-01 4.78471726e-01 1.26093894e-01 3.91582340e-01 -2.68267542e-01 -6.03451550e-01 4.30659562e-01 -4.47224259e-01 -5.56760252e-01 -7.33957347e-03 -7.21499696e-02 6.88690901e-01 5.66981614e-01 2.64097959e-01 -6.89118683e-01 1.97730333e-01 -5.03564537e-01 1.75779104e-01 -1.20241277e-01 2.59024203e-01 3.07540447e-01 -1.14707291e+00 -1.22721314e+00 -2.70069480e-01 1.01898573e-01 2.64622152e-01 -9.22880247e-02 1.51115820e-01 -2.25131899e-01 6.31222248e-01 3.32961470e-01 -5.22587359e-01 -1.24790311e+00 2.96686798e-01 -1.77559163e-02 -1.05929327e+00 -5.47770560e-01 9.80464280e-01 5.92137933e-01 -5.90899348e-01 -2.32107788e-01 -9.49196815e-01 -1.81524977e-01 -3.04492116e-01 5.62586308e-01 -1.01465337e-01 3.47699195e-01 -7.71149993e-01 -5.28890491e-01 5.98287702e-01 -1.93564609e-01 -9.18715224e-02 1.03926337e+00 -1.03765674e-01 -3.68091017e-01 1.57369882e-01 8.43890786e-01 1.33346468e-01 -9.01935399e-01 -4.33016837e-01 5.95528722e-01 -2.48713598e-01 -8.51680934e-02 -7.08478570e-01 -6.33818626e-01 7.53717899e-01 1.19343370e-01 4.89875287e-01 9.51764524e-01 4.47021425e-01 1.06934869e+00 3.23776811e-01 5.95821977e-01 -1.14582086e+00 -4.49380577e-01 8.39754343e-01 7.16779113e-01 -1.41280949e+00 -1.23358481e-01 -1.10361767e+00 -6.12020969e-01 1.03964078e+00 5.78191340e-01 -1.18166178e-01 4.09968346e-01 6.07188106e-01 1.53044611e-01 -2.63893157e-01 -9.80565548e-01 -3.39796722e-01 2.04183891e-01 4.94675487e-01 1.03150296e+00 5.04870236e-01 -7.99987316e-01 1.24348235e+00 -7.15540171e-01 -7.47084677e-01 4.54287499e-01 7.96754718e-01 -8.05460095e-01 -1.65847516e+00 -2.56145298e-01 2.93552697e-01 -8.40054452e-01 -5.68125129e-01 -5.64906478e-01 4.69841242e-01 1.19544171e-01 1.50184381e+00 -7.03691021e-02 -1.79731280e-01 4.83505040e-01 1.46368518e-02 5.55133164e-01 -1.19859123e+00 -7.25074053e-01 -1.87235937e-01 7.34020412e-01 -3.45104098e-01 -6.28358424e-01 -6.77473664e-01 -1.42767739e+00 -1.29153624e-01 -4.53849435e-01 5.71983814e-01 6.20741665e-01 1.31491697e+00 1.67366445e-01 4.11898226e-01 4.70716894e-01 -2.96641409e-01 -6.37297332e-01 -1.01163077e+00 -2.85830230e-01 2.57735580e-01 2.50421241e-02 -5.80014884e-01 -2.31357902e-01 6.22600764e-02]
[10.354726791381836, 9.679966926574707]
093d869a-0009-4f32-95c6-d22b328627e0
approximal-operator-with-application-to-audio
2005.01437
null
https://arxiv.org/abs/2005.01437v3
https://arxiv.org/pdf/2005.01437v3.pdf
Approximal operator with application to audio inpainting
In their recent evaluation of time-frequency representations and structured sparsity approaches to audio inpainting, Lieb and Stark (2018) have used a particular mapping as a proximal operator. This operator serves as the fundamental part of an iterative numerical solver. However, their mapping is improperly justified. The present article proves that their mapping is indeed a proximal operator, and also derives its proper counterpart. Furthermore, it is rationalized that Lieb and Stark's operator can be understood as an approximation of the proper mapping. Surprisingly, in most cases, such an approximation (referred to as the approximal operator) is shown to provide even better numerical results in audio inpainting compared to its proper counterpart, while being computationally much more effective.
['Pavel Rajmic', 'Ondřej Mokrý']
2020-05-04
null
null
null
null
['audio-inpainting']
['audio']
[ 1.64421976e-01 3.61535490e-01 -2.12093949e-01 2.84777641e-01 -8.01798344e-01 -4.49281156e-01 3.77971560e-01 -6.37234449e-02 -6.94682449e-02 7.78963327e-01 4.55728829e-01 -1.65118888e-01 -2.58017927e-01 -5.69296300e-01 -7.98801422e-01 -6.03345215e-01 3.51440683e-02 -3.01263221e-02 -3.02190185e-01 -5.25527656e-01 1.73469841e-01 2.91017354e-01 -1.36521995e+00 -3.15224491e-02 9.13419366e-01 1.02886593e+00 -7.96619579e-02 4.16056186e-01 9.60861966e-02 9.09946501e-01 -5.27589560e-01 -5.07214010e-01 5.42562783e-01 -7.24873066e-01 -9.27225411e-01 6.74865916e-02 4.42276806e-01 -1.35768220e-01 -2.36145705e-01 1.09378350e+00 3.47857296e-01 2.49631286e-01 2.96393752e-01 -1.07091880e+00 -5.25156021e-01 6.30435824e-01 -4.41685975e-01 6.27170131e-02 6.60228431e-01 -2.19396278e-01 1.05071449e+00 -1.18597758e+00 5.66852808e-01 7.98292041e-01 1.17774296e+00 2.59693742e-01 -1.32131386e+00 -3.90014023e-01 -2.81364501e-01 1.59067869e-01 -1.59231627e+00 -3.92956257e-01 1.00377929e+00 -1.87678322e-01 3.26957405e-01 6.05234683e-01 1.02553511e+00 9.16671634e-01 -1.22017944e-02 9.92373049e-01 1.07755697e+00 -6.56621099e-01 2.13135347e-01 -3.12738940e-02 -1.98369294e-01 4.22782093e-01 6.86234012e-02 2.63463885e-01 -8.24049950e-01 -4.68446374e-01 8.47128451e-01 -3.01736116e-01 -7.30527639e-01 -1.89202800e-01 -8.97338986e-01 9.02801931e-01 2.54062444e-01 3.54672849e-01 -5.50251007e-01 1.49372622e-01 3.78408283e-01 4.81344759e-01 8.56857955e-01 5.07385373e-01 1.16430067e-01 -5.53877413e-01 -1.22355831e+00 6.14115238e-01 9.02363181e-01 6.95065498e-01 4.79222745e-01 4.92421359e-01 -3.01322527e-03 7.37117052e-01 -6.10032193e-02 1.08775951e-01 3.46291095e-01 -1.36388683e+00 1.53058887e-01 -1.16700027e-02 2.54258275e-01 -1.09317780e+00 -2.24983618e-01 -8.67683887e-01 -9.15132642e-01 4.27461527e-02 7.14631736e-01 -7.77770113e-03 -6.10351041e-02 1.87636781e+00 2.92119592e-01 7.02515483e-01 -2.83577532e-01 1.20237720e+00 1.72734469e-01 7.51149714e-01 -2.19524130e-01 -5.11078119e-01 1.00127363e+00 -6.95839286e-01 -1.02753937e+00 7.98224136e-02 2.41811097e-01 -9.78448570e-01 1.04766810e+00 7.41297364e-01 -1.62499022e+00 -4.17486221e-01 -1.03058839e+00 -1.44613862e-01 2.74836779e-01 -2.57942706e-01 8.87652397e-01 5.05836129e-01 -1.21480048e+00 8.99129152e-01 -4.21363443e-01 -2.38272548e-01 1.09461561e-01 -1.03101261e-01 -1.37818828e-01 2.82578856e-01 -1.21392226e+00 9.25484896e-01 -1.98984131e-01 7.34664947e-02 -5.56222141e-01 -1.16817188e+00 -5.15338182e-01 1.01165690e-01 3.39399904e-01 -8.84358466e-01 1.26927733e+00 -1.07377362e+00 -1.69665968e+00 8.39433074e-01 -1.62325606e-01 -9.54251707e-01 7.09468722e-01 -4.21171606e-01 -7.00691864e-02 4.18769896e-01 -3.46444771e-02 1.65035710e-01 1.42750919e+00 -9.40556586e-01 -1.73996523e-01 -1.01367146e-01 2.82195717e-01 4.79123220e-02 -2.73452252e-01 -1.41534477e-01 8.53420198e-02 -1.42509019e+00 3.83088648e-01 -7.19566584e-01 -1.13441415e-01 1.14691257e-01 -2.24261954e-01 -2.85510309e-02 4.77537334e-01 -9.98928010e-01 1.47697544e+00 -2.44147134e+00 4.41203475e-01 1.42847747e-01 2.05152184e-01 1.94101501e-02 8.42904150e-02 7.78821170e-01 -5.03233552e-01 -4.02794071e-02 -4.08447683e-01 -4.26044911e-01 6.40221089e-02 -4.81377952e-02 -8.11174929e-01 7.84625530e-01 1.24051116e-01 7.40635991e-01 -8.30252588e-01 -1.31417885e-01 1.25466302e-01 7.16618359e-01 -6.57930076e-01 -8.96145999e-02 6.11734949e-02 5.62203944e-01 -5.58592901e-02 5.79754949e-01 5.80163956e-01 2.73975022e-02 -9.05633718e-02 -3.23462427e-01 -3.05032790e-01 2.59676337e-01 -1.27991199e+00 1.77592564e+00 -5.71977794e-01 8.11339438e-01 6.20784402e-01 -1.41284263e+00 7.34664023e-01 6.77978098e-01 7.98461854e-01 -4.86089408e-01 -1.36755453e-02 4.81139064e-01 -4.70459342e-01 -3.23889852e-01 7.01045990e-01 -5.30326188e-01 2.19206154e-01 4.72456038e-01 -2.58706212e-01 -2.61199474e-01 1.43987104e-01 1.85587540e-01 9.96328890e-01 3.14616233e-01 4.33012128e-01 -6.16358757e-01 5.85875690e-01 -9.27135423e-02 3.77642602e-01 6.11073494e-01 -6.92283437e-02 9.13893521e-01 4.40177113e-01 -2.56289333e-01 -1.20640755e+00 -9.73255932e-01 -3.41711968e-01 7.10629463e-01 -2.40074307e-01 -6.47807002e-01 -8.48190010e-01 -4.41838354e-02 -6.03763796e-02 4.62631434e-01 -5.46682119e-01 -2.09564567e-01 -6.62003994e-01 -2.72521108e-01 7.91090906e-01 3.44003677e-01 2.78910160e-01 -6.36856973e-01 -7.67102242e-01 2.88800061e-01 -4.82962281e-01 -7.65609801e-01 -6.15448296e-01 2.19828695e-01 -9.97806609e-01 -9.15061295e-01 -1.08722067e+00 -3.67500901e-01 1.99278176e-01 2.06165344e-01 1.02223885e+00 3.22141722e-02 5.64760305e-02 4.73931432e-01 -6.12899721e-01 -2.25158334e-01 -4.97256398e-01 -2.46130005e-01 -2.61103064e-02 3.62201810e-01 -2.70483971e-01 -1.06608999e+00 -4.34016377e-01 3.95686068e-02 -1.13342941e+00 -6.62427470e-02 1.36912614e-01 1.03622031e+00 4.87541646e-01 -2.20652759e-01 5.96813321e-01 -6.18933976e-01 7.69680977e-01 -3.26808006e-01 -5.97743273e-01 -1.62449002e-01 -6.05490565e-01 -4.18653805e-03 9.25942421e-01 -3.91031444e-01 -7.67670691e-01 -1.82385176e-01 -2.60587156e-01 -6.66071117e-01 3.58548164e-01 7.90013134e-01 2.85834461e-01 -3.29726368e-01 5.85590899e-01 4.82301503e-01 9.54985544e-02 -7.74816930e-01 2.97841311e-01 3.09613556e-01 5.21135569e-01 -7.39918113e-01 9.97602820e-01 7.12971926e-01 1.90236092e-01 -9.26006019e-01 -1.06010139e+00 -2.15124503e-01 -5.48503995e-02 -2.19168752e-01 3.30354035e-01 -7.57936418e-01 -6.05652452e-01 1.33931100e-01 -9.14961040e-01 -1.96742862e-01 -9.48100328e-01 4.83343542e-01 -8.69402885e-01 5.74551284e-01 -6.85579002e-01 -9.83864903e-01 2.83025838e-02 -8.33763301e-01 9.85232353e-01 -2.89401561e-01 -4.96250123e-01 -9.83348846e-01 2.02150848e-02 6.62820190e-02 6.49183631e-01 4.49384034e-01 7.24848688e-01 -5.97663708e-02 -1.23335972e-01 -2.54853159e-01 1.32594839e-01 3.87003541e-01 -8.34985897e-02 9.67951342e-02 -9.63894844e-01 -1.85937837e-01 7.73068607e-01 4.09485139e-02 5.71707785e-01 5.75479686e-01 9.00831401e-01 -5.37894130e-01 3.21506709e-01 9.72523749e-01 1.40862346e+00 -1.89017028e-01 5.21339774e-01 2.31789410e-01 1.81343168e-01 6.23021245e-01 5.34523368e-01 8.04733992e-01 -8.28573033e-02 9.56575334e-01 2.18765050e-01 1.27863333e-01 -2.65933841e-01 -4.21222776e-01 5.45555413e-01 1.14981246e+00 -3.41937006e-01 3.59838665e-01 -4.22910899e-01 3.15600425e-01 -1.82164586e+00 -1.16119838e+00 -2.07306430e-01 2.23913527e+00 1.00585365e+00 -1.59996182e-01 4.07979012e-01 7.33231425e-01 7.06714153e-01 1.40935346e-01 -2.17127442e-01 -4.65882689e-01 -3.17226380e-01 7.95596957e-01 3.81469369e-01 6.84427977e-01 -7.08969235e-01 4.27823305e-01 7.40579605e+00 1.08838546e+00 -1.02548587e+00 4.36605930e-01 1.84155822e-01 -9.45313349e-02 -5.15079856e-01 2.65344799e-01 -1.10674836e-01 5.22788525e-01 9.06969190e-01 -4.87644166e-01 9.01038647e-01 5.96245587e-01 4.62272137e-01 -9.46709663e-02 -8.94844532e-01 1.10674834e+00 1.71510479e-03 -1.50849307e+00 -1.73976481e-01 9.66837034e-02 6.13795042e-01 -6.99329734e-01 1.49536416e-01 5.79942986e-02 -5.23987710e-01 -1.11492074e+00 1.22682178e+00 6.23151362e-01 4.97916162e-01 -9.50031161e-01 4.38355207e-01 3.68828684e-01 -1.21417475e+00 1.16931081e-01 -1.02200031e-01 -5.88650286e-01 3.56967300e-01 9.20617759e-01 -2.50231981e-01 7.75792599e-01 3.62912148e-01 8.36055577e-01 -1.10660374e-01 1.04759181e+00 -7.44845495e-02 9.25433040e-01 -4.02255535e-01 4.89220649e-01 2.66453952e-01 -6.75212681e-01 1.02571726e+00 6.78999066e-01 5.63048720e-01 9.91827697e-02 -9.38394144e-02 9.48378921e-01 1.24383859e-01 2.14981794e-01 -6.23219609e-01 -1.55523077e-01 3.72582763e-01 8.58801305e-01 -4.31870311e-01 -1.55158252e-01 -3.32612574e-01 8.97962749e-01 -3.32959229e-03 5.38771212e-01 -9.44735527e-01 -2.63359785e-01 7.07328320e-01 3.18463624e-01 1.81652144e-01 -8.10995325e-02 -6.23386562e-01 -9.97445047e-01 1.02125555e-01 -8.92540157e-01 7.33898506e-02 -8.56943130e-01 -1.07145095e+00 3.26962411e-01 -1.03364743e-01 -1.76996756e+00 -2.14795977e-01 -2.44448185e-01 -3.13901186e-01 9.03260827e-01 -1.23881638e+00 -6.83512449e-01 -6.66298941e-02 6.42164052e-01 2.01580539e-01 1.97716713e-01 8.12402606e-01 6.21163666e-01 -2.56174684e-01 4.96982843e-01 8.04872587e-02 -4.02478933e-01 4.57633108e-01 -1.04933465e+00 -1.87972203e-01 8.95203471e-01 3.15362513e-01 8.71835828e-01 1.28990495e+00 -2.45758772e-01 -1.70781624e+00 -6.13937259e-01 7.87689447e-01 1.39879221e-02 8.16090882e-01 6.83466345e-02 -8.93044412e-01 6.85495198e-01 4.38433349e-01 -1.85389757e-01 5.61563611e-01 -2.26857662e-01 -1.54863417e-01 -9.83732864e-02 -1.09908712e+00 5.60759664e-01 9.82103288e-01 -7.89600432e-01 -5.13284862e-01 5.00379503e-01 5.81199646e-01 -5.87942719e-01 -9.08394933e-01 2.08121955e-01 5.36082268e-01 -1.48752975e+00 1.04713190e+00 -2.15152413e-01 5.28909266e-01 -3.14930707e-01 -2.22930983e-01 -1.09226501e+00 -5.30837066e-02 -1.32604325e+00 -5.25283277e-01 9.96091068e-01 -1.57588124e-01 -6.83480740e-01 5.84726512e-01 8.23388994e-02 -4.08517420e-01 -1.00902259e+00 -1.37489605e+00 -1.07666409e+00 1.85939580e-01 -7.99654841e-01 3.02058220e-01 9.92683172e-01 2.81404853e-01 -5.73685020e-02 -6.77020192e-01 -1.47791103e-01 5.76674044e-01 1.09861769e-01 5.13502419e-01 -1.02446163e+00 -4.59562480e-01 -6.17804110e-01 -2.61445165e-01 -1.13676691e+00 2.72985637e-01 -9.68805492e-01 -3.50227296e-01 -1.05260611e+00 -2.95715421e-01 -4.30113822e-01 4.31310292e-03 -2.24399399e-02 1.90243363e-01 6.39078736e-01 3.01248491e-01 2.66778409e-01 2.27523208e-01 5.78671634e-01 1.37166667e+00 -1.02472240e-02 -7.79059604e-02 2.08436519e-01 -8.54541779e-01 6.87694073e-01 4.73723173e-01 -3.79736662e-01 -3.05991501e-01 -8.62231255e-02 8.03079128e-01 4.26194459e-01 7.04440355e-01 -9.18974936e-01 1.45308092e-01 9.03101042e-02 -2.24127755e-01 -3.28105479e-01 6.52834773e-01 -7.67525613e-01 5.76586187e-01 5.12284517e-01 -3.26339275e-01 -1.05230689e-01 6.15223572e-02 4.27085280e-01 -6.43296301e-01 -5.99771559e-01 6.33436620e-01 -7.39810988e-02 -2.50966221e-01 2.17570830e-02 -3.23825568e-01 2.48628139e-01 5.83686173e-01 -4.93804425e-01 3.31476003e-01 -9.66358781e-01 -8.61382723e-01 -5.82137167e-01 4.54318404e-01 -3.03307295e-01 5.41311681e-01 -1.52741539e+00 -7.35732496e-01 4.28615101e-02 -4.44064140e-01 -2.74184525e-01 1.76080033e-01 1.59349537e+00 -5.42500973e-01 3.20730537e-01 1.99943677e-01 -4.02510405e-01 -6.82488739e-01 3.29314530e-01 1.55117556e-01 -1.40742287e-01 -9.08293426e-01 9.53058004e-01 -7.38569275e-02 1.95395038e-01 2.81065673e-01 -3.06904256e-01 3.30180883e-01 1.71609655e-01 5.26865900e-01 6.72811270e-01 -7.55226389e-02 -5.35020709e-01 -3.29835057e-01 6.52342260e-01 7.31243134e-01 -3.53298753e-01 1.27673566e+00 -1.27257556e-01 -3.33948731e-01 7.55438328e-01 1.10129750e+00 5.99656343e-01 -1.14260077e+00 -1.51522830e-01 -7.32809603e-02 -5.44928432e-01 -7.22438321e-02 -2.96713978e-01 -1.07395375e+00 7.50357211e-01 1.00552581e-01 7.53437102e-01 1.43120635e+00 -3.72184873e-01 1.03729784e+00 -7.25535229e-02 5.80069602e-01 -9.12202179e-01 -2.08984688e-01 4.30962950e-01 1.24799204e+00 -6.48050785e-01 2.11439967e-01 -6.79267943e-01 -2.42255703e-01 1.24794257e+00 -1.17769942e-01 -4.91509914e-01 5.50673127e-01 3.98283362e-01 -2.28098318e-01 2.23857597e-01 -4.69040990e-01 -2.51510609e-02 1.21995173e-01 4.95235682e-01 5.30806422e-01 -1.90264523e-01 -8.55299771e-01 5.17071009e-01 -5.68516970e-01 1.13440365e-01 5.14789820e-01 7.23705649e-01 -1.25322714e-01 -1.01270008e+00 -6.77519023e-01 7.41120130e-02 -5.71630418e-01 -3.41222972e-01 -1.56453960e-02 6.73835099e-01 -1.64539423e-02 8.85632217e-01 -2.74490625e-01 -1.43129200e-01 1.91696405e-01 2.05598488e-01 7.10068882e-01 -3.72093469e-01 -9.15370882e-01 1.62250519e-01 -1.41015753e-01 -7.31529415e-01 -6.61203146e-01 -3.80074024e-01 -9.91222799e-01 -5.41306436e-01 -1.45062819e-01 5.24305105e-01 3.26553255e-01 1.02315974e+00 1.58855036e-01 4.42891181e-01 7.13483691e-01 -8.44594479e-01 -8.39162588e-01 -9.89715278e-01 -8.15312386e-01 4.17579740e-01 4.52816248e-01 -5.91581881e-01 -6.11180425e-01 -1.37501759e-02]
[15.491037368774414, 5.563834190368652]
a333097d-6384-496c-b53d-77cfc26f686e
multimodal-speech-emotion-recognition-and
1904.06022
null
http://arxiv.org/abs/1904.06022v1
http://arxiv.org/pdf/1904.06022v1.pdf
Multimodal Speech Emotion Recognition and Ambiguity Resolution
Identifying emotion from speech is a non-trivial task pertaining to the ambiguous definition of emotion itself. In this work, we adopt a feature-engineering based approach to tackle the task of speech emotion recognition. Formalizing our problem as a multi-class classification problem, we compare the performance of two categories of models. For both, we extract eight hand-crafted features from the audio signal. In the first approach, the extracted features are used to train six traditional machine learning classifiers, whereas the second approach is based on deep learning wherein a baseline feed-forward neural network and an LSTM-based classifier are trained over the same features. In order to resolve ambiguity in communication, we also include features from the text domain. We report accuracy, f-score, precision, and recall for the different experiment settings we evaluated our models in. Overall, we show that lighter machine learning based models trained over a few hand-crafted features are able to achieve performance comparable to the current deep learning based state-of-the-art method for emotion recognition.
['Gaurav Sahu']
2019-04-12
null
null
null
null
['multimodal-emotion-recognition', 'multimodal-emotion-recognition']
['computer-vision', 'speech']
[ 4.38622475e-01 1.18888542e-01 3.70361894e-01 -7.23325491e-01 -1.15945315e+00 -4.32635367e-01 5.79251528e-01 2.24313498e-01 -5.67058086e-01 4.20741290e-01 1.44834459e-01 -1.90050781e-01 -1.25508875e-01 -3.42356026e-01 -4.21377271e-01 -6.32464886e-01 9.78438109e-02 1.69671819e-01 -2.15696856e-01 -3.01424503e-01 2.22003207e-01 3.51161093e-01 -1.79155302e+00 6.49840593e-01 4.73441929e-01 1.86165702e+00 -3.39482874e-01 7.14374065e-01 -3.55384529e-01 7.73169696e-01 -9.02451932e-01 -2.67408937e-01 -1.09070107e-01 -3.20482701e-01 -8.69115233e-01 5.16400039e-02 1.05756700e-01 1.18789010e-01 8.05214867e-02 6.74979150e-01 4.74280089e-01 1.15654737e-01 6.24355912e-01 -1.30290091e+00 -1.15435474e-01 4.67315525e-01 -6.99606761e-02 1.04789540e-01 5.70732057e-01 -4.98798907e-01 1.00870895e+00 -1.10984790e+00 2.39965051e-01 1.04402983e+00 6.94539309e-01 5.27427554e-01 -9.52975631e-01 -5.62827289e-01 1.50269017e-01 2.60399610e-01 -1.21958995e+00 -8.26946378e-01 8.94704223e-01 -5.05598128e-01 1.17075646e+00 3.22292358e-01 3.78963977e-01 1.32634902e+00 5.49138151e-03 6.68268979e-01 1.18487847e+00 -7.64363945e-01 4.85585839e-01 3.69741917e-01 1.72094002e-01 5.64070225e-01 -4.57216144e-01 1.04799166e-01 -7.26734817e-01 -3.22221994e-01 2.76999455e-02 -3.03746521e-01 -2.81985402e-01 -1.28505051e-01 -8.91994417e-01 8.75902951e-01 1.73477479e-03 8.67095530e-01 -6.01767719e-01 7.29233027e-02 7.06123412e-01 6.40683472e-01 7.75445819e-01 4.47319329e-01 -9.07095909e-01 -5.93136907e-01 -9.25245881e-01 -1.43619224e-01 1.11206114e+00 4.39868391e-01 5.55809379e-01 2.46309981e-01 -7.06742182e-02 9.62720096e-01 1.70492694e-01 -1.00472532e-01 7.95475066e-01 -5.84805071e-01 2.40752459e-01 4.33920652e-01 -2.39283014e-02 -9.90692735e-01 -4.97876465e-01 -4.43379760e-01 -5.49563646e-01 9.26331058e-02 2.67164618e-01 -5.27473509e-01 -7.28179276e-01 1.58922338e+00 5.51071987e-02 1.25727236e-01 4.99743730e-01 6.33742690e-01 8.22377443e-01 6.72533751e-01 -9.19173956e-02 -2.84749210e-01 1.29738677e+00 -9.68906581e-01 -7.76511133e-01 -1.36727139e-01 4.94767219e-01 -7.43803382e-01 7.84270167e-01 8.35400879e-01 -7.13114381e-01 -3.91454399e-01 -1.15951967e+00 3.61504585e-01 -7.91983783e-01 4.94190454e-01 3.56360435e-01 8.18547964e-01 -9.53074217e-01 6.32269323e-01 -4.47360575e-01 -2.39328444e-01 1.20966658e-02 4.86862004e-01 -5.94138741e-01 4.44672763e-01 -1.36123538e+00 9.50156391e-01 2.58770347e-01 2.12765411e-01 -5.20848513e-01 -1.78392753e-01 -8.52454364e-01 4.37450886e-01 3.09863120e-01 -2.53918201e-01 1.48943019e+00 -1.58365095e+00 -2.06316900e+00 8.37398648e-01 -2.33791187e-01 -4.69497740e-01 1.20639622e-01 -2.11368740e-01 -7.22049475e-01 1.93917781e-01 -3.34463775e-01 1.45462900e-01 1.23988438e+00 -1.24624777e+00 -6.97272420e-01 -2.24777594e-01 -2.64682531e-01 -1.36391789e-01 -5.58219552e-01 3.93804282e-01 1.20011538e-01 -5.65871000e-01 -1.11409590e-01 -7.17486441e-01 8.76013264e-02 -3.36297065e-01 -1.95414335e-01 -3.52449566e-01 9.17320251e-01 -5.44070661e-01 1.17397654e+00 -2.38922954e+00 1.90224081e-01 2.60759473e-01 1.55497968e-01 3.69699806e-01 -1.91910759e-01 5.29783785e-01 -2.85966188e-01 9.59610417e-02 -9.97920856e-02 -6.22938097e-01 3.33118975e-01 1.38968512e-01 -3.70290428e-01 2.14350566e-01 4.90888208e-01 4.96056467e-01 -5.18619955e-01 -8.99172798e-02 1.74297705e-01 6.55638933e-01 -1.88950688e-01 4.53244209e-01 2.99138185e-02 3.12376291e-01 -3.00737500e-01 3.64766747e-01 2.63896585e-01 2.22742900e-01 1.66423827e-01 -1.32621378e-01 -2.15361044e-02 5.41167617e-01 -1.13949525e+00 1.55140591e+00 -9.62227821e-01 8.20305467e-01 1.38261512e-01 -1.37292755e+00 1.16718566e+00 8.99762154e-01 4.47622091e-01 -4.79730546e-01 4.70576018e-01 4.26874548e-01 -4.06718254e-02 -5.79950213e-01 4.15241241e-01 -1.94639653e-01 -3.74624521e-01 4.35225695e-01 5.87512553e-01 -2.67360136e-02 -1.82229757e-01 -3.40014666e-01 1.20146167e+00 -2.27484450e-01 3.62545758e-01 2.67166998e-02 8.25404525e-01 -4.70542967e-01 4.09694284e-01 5.23441613e-01 -1.95002720e-01 3.68360251e-01 5.77113867e-01 -5.89753330e-01 -5.61366081e-01 -5.64592361e-01 7.21156746e-02 1.44843113e+00 -4.77742136e-01 -5.89836776e-01 -9.49687779e-01 -8.60985637e-01 -3.10459524e-01 6.01561785e-01 -7.15390384e-01 -2.72237420e-01 -2.43726179e-01 -3.66866857e-01 7.90103674e-01 3.54753226e-01 1.76858559e-01 -1.07829309e+00 -8.29178631e-01 4.11244690e-01 -2.42696747e-01 -1.27403820e+00 5.82207739e-02 6.76549017e-01 -2.31130078e-01 -7.39002705e-01 -4.73121583e-01 -6.61595583e-01 1.35244042e-01 -3.23219240e-01 1.05007565e+00 1.16477301e-02 -8.61757249e-02 6.47399485e-01 -7.87403226e-01 -7.53787875e-01 -4.91871834e-01 1.27682522e-01 -3.14835571e-02 6.13207638e-01 5.47061503e-01 -5.99388182e-01 -1.46502122e-01 1.47399791e-02 -8.66336405e-01 -4.99388397e-01 6.25001431e-01 1.01163185e+00 2.33178794e-01 3.30320261e-02 7.82933176e-01 -3.09300542e-01 9.37015772e-01 -3.45226318e-01 -1.12244889e-01 2.86280960e-01 -3.14069659e-01 2.52872676e-01 6.30191624e-01 -6.70136929e-01 -7.87172139e-01 4.48107451e-01 -5.57743728e-01 -2.43792042e-01 -5.46362698e-01 6.74758255e-01 -1.20300964e-01 -1.66007027e-01 3.37108850e-01 1.77427202e-01 -1.26743048e-01 -4.81944382e-01 2.96330899e-01 1.25078058e+00 2.75756836e-01 -5.03122091e-01 1.87431619e-01 -3.30442600e-02 -3.42110038e-01 -9.05150771e-01 -8.83949995e-01 -3.96347255e-01 -5.20865917e-01 -2.98252910e-01 7.55717695e-01 -6.93658769e-01 -5.27948260e-01 5.03878176e-01 -1.38077521e+00 -5.15657216e-02 -1.99388027e-01 4.86754805e-01 -7.32808471e-01 1.14862733e-01 -4.43399489e-01 -1.13045263e+00 -5.00247359e-01 -1.04952669e+00 1.41353083e+00 -5.34220487e-02 -4.74904567e-01 -8.81628573e-01 7.64193833e-02 1.42546725e-02 5.48185706e-01 2.34195888e-01 9.42181170e-01 -1.41408813e+00 4.05298680e-01 -3.59596312e-01 9.16002393e-02 5.50892055e-01 1.17808603e-01 -1.08300932e-01 -1.42008781e+00 -7.95862377e-02 3.82754385e-01 -6.12130284e-01 8.59573066e-01 -4.27618176e-02 1.08560741e+00 -2.87056416e-01 2.62824912e-02 1.46411002e-01 1.01783919e+00 3.45028400e-01 3.39441568e-01 3.03038299e-01 1.11253917e-01 7.78874934e-01 5.65964699e-01 6.85920119e-01 1.84098959e-01 7.59883583e-01 2.88753986e-01 -1.36864141e-01 4.77474540e-01 1.18304633e-01 3.45858574e-01 6.87472999e-01 2.45599985e-01 -2.47305855e-01 -7.71628320e-01 4.17014778e-01 -1.85558784e+00 -8.15785050e-01 3.69305253e-01 1.85253417e+00 6.86865330e-01 1.47554070e-01 1.34181619e-01 7.97366798e-01 4.95238274e-01 9.20914114e-02 -4.60561961e-02 -9.91646588e-01 2.40769044e-01 5.66277862e-01 -2.36400500e-01 4.74148273e-01 -1.31942093e+00 7.55264223e-01 5.93837261e+00 6.84551477e-01 -1.67348278e+00 8.71432796e-02 4.41565335e-01 -8.29453114e-03 3.27063315e-02 -4.23505157e-01 -4.17436302e-01 2.25365013e-01 1.37604618e+00 1.40269265e-01 3.29015285e-01 8.13602567e-01 4.67277244e-02 2.49223784e-02 -1.27437186e+00 1.24425387e+00 3.87760937e-01 -8.28193903e-01 -3.31847429e-01 -1.67896375e-01 1.22219160e-01 -2.03575507e-01 3.81189957e-02 4.55520123e-01 -2.01354444e-01 -1.13750160e+00 8.79900753e-01 4.68108892e-01 5.26311278e-01 -8.19746435e-01 9.34966445e-01 3.17940593e-01 -9.25538540e-01 -2.61841774e-01 9.52240750e-02 -2.67770559e-01 9.07381698e-02 7.32029200e-01 -7.06854105e-01 6.59372389e-01 7.29802907e-01 3.20701033e-01 -2.15230599e-01 7.10506558e-01 -1.28026307e-01 5.61934412e-01 -2.58910418e-01 -1.29980996e-01 3.56608421e-01 2.41773576e-01 3.62154037e-01 1.51662791e+00 5.12632132e-01 -1.58127949e-01 1.36330023e-01 5.33272266e-01 -1.61545388e-02 2.30420902e-01 -5.15520453e-01 -2.99785733e-01 2.64688015e-01 1.50242007e+00 -4.72200781e-01 -3.49734932e-01 -3.22778374e-01 1.03002393e+00 4.10759538e-01 2.20332131e-01 -6.68635130e-01 -8.06251287e-01 6.64058924e-01 -4.96572465e-01 5.41470766e-01 -8.71650428e-02 2.50347257e-02 -1.11449814e+00 7.82167912e-02 -1.03298831e+00 1.61468357e-01 -5.77786863e-01 -1.18861806e+00 1.27796006e+00 -3.32776368e-01 -1.02828777e+00 -8.55486631e-01 -7.87222385e-01 -6.43798411e-01 7.43201137e-01 -1.37954688e+00 -9.45052803e-01 -1.54612616e-01 4.37284678e-01 6.11126363e-01 -3.49070966e-01 1.42315471e+00 3.32265377e-01 -4.41882849e-01 5.34329355e-01 7.29095489e-02 2.50261724e-01 6.76097989e-01 -1.20697165e+00 1.34747371e-01 4.72858250e-01 6.16103292e-01 2.78097272e-01 7.21096098e-01 5.08656539e-02 -1.13645351e+00 -7.62995481e-01 1.22195065e+00 -8.03164691e-02 6.96355045e-01 -5.63661754e-01 -7.68695772e-01 5.25450647e-01 2.73642451e-01 1.37863994e-01 1.10918033e+00 2.48190165e-01 -4.20592397e-01 -1.19031049e-01 -1.07011688e+00 -6.26888871e-02 3.24166059e-01 -8.15563738e-01 -8.68085206e-01 -1.52843580e-01 5.43777764e-01 -2.43399575e-01 -1.00879395e+00 3.88863832e-01 7.83127785e-01 -9.97262716e-01 5.94462335e-01 -8.23183715e-01 2.48216286e-01 1.63905993e-01 -4.46057469e-01 -1.70856142e+00 1.74487412e-01 -5.93874872e-01 -7.30781928e-02 1.34485424e+00 7.16305912e-01 -7.40402579e-01 3.74447614e-01 4.67356592e-01 -1.69265911e-01 -1.01610744e+00 -1.18923223e+00 -6.34143710e-01 -1.10384375e-02 -6.77285910e-01 5.36719441e-01 8.83972406e-01 3.01762551e-01 5.50552666e-01 -2.98780560e-01 -3.47896926e-02 -8.42294842e-02 1.21462658e-01 4.54263479e-01 -1.39142907e+00 -3.57833713e-01 -4.12336051e-01 -5.28765440e-01 -5.74588239e-01 6.84542298e-01 -4.65586513e-01 2.05418885e-01 -1.10793746e+00 -4.21054453e-01 -1.06453724e-01 -4.99464184e-01 5.82278192e-01 2.96541691e-01 3.36790048e-02 2.15157926e-01 -3.71822864e-01 -5.45434475e-01 5.84700644e-01 3.11341792e-01 -1.88648045e-01 -2.39904583e-01 1.71013668e-01 -6.62360370e-01 7.44151473e-01 9.55805719e-01 -5.80793381e-01 -9.96362269e-02 -2.03246787e-01 9.63761732e-02 2.46561728e-02 2.90036857e-01 -1.03498149e+00 1.80773944e-01 1.68931782e-01 2.49105811e-01 -1.17558062e-01 6.92475498e-01 -8.99778605e-01 -2.64484286e-01 2.47743636e-01 -4.98460084e-01 -3.35926078e-02 3.49484593e-01 2.30543330e-01 -7.38381386e-01 -3.24604452e-01 3.82453382e-01 8.51311758e-02 -6.05234802e-01 -2.27191195e-01 -7.94598639e-01 -1.91608220e-01 7.37914979e-01 6.38773888e-02 -2.12048423e-02 -6.43042922e-01 -1.05403578e+00 -3.80887330e-01 -2.35440090e-01 6.41891360e-01 7.23870814e-01 -1.09940171e+00 -4.60147113e-01 2.80490100e-01 2.20820650e-01 -6.30797207e-01 -1.32258430e-01 7.89077878e-01 1.75183013e-01 4.62364256e-01 -8.24753791e-02 -3.76713693e-01 -1.48919940e+00 4.41194803e-01 6.24468625e-01 -3.27269375e-01 -1.16291404e-01 6.65944457e-01 -3.22656304e-01 -5.25440097e-01 5.97062171e-01 -4.18630630e-01 -3.21500480e-01 3.92387360e-01 4.94280636e-01 7.68679380e-02 6.41064346e-01 -7.15347588e-01 -4.83652085e-01 3.69845569e-01 1.62620455e-01 -4.79904175e-01 1.35454857e+00 3.58260162e-02 2.82646436e-02 7.29530811e-01 1.43923783e+00 -9.09400359e-02 -5.72709560e-01 -2.05035701e-01 3.34717721e-01 -2.11211145e-02 2.08404869e-01 -1.10187101e+00 -8.09418321e-01 1.11853623e+00 6.99428618e-01 6.81746960e-01 1.38828516e+00 -1.73839390e-01 5.78039706e-01 5.84102035e-01 2.35248283e-01 -1.27890003e+00 7.61089176e-02 7.26312578e-01 1.01654136e+00 -1.17026091e+00 -5.38865685e-01 -3.16905737e-01 -6.53331339e-01 1.45369053e+00 2.77065873e-01 8.73554200e-02 6.89641654e-01 4.38254088e-01 3.32666427e-01 -1.41165704e-01 -9.67254996e-01 -3.49210948e-01 4.17232186e-01 4.22845453e-01 6.40254915e-01 -1.66020051e-01 -2.19020128e-01 1.12553632e+00 -3.10439497e-01 -1.54695781e-02 2.89247572e-01 8.99725020e-01 -4.10178035e-01 -1.30515921e+00 -2.95183122e-01 3.45895678e-01 -7.51489460e-01 5.58049977e-02 -9.60540354e-01 4.79346126e-01 9.84760467e-03 1.35492253e+00 -2.13428754e-02 -7.45989740e-01 4.80853289e-01 5.38037121e-01 3.37528020e-01 -4.48370963e-01 -9.28214967e-01 -1.14439368e-01 5.29329658e-01 -4.51599270e-01 -5.62811613e-01 -4.25715506e-01 -9.62079704e-01 3.46577525e-01 -4.15004969e-01 4.61717129e-01 1.11681557e+00 1.35642993e+00 4.91124183e-01 6.99971795e-01 8.51856411e-01 -1.04445815e+00 -7.72485733e-01 -1.13173091e+00 -4.62519348e-01 2.16988608e-01 5.77722371e-01 -7.10510135e-01 -5.12477636e-01 -1.49410605e-01]
[13.50604248046875, 5.691328525543213]
eeb2ac89-a724-4b71-8147-0ec81d6244ca
more-photos-are-all-you-need-semi-supervised
2103.1399
null
https://arxiv.org/abs/2103.13990v1
https://arxiv.org/pdf/2103.13990v1.pdf
More Photos are All You Need: Semi-Supervised Learning for Fine-Grained Sketch Based Image Retrieval
A fundamental challenge faced by existing Fine-Grained Sketch-Based Image Retrieval (FG-SBIR) models is the data scarcity -- model performances are largely bottlenecked by the lack of sketch-photo pairs. Whilst the number of photos can be easily scaled, each corresponding sketch still needs to be individually produced. In this paper, we aim to mitigate such an upper-bound on sketch data, and study whether unlabelled photos alone (of which they are many) can be cultivated for performances gain. In particular, we introduce a novel semi-supervised framework for cross-modal retrieval that can additionally leverage large-scale unlabelled photos to account for data scarcity. At the centre of our semi-supervision design is a sequential photo-to-sketch generation model that aims to generate paired sketches for unlabelled photos. Importantly, we further introduce a discriminator guided mechanism to guide against unfaithful generation, together with a distillation loss based regularizer to provide tolerance against noisy training samples. Last but not least, we treat generation and retrieval as two conjugate problems, where a joint learning procedure is devised for each module to mutually benefit from each other. Extensive experiments show that our semi-supervised model yields significant performance boost over the state-of-the-art supervised alternatives, as well as existing methods that can exploit unlabelled photos for FG-SBIR.
['Yi-Zhe Song', 'Tao Xiang', 'Yongxin Yang', 'Aneeshan Sain', 'Pinaki Nath Chowdhury', 'Ayan Kumar Bhunia']
2021-03-25
null
http://openaccess.thecvf.com//content/CVPR2021/html/Bhunia_More_Photos_Are_All_You_Need_Semi-Supervised_Learning_for_Fine-Grained_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Bhunia_More_Photos_Are_All_You_Need_Semi-Supervised_Learning_for_Fine-Grained_CVPR_2021_paper.pdf
cvpr-2021-1
['sketch-based-image-retrieval', 'semi-supervised-sketch-based-image-retrieval']
['computer-vision', 'computer-vision']
[ 6.09765530e-01 2.31055412e-02 -2.79843837e-01 -2.55844802e-01 -1.32494664e+00 -7.43762314e-01 9.92345393e-01 -3.47786754e-01 -8.71039703e-02 5.63977420e-01 2.79240578e-01 5.99523038e-02 -1.83143899e-01 -6.49688065e-01 -7.22376347e-01 -7.13201761e-01 4.32878196e-01 5.05293012e-01 -1.69686913e-01 -1.84638336e-01 2.89985061e-01 5.36126435e-01 -1.71432507e+00 4.11761999e-01 7.18568861e-01 1.01407063e+00 2.20444605e-01 5.44963241e-01 -5.71482144e-02 5.29957235e-01 -4.52545285e-01 -5.85406840e-01 5.48623681e-01 -4.06759560e-01 -6.96943104e-01 5.12213171e-01 1.03535342e+00 -6.46084487e-01 -3.79693955e-01 7.71254599e-01 8.36629808e-01 -4.98484895e-02 9.23138440e-01 -1.44808722e+00 -9.12598133e-01 3.99205834e-01 -4.81085449e-01 -3.55858833e-01 3.10288757e-01 1.53238535e-01 1.26045644e+00 -1.11636496e+00 8.32281351e-01 1.32020497e+00 2.87143052e-01 7.46196866e-01 -1.34621978e+00 -7.01911032e-01 1.60155952e-01 -2.88805962e-01 -1.54907739e+00 -7.55868435e-01 9.59680021e-01 -2.39511698e-01 3.94777805e-01 2.59308875e-01 2.45629400e-01 1.24226820e+00 -4.98861253e-01 1.25787330e+00 1.04525840e+00 -6.62221849e-01 1.04330741e-01 2.37798914e-01 -4.30209875e-01 7.43677378e-01 -6.17401041e-02 2.14488417e-01 -6.80766523e-01 -2.47610435e-01 9.30330217e-01 6.17186613e-02 -2.13495925e-01 -7.41695404e-01 -1.07673359e+00 8.84385943e-01 5.22572219e-01 8.37366655e-02 -1.46040887e-01 1.11280918e-01 2.37473607e-01 4.53661829e-01 4.58698988e-01 5.32938957e-01 -1.53309301e-01 3.16118032e-01 -1.30920851e+00 2.64152467e-01 6.03374898e-01 1.15243256e+00 9.59850132e-01 -1.58553675e-01 -4.81982827e-01 1.07588482e+00 2.05191597e-01 6.15750194e-01 3.01480860e-01 -9.66496766e-01 4.27264512e-01 4.62936997e-01 -3.30851637e-02 -9.62315083e-01 4.01646256e-01 -9.45576802e-02 -9.28215623e-01 2.07712851e-03 2.04861030e-01 4.57491785e-01 -9.96341407e-01 1.70236564e+00 6.76747272e-03 4.11872985e-03 -1.36742920e-01 9.45831299e-01 5.90724587e-01 4.26258683e-01 8.94912798e-03 4.97189686e-02 1.05624902e+00 -1.23774207e+00 -2.46441290e-01 -2.66222507e-01 3.13075364e-01 -9.25280631e-01 1.19030917e+00 1.97931454e-01 -1.05112064e+00 -5.11768460e-01 -1.00609910e+00 -2.27346122e-01 -3.66027027e-01 3.91411006e-01 3.91622782e-01 5.05811036e-01 -1.13071346e+00 5.89888752e-01 -3.04815739e-01 -3.03030133e-01 5.44921756e-01 1.33733928e-01 -5.66142499e-01 -5.14512062e-01 -9.32141662e-01 5.81795931e-01 2.01754615e-01 1.23860054e-01 -8.44480038e-01 -6.15041077e-01 -7.97719419e-01 -1.16046555e-02 4.26937044e-01 -8.08243573e-01 9.52068448e-01 -1.15501678e+00 -1.32646012e+00 1.15089273e+00 3.19691515e-03 -8.74771625e-02 8.52041721e-01 -2.90992539e-02 -4.89037707e-02 3.62323344e-01 2.39224300e-01 1.21991384e+00 1.41791558e+00 -1.73980975e+00 -2.26527527e-01 -2.07223892e-01 1.09511493e-02 2.55742222e-01 -5.44268429e-01 -9.37342867e-02 -8.83022606e-01 -9.46288705e-01 -1.56739607e-01 -1.12247682e+00 -1.83148310e-01 5.43824613e-01 -3.98501515e-01 -2.41600588e-01 7.25412548e-01 -3.52645725e-01 1.01217926e+00 -2.13043237e+00 1.83109537e-01 2.74933189e-01 -1.05783507e-01 4.38810110e-01 -8.09249282e-01 7.87858546e-01 8.15206990e-02 1.94423020e-01 -2.43775710e-01 -7.75969625e-01 1.86214015e-01 3.94468665e-01 -7.64162600e-01 1.53498471e-01 6.75520778e-01 1.01595354e+00 -1.02855980e+00 -6.44111931e-01 2.04027086e-01 5.32739162e-01 -4.54846948e-01 2.95073807e-01 -3.29491228e-01 1.91418618e-01 -4.01848793e-01 8.90170395e-01 6.96835577e-01 -3.09488595e-01 2.16171533e-01 -2.30726406e-01 2.98176736e-01 -2.06033718e-02 -1.15240157e+00 1.95855117e+00 -5.06593823e-01 2.31137872e-01 1.10779777e-01 -7.29956150e-01 9.19967771e-01 1.44225299e-01 1.04221068e-01 -7.53404200e-01 -1.82885319e-01 3.61003458e-01 -7.85692275e-01 -6.87772632e-02 7.26071596e-01 -2.12694734e-01 -1.73316956e-01 7.23165989e-01 1.77069977e-02 -4.34198081e-01 3.21707666e-01 5.26535749e-01 8.39633882e-01 3.58439982e-01 -7.89760873e-02 4.51880693e-02 4.41723287e-01 -2.08299711e-01 2.75587350e-01 9.45001602e-01 -3.73271219e-02 1.18752456e+00 1.47938177e-01 -1.57113045e-01 -1.26017392e+00 -1.02002549e+00 8.39632284e-03 1.24483299e+00 2.14867383e-01 -3.13358665e-01 -4.04962927e-01 -9.21443880e-01 1.08814649e-01 1.96382433e-01 -4.91861284e-01 -1.76695094e-01 -3.90196502e-01 -3.58424187e-01 6.33038580e-01 4.87729728e-01 2.64176995e-01 -1.22525692e+00 -1.30131766e-01 -7.81515241e-02 -1.41579032e-01 -9.89795864e-01 -7.43785858e-01 -1.47181913e-01 -5.90881050e-01 -8.98991764e-01 -1.07700455e+00 -8.16654563e-01 9.43758845e-01 6.32977009e-01 1.35556459e+00 3.78092587e-01 -4.71686751e-01 5.80740571e-01 -5.27003407e-01 -5.82999848e-02 -4.47929621e-01 6.07557260e-02 -1.06104940e-01 1.27052858e-01 -9.77659151e-02 -5.38914740e-01 -8.44333827e-01 4.87701952e-01 -1.43801653e+00 4.70500961e-02 9.60833848e-01 1.15073264e+00 5.97831368e-01 -2.40825564e-01 5.13992965e-01 -8.44412446e-01 5.10916948e-01 -2.22570822e-01 -2.99527198e-01 6.80848122e-01 -6.22945309e-01 2.72606462e-01 4.50696021e-01 -5.35603940e-01 -8.83488178e-01 2.64966518e-01 2.49693587e-01 -8.25085163e-01 -2.89033651e-02 1.95971593e-01 -2.15869814e-01 -1.96741790e-01 5.19825637e-01 3.37277144e-01 2.05786496e-01 -4.77972150e-01 8.45557034e-01 7.84177840e-01 4.49715316e-01 -8.68778408e-01 1.05653667e+00 5.45325518e-01 -1.23067737e-01 -7.45120823e-01 -8.30289304e-01 -5.70158958e-01 -5.00786602e-01 -1.07594118e-01 2.85274386e-01 -1.22483385e+00 -2.04290301e-01 2.76323020e-01 -9.03586924e-01 -2.98373997e-01 -2.46922880e-01 -1.65822804e-01 -5.93284965e-01 5.95791459e-01 -5.30874968e-01 -8.55678558e-01 -4.63399410e-01 -9.97165501e-01 1.78462279e+00 -5.25911115e-02 1.65724754e-02 -7.48991072e-01 1.80649366e-02 5.96608222e-01 4.76815462e-01 -3.35940048e-02 6.24390185e-01 -4.22169626e-01 -9.26973283e-01 -3.50690275e-01 -7.19717979e-01 5.08104444e-01 -4.22431193e-02 -3.33232507e-02 -1.02507663e+00 -5.43274164e-01 -6.46236241e-01 -9.78782773e-01 1.14494169e+00 -2.97152251e-01 1.09424472e+00 -4.79901433e-01 -1.52208954e-01 3.37096453e-01 1.46520603e+00 -6.10026062e-01 7.14648902e-01 4.51254584e-02 6.79592192e-01 6.55712605e-01 6.44553125e-01 4.21996564e-01 2.47002110e-01 8.22784305e-01 2.79822677e-01 -1.51471272e-01 -4.89985943e-01 -7.55510986e-01 3.00890416e-01 6.69477105e-01 3.56816649e-01 -2.99407631e-01 -4.03655559e-01 6.94078267e-01 -1.81428468e+00 -9.79214013e-01 3.69292140e-01 2.25745249e+00 1.01485670e+00 -3.17372203e-01 9.31695625e-02 1.78073138e-01 5.81022382e-01 5.03636658e-01 -4.00664181e-01 8.32009539e-02 -3.25632483e-01 2.37908542e-01 2.60780871e-01 3.51260245e-01 -1.09514260e+00 1.04743958e+00 6.00706863e+00 1.27651656e+00 -1.07394755e+00 -3.47335637e-01 6.42082632e-01 -1.83512390e-01 -5.80428779e-01 3.01971640e-02 -5.75859070e-01 4.44362044e-01 3.64657640e-01 1.97959602e-01 5.62435508e-01 7.76621938e-01 -2.10520998e-01 7.80560635e-03 -1.18755341e+00 1.03424513e+00 4.07004327e-01 -1.23973036e+00 5.80093563e-01 -1.34499120e-02 1.02288043e+00 -2.26234406e-01 2.30197474e-01 2.06315637e-01 3.52551132e-01 -8.10631037e-01 7.04974890e-01 4.19676244e-01 1.23064244e+00 -5.47188759e-01 2.34950036e-01 2.76492387e-01 -1.10601544e+00 1.64616406e-02 -4.04998481e-01 2.96746403e-01 7.12569505e-02 6.17826581e-01 -5.43519378e-01 8.24244142e-01 3.15498173e-01 6.02284253e-01 -7.32674718e-01 7.25868225e-01 -3.05958211e-01 1.56229839e-01 -3.05794835e-01 2.09731653e-01 2.33141124e-01 -8.76519307e-02 2.76320279e-01 1.11475289e+00 3.02889109e-01 -1.63516000e-01 3.40433449e-01 7.32944012e-01 -3.53720725e-01 -5.82731422e-03 -7.77040362e-01 -2.29910478e-01 6.37733638e-01 1.33954155e+00 -4.34334069e-01 -3.61194849e-01 -1.50221676e-01 1.39965546e+00 5.25350511e-01 4.00522113e-01 -2.73138404e-01 -1.47762641e-01 3.27329218e-01 5.29541373e-02 4.23966825e-01 1.21586427e-01 -5.17832786e-02 -1.33565497e+00 2.78303117e-01 -9.83823717e-01 4.03577000e-01 -1.02222896e+00 -1.85245049e+00 4.57148910e-01 -3.11389565e-01 -1.30353439e+00 -3.45161706e-01 -3.34210366e-01 -4.17068690e-01 8.08970094e-01 -2.00363970e+00 -1.78907752e+00 -1.15936168e-01 6.57500982e-01 5.06844521e-01 -5.35733476e-02 8.01549375e-01 4.73181874e-01 -3.58730167e-01 9.42632437e-01 7.30815679e-02 1.59779772e-01 1.23245752e+00 -1.14457250e+00 1.85707718e-01 7.22677529e-01 4.68953162e-01 6.16377652e-01 3.73675346e-01 -5.10535955e-01 -1.56520319e+00 -1.16426516e+00 9.56629813e-01 -5.00493169e-01 5.32696843e-01 -4.39942896e-01 -5.85753322e-01 3.04583222e-01 -9.61672887e-02 3.64099503e-01 4.88694459e-01 -1.18039265e-01 -8.90038669e-01 -9.89292338e-02 -1.02787197e+00 6.27082527e-01 1.13190186e+00 -9.91754293e-01 -2.96348870e-01 3.54961127e-01 4.42820311e-01 -7.63159767e-02 -5.74241877e-01 2.74110079e-01 7.59169996e-01 -9.14170384e-01 1.29143047e+00 -3.57671499e-01 8.65417480e-01 -2.40028292e-01 -2.08691210e-01 -1.07667840e+00 -9.54977274e-02 -8.06411862e-01 1.14971332e-01 1.58324230e+00 1.15366742e-01 -1.98280245e-01 1.00349116e+00 6.80057883e-01 2.80816913e-01 -5.34411788e-01 -5.36648571e-01 -9.50660884e-01 -3.27125601e-02 -5.48493629e-03 5.06753564e-01 8.34626675e-01 -2.82239556e-01 4.06343728e-01 -7.52094448e-01 -1.48091480e-01 7.22855806e-01 3.88751805e-01 1.01946127e+00 -1.02788603e+00 -2.92977720e-01 -4.51740324e-01 -1.64421007e-01 -1.13874912e+00 1.71923965e-01 -8.74549270e-01 2.00428978e-01 -1.30053484e+00 4.14729148e-01 -7.93579996e-01 -2.24329680e-01 5.42449415e-01 -4.34303313e-01 7.87787914e-01 6.33621812e-01 4.91431445e-01 -8.63741696e-01 7.42494285e-01 1.23972070e+00 -1.72385752e-01 2.66002920e-02 -2.04160139e-01 -7.88650513e-01 2.23612711e-01 3.86918128e-01 -2.47742906e-01 -5.22973537e-01 -4.94981349e-01 1.92614809e-01 9.66751799e-02 6.22010231e-01 -5.96650183e-01 1.77863896e-01 2.23083254e-02 1.78736776e-01 -3.03187221e-01 4.35060799e-01 -7.36141026e-01 2.62824334e-02 -7.48368278e-02 -6.15232110e-01 -2.99364567e-01 -2.09428921e-01 8.48054767e-01 -3.47530156e-01 -1.30405962e-01 5.03116548e-01 -1.52794108e-01 -3.78003389e-01 5.82745075e-01 1.17888756e-01 7.45806769e-02 5.50673068e-01 -3.05845290e-02 -3.43443662e-01 -4.13581997e-01 -2.77550578e-01 2.43445992e-01 8.17912638e-01 5.19487858e-01 5.49715579e-01 -1.69235885e+00 -7.60113835e-01 2.26165310e-01 5.92893720e-01 1.38370432e-02 3.08504820e-01 3.04835171e-01 -6.77496716e-02 2.90970296e-01 -4.28796150e-02 -3.06207508e-01 -1.24973917e+00 7.09642649e-01 -2.10627377e-01 -5.85227668e-01 -2.86228806e-01 8.89306784e-01 1.32354990e-01 -6.44228160e-01 4.06720608e-01 3.15603048e-01 4.02855009e-01 2.26797223e-01 4.92841750e-01 7.71431401e-02 -4.48544603e-03 -4.96468842e-01 -1.09115854e-01 5.73367119e-01 -2.56049424e-01 -2.40547076e-01 1.30079556e+00 -1.97487637e-01 1.88199300e-02 2.78396495e-02 1.35536599e+00 -8.95678103e-02 -1.44034219e+00 -5.95716834e-01 -1.99174985e-01 -7.22931385e-01 -1.34735852e-01 -9.56791520e-01 -9.95021641e-01 8.90774965e-01 2.61405170e-01 3.05607785e-02 1.12259519e+00 2.08202247e-02 9.14088130e-01 3.57452214e-01 4.06538129e-01 -1.01916385e+00 4.34055567e-01 7.65798539e-02 1.08230114e+00 -1.51422727e+00 1.77662015e-01 -3.20873260e-01 -6.25612736e-01 8.71100128e-01 2.68000960e-01 -2.75107861e-01 2.66994298e-01 -4.47770506e-02 4.53088917e-02 -8.10134038e-02 -7.36389995e-01 -3.08305711e-01 5.18059313e-01 5.07502198e-01 9.77145433e-02 -7.69179314e-02 5.33664636e-02 1.20418869e-01 2.16760591e-01 1.48972988e-01 4.35901247e-02 1.04508936e+00 -1.36341795e-01 -1.65726352e+00 -3.67234200e-01 3.78566682e-01 -1.30180329e-01 -2.05201358e-01 -7.66620755e-01 5.22512615e-01 -2.30545998e-01 7.96529055e-01 -9.46865678e-02 -1.80457026e-01 1.55752227e-01 -1.50669683e-02 6.43954515e-01 -4.65681344e-01 -3.82535666e-01 1.16299666e-01 -2.58309953e-02 -5.19652128e-01 -5.86437523e-01 -3.85466307e-01 -4.54437822e-01 -1.78804204e-01 -4.80243623e-01 -1.69697389e-01 5.55263579e-01 7.01361835e-01 6.41384900e-01 -9.54346061e-02 1.01999629e+00 -1.26764512e+00 -8.19897890e-01 -6.39233947e-01 -5.25235176e-01 6.94976330e-01 2.95788735e-01 -5.03876388e-01 -5.21373212e-01 1.35299355e-01]
[11.609463691711426, 0.6404974460601807]
6528f49b-9d41-4528-a5af-c5f17233f7e6
coarse-to-fine-a-hierarchical-diffusion-model
2305.13266
null
https://arxiv.org/abs/2305.13266v2
https://arxiv.org/pdf/2305.13266v2.pdf
Coarse-to-Fine: a Hierarchical Diffusion Model for Molecule Generation in 3D
Generating desirable molecular structures in 3D is a fundamental problem for drug discovery. Despite the considerable progress we have achieved, existing methods usually generate molecules in atom resolution and ignore intrinsic local structures such as rings, which leads to poor quality in generated structures, especially when generating large molecules. Fragment-based molecule generation is a promising strategy, however, it is nontrivial to be adapted for 3D non-autoregressive generations because of the combinational optimization problems. In this paper, we utilize a coarse-to-fine strategy to tackle this problem, in which a Hierarchical Diffusion-based model (i.e.~HierDiff) is proposed to preserve the validity of local segments without relying on autoregressive modeling. Specifically, HierDiff first generates coarse-grained molecule geometries via an equivariant diffusion process, where each coarse-grained node reflects a fragment in a molecule. Then the coarse-grained nodes are decoded into fine-grained fragments by a message-passing process and a newly designed iterative refined sampling module. Lastly, the fine-grained fragments are then assembled to derive a complete atomic molecular structure. Extensive experiments demonstrate that HierDiff consistently improves the quality of molecule generation over existing methods
['Yanyan Lan', 'WeiYing Ma', 'Hao Zhou', 'Bowen Gao', 'Jingjing Gong', 'Minkai Xu', 'Yuxuan Song', 'Bo Qiang']
2023-05-05
null
null
null
null
['drug-discovery']
['medical']
[ 3.70054811e-01 -1.30253240e-01 -1.23610504e-01 -3.32665294e-02 -7.55444109e-01 -3.58239859e-01 4.06463951e-01 1.73685014e-01 -4.69440371e-02 1.31864476e+00 1.09884471e-01 -2.18361184e-01 -1.83756296e-02 -1.23598659e+00 -7.49602616e-01 -1.06450284e+00 4.19949517e-02 6.12422109e-01 2.35162094e-01 -2.65639663e-01 4.57511574e-01 7.41314113e-01 -1.20050454e+00 2.20138207e-01 1.37795258e+00 3.73042226e-01 2.08732158e-01 4.29148585e-01 -2.57582784e-01 4.13255423e-01 -4.96597350e-01 -2.42717564e-01 4.98417057e-02 -9.42212641e-01 -6.89644396e-01 8.44883993e-02 6.13940321e-02 -1.09029889e-01 5.58200926e-02 1.02985680e+00 6.26003623e-01 2.37812653e-01 9.92100537e-01 -5.07933438e-01 -6.02932274e-01 4.35659140e-01 -6.18491113e-01 -2.58696169e-01 3.59242946e-01 1.81144699e-01 7.66188085e-01 -1.15020144e+00 7.78687060e-01 1.46416628e+00 4.25445735e-01 4.33837593e-01 -1.50757110e+00 -6.98282123e-01 1.73179463e-01 -2.45257795e-01 -1.56825519e+00 -3.72677177e-01 8.76148880e-01 -5.82463622e-01 8.45481694e-01 1.21756315e-01 8.78711283e-01 8.78502905e-01 7.48949170e-01 1.92206055e-01 1.08782470e+00 -2.58504093e-01 5.15599489e-01 -2.05633864e-01 -5.96215725e-02 7.53760755e-01 5.06931424e-01 1.81244146e-02 -6.04108930e-01 -4.87107426e-01 1.07497644e+00 6.24027587e-02 -1.09083846e-01 -2.76425630e-01 -1.18395543e+00 9.96301293e-01 5.80809653e-01 1.38460293e-01 -9.37609971e-01 -2.01721992e-02 -3.15653868e-02 -1.18879288e-01 4.71331090e-01 6.05142176e-01 -6.45409971e-02 2.43352298e-02 -1.09034705e+00 6.88353121e-01 6.64256036e-01 9.17784750e-01 9.25797284e-01 1.49431556e-01 -1.97668776e-01 5.42937815e-01 3.31293941e-01 1.49176612e-01 7.30310753e-02 -5.92408419e-01 2.57724762e-01 5.41314542e-01 1.45773545e-01 -8.74446630e-01 -2.81546444e-01 -4.39669639e-01 -1.30785596e+00 1.76559612e-01 4.87817936e-02 -2.21501991e-01 -9.11485016e-01 1.33996427e+00 7.74391234e-01 5.04980013e-02 6.74384832e-03 7.37216055e-01 6.39731348e-01 1.03068662e+00 2.30464160e-01 -8.18640888e-01 1.01273465e+00 -9.44681287e-01 -5.04329860e-01 5.34284234e-01 4.04349864e-01 -8.26728821e-01 5.84318280e-01 5.01435935e-01 -1.34096479e+00 -5.97660065e-01 -1.00241959e+00 7.80177265e-02 -1.65577754e-01 -2.76359379e-01 7.68305063e-01 5.91636956e-01 -7.14819372e-01 8.14168215e-01 -6.90802455e-01 1.40399173e-01 4.25556242e-01 4.17540759e-01 -2.24109843e-01 3.53201851e-02 -1.08442438e+00 4.48084831e-01 5.68774164e-01 1.96574867e-01 -1.05177343e+00 -7.14737594e-01 -6.30254924e-01 -2.09844172e-01 2.05074310e-01 -1.34892082e+00 7.75172293e-01 -5.66392004e-01 -1.85330510e+00 1.51734293e-01 -5.91484964e-01 -2.55871624e-01 4.14571494e-01 3.50390434e-01 8.84280633e-03 -5.15670925e-02 3.73736620e-02 7.18557119e-01 1.01595640e+00 -1.28960490e+00 -4.52625006e-01 -3.02532285e-01 -8.96231830e-02 3.49035442e-01 4.28230837e-02 -2.42948532e-01 -2.48016074e-01 -7.04822421e-01 3.17123681e-01 -9.48916614e-01 -8.83170784e-01 -4.54546571e-01 -6.40557945e-01 -1.20230339e-01 3.03821921e-01 -2.91380018e-01 1.37106538e+00 -1.54488051e+00 7.17760623e-01 4.25013274e-01 3.63916934e-01 6.29812777e-02 -8.82238150e-02 7.64766872e-01 -1.68386772e-01 3.46076638e-01 -4.44363773e-01 -7.31307715e-02 -3.87867540e-01 -1.70555145e-01 -1.23970352e-01 2.58882314e-01 2.39884183e-01 6.89849436e-01 -1.10957706e+00 -4.76757199e-01 1.40708581e-01 6.73910141e-01 -9.40105319e-01 5.49456477e-02 -5.80652833e-01 8.22436571e-01 -1.06100452e+00 6.79481268e-01 8.77509356e-01 -2.29675874e-01 3.86430100e-02 -3.67741644e-01 -5.20676851e-01 1.62082896e-01 -1.01365650e+00 1.63041317e+00 -9.58632678e-02 -3.96172851e-01 -2.03090563e-01 -5.97707987e-01 1.11616850e+00 2.61485398e-01 5.36637545e-01 -2.80866891e-01 -2.38337755e-01 3.54328603e-01 -3.16027016e-03 -1.04821868e-01 7.68740475e-01 -6.88540637e-01 5.49545661e-02 1.52050570e-01 -3.35343093e-01 -5.42235613e-01 2.37228960e-01 1.07088454e-01 7.57327914e-01 2.73136586e-01 3.36735100e-01 -2.09594652e-01 5.32292604e-01 3.10502112e-01 6.36081398e-01 6.14841223e-01 2.67380744e-01 7.10277796e-01 4.52145964e-01 -4.65029418e-01 -1.16212499e+00 -8.32110226e-01 -1.60055622e-01 6.59920633e-01 1.74257517e-01 -6.00741208e-01 -9.68843639e-01 -5.12061000e-01 -1.58627450e-01 4.96357411e-01 -3.86178702e-01 -1.88713968e-01 -4.78274405e-01 -1.31860733e+00 2.23678246e-01 -5.30443676e-02 2.85772383e-01 -1.09995890e+00 -3.21031664e-03 8.16124141e-01 2.72092313e-01 -3.31240952e-01 -4.15529877e-01 -6.01303428e-02 -1.00225127e+00 -7.80944049e-01 -8.68356586e-01 -5.79693675e-01 7.98822224e-01 3.76029342e-01 7.16983795e-01 4.07430194e-02 -1.85670227e-01 -6.24971449e-01 -3.12786102e-01 -1.09950706e-01 -5.03533781e-01 3.05672437e-01 3.83733120e-03 6.56474382e-02 -1.14502557e-01 -5.54788828e-01 -7.80308962e-01 3.46150622e-03 -8.79955411e-01 3.28473359e-01 7.98811972e-01 9.63279843e-01 1.21509528e+00 4.36597645e-01 7.04246163e-01 -1.09984708e+00 7.92328000e-01 -4.27345335e-01 -6.17794275e-01 6.44210773e-03 -4.55933779e-01 2.50882536e-01 8.14308584e-01 -3.02444130e-01 -1.20403540e+00 2.03336254e-01 -5.53808451e-01 -1.14151090e-01 -9.57111493e-02 6.98545218e-01 -4.03213769e-01 -2.52868086e-01 5.36663771e-01 4.00595158e-01 -1.72219783e-01 -3.97309363e-01 3.20777714e-01 4.26971883e-01 -5.49863316e-02 -8.60925317e-01 6.31308377e-01 2.62993366e-01 3.28127474e-01 -1.08911252e+00 -4.32736009e-01 7.94679020e-03 -7.49382794e-01 1.31453559e-01 1.05020058e+00 -7.51110613e-01 -6.91230237e-01 4.82401729e-01 -1.23230922e+00 -1.39543727e-01 -9.80940536e-02 4.98616904e-01 -3.90007496e-01 4.37444419e-01 -6.74904406e-01 -6.48400068e-01 -4.24105734e-01 -1.68032837e+00 1.18894219e+00 3.11929703e-01 -2.35463426e-01 -8.02242100e-01 3.00136507e-01 2.41707459e-01 2.38770559e-01 4.20106769e-01 1.27016139e+00 -9.70826596e-02 -9.94774520e-01 -6.21616356e-02 6.27094582e-02 -2.07683891e-01 3.15525383e-01 2.36634955e-01 -4.32481080e-01 -3.48750561e-01 -1.81375653e-01 -4.75224815e-02 6.80855870e-01 5.50521195e-01 1.09745002e+00 -2.41299093e-01 -5.52740633e-01 5.92372239e-01 1.12558186e+00 5.15824020e-01 6.92715943e-01 6.42882064e-02 9.81512308e-01 4.12984937e-01 6.71797812e-01 5.68032980e-01 1.29188672e-01 6.50720537e-01 1.50492683e-01 -2.26614028e-01 -1.96762737e-02 -5.35907686e-01 1.78906307e-01 8.62993479e-01 -3.59686971e-01 -3.68213326e-01 -5.34619927e-01 7.58756101e-02 -1.60718012e+00 -1.00049913e+00 -3.56592834e-01 2.23748779e+00 1.12320530e+00 1.08193956e-01 2.10783765e-01 -1.97930679e-01 6.54632986e-01 1.80322707e-01 -7.45329320e-01 -3.04707736e-01 3.05493572e-03 3.90631348e-01 2.92193294e-01 7.40835845e-01 -6.64722800e-01 1.32199430e+00 6.48189068e+00 1.26192653e+00 -1.14336288e+00 -2.27378756e-01 7.26126730e-01 1.67321846e-01 -7.00204611e-01 2.64218152e-01 -1.22378492e+00 5.08998334e-01 6.21343493e-01 -1.94561750e-01 2.04505652e-01 5.67512274e-01 6.60369754e-01 5.38351201e-02 -7.13566601e-01 6.61940992e-01 -2.23478332e-01 -1.71742940e+00 7.02609658e-01 3.64853978e-01 1.02086926e+00 -6.49115622e-01 -1.27212390e-01 -1.02921424e-03 2.14908153e-01 -1.14675927e+00 4.23865139e-01 6.74432874e-01 6.90289795e-01 -1.21703410e+00 2.73606569e-01 4.77026165e-01 -1.24711013e+00 5.98231256e-01 -6.44917548e-01 1.07925639e-01 4.07505631e-01 9.00688827e-01 -7.55159914e-01 7.82449007e-01 6.46753833e-02 6.35214210e-01 -6.25077635e-02 9.72262323e-01 5.03238328e-02 4.23089713e-01 -9.69729125e-02 -2.13267013e-01 2.45114073e-01 -1.02168024e+00 5.57738245e-01 9.06765103e-01 4.55434442e-01 3.64044249e-01 3.62617016e-01 1.11030447e+00 -1.06599696e-01 1.18881799e-01 -6.28368676e-01 -4.35595103e-02 4.90748852e-01 1.11452758e+00 -6.50955856e-01 -4.34007168e-01 -1.41028911e-01 1.00832498e+00 2.11870193e-01 4.79416162e-01 -6.75818324e-01 -3.78441244e-01 4.87507284e-01 2.65349984e-01 2.83775210e-01 -4.62837517e-01 -3.61987017e-02 -9.93821323e-01 -3.61659169e-01 -1.15944529e+00 1.06078414e-02 -4.70209122e-01 -1.07585883e+00 7.52785504e-01 -1.77474871e-01 -8.77193451e-01 -9.85357463e-02 -1.78310692e-01 -4.59657729e-01 1.14575303e+00 -1.23411691e+00 -9.57430542e-01 -8.71543363e-02 3.04328144e-01 6.67205453e-01 9.16593000e-02 8.28361392e-01 3.97957414e-02 -7.70101488e-01 3.63387287e-01 2.00295433e-01 -7.05936491e-01 5.27417839e-01 -9.93969500e-01 3.54520380e-01 5.72302699e-01 -2.75560439e-01 8.90500128e-01 6.04220688e-01 -1.25305307e+00 -1.53675866e+00 -1.29997146e+00 7.10357487e-01 -7.76116699e-02 3.95461410e-01 -2.31251270e-01 -1.02709222e+00 2.24020690e-01 -2.42180470e-02 -4.61920798e-01 5.49744546e-01 -1.77136362e-01 1.18407011e-01 1.31277010e-01 -1.03834784e+00 7.39574134e-01 1.13144481e+00 -2.52935946e-01 -1.48231447e-01 4.92687225e-01 8.89099479e-01 -4.99624193e-01 -1.18538332e+00 4.08371806e-01 4.43645597e-01 -9.43524301e-01 1.11470139e+00 -5.34937918e-01 3.47033322e-01 -6.94227397e-01 2.02177182e-01 -1.31063163e+00 -6.42861784e-01 -9.92235541e-01 1.08845249e-01 1.09796321e+00 6.46548212e-01 -6.20051920e-01 1.05385613e+00 3.91851336e-01 -3.07786822e-01 -8.73976648e-01 -6.78481936e-01 -4.91054267e-01 3.35448533e-01 2.00489327e-01 8.91989946e-01 6.55739069e-01 -2.41396815e-01 6.77645266e-01 -4.92151111e-01 1.65566579e-01 6.55583918e-01 3.94967079e-01 6.93234026e-01 -9.79832351e-01 -3.46215874e-01 -3.95511955e-01 8.33373889e-03 -1.39317155e+00 -3.49731557e-02 -6.92673922e-01 -4.56828661e-02 -1.49732649e+00 3.41815382e-01 -7.28661597e-01 1.39027268e-01 -1.14799209e-01 -3.37273091e-01 9.60385352e-02 -1.81500927e-01 4.36845571e-01 -4.05955136e-01 9.92488444e-01 1.76728690e+00 -1.08974418e-02 -6.83858871e-01 8.60247985e-02 -6.71147346e-01 4.83289868e-01 5.96456110e-01 -5.01721680e-01 -3.65736395e-01 2.55117789e-02 1.28309965e-01 4.51689988e-01 -1.96347609e-02 -9.16694760e-01 -8.28600954e-03 -4.38273013e-01 4.61750954e-01 -9.52736259e-01 4.53887880e-01 -3.59524310e-01 7.23177075e-01 5.48122644e-01 -1.31198421e-01 -1.32318541e-01 -1.53675243e-01 6.67316973e-01 -2.54993767e-01 -2.98877358e-01 8.23625565e-01 -3.07408363e-01 -3.20492685e-02 8.38341296e-01 -4.98612672e-01 -4.56411690e-01 9.95540738e-01 -3.23078483e-01 1.45699352e-01 3.02112028e-02 -5.77695489e-01 -5.14765121e-02 8.29575956e-01 -1.01391122e-01 6.68990791e-01 -1.27545965e+00 -6.99616373e-01 2.64471501e-01 -1.66867942e-01 6.37262344e-01 3.35976601e-01 6.23160541e-01 -7.25915611e-01 4.47417408e-01 1.20898493e-01 -4.75510210e-01 -1.02962661e+00 6.70637369e-01 2.44515643e-01 -3.47316414e-01 -3.00967783e-01 7.16166675e-01 6.04445457e-01 -1.46213979e-01 -2.92625546e-01 -2.45851621e-01 -5.24061639e-03 1.01208486e-01 5.11579394e-01 2.16750085e-01 1.11607157e-01 -6.55223191e-01 -7.64452592e-02 8.58938277e-01 -4.16067004e-01 1.28734931e-01 1.34064245e+00 1.07412383e-01 -2.72567421e-01 8.54791328e-02 9.50650156e-01 2.18328983e-01 -1.26848876e+00 1.73313111e-01 -4.00710106e-01 -3.16097856e-01 -4.93704416e-02 -1.95394799e-01 -6.00556970e-01 7.37820864e-01 4.39701192e-02 -4.16530333e-02 7.11371243e-01 -1.70250416e-01 8.92115414e-01 3.35789233e-01 5.59107602e-01 -6.58261657e-01 6.95539862e-02 4.44780409e-01 7.64086783e-01 -6.89553440e-01 1.74962014e-01 -7.51220584e-01 -4.61377263e-01 1.00520384e+00 3.74598622e-01 -7.09990188e-02 3.45248163e-01 -1.03576832e-01 -5.55298209e-01 -2.39301920e-01 -5.78753173e-01 -2.13662572e-02 1.16913319e-01 5.00766337e-01 6.96494401e-01 -6.78474130e-03 -4.29320842e-01 4.39812869e-01 -6.94528148e-02 1.18753612e-02 2.77676225e-01 9.33190703e-01 -6.37127161e-01 -1.59602952e+00 -4.32026237e-01 3.29376489e-01 -1.88792780e-01 -4.10904378e-01 -4.23020184e-01 5.18125594e-01 2.76636183e-01 7.41168678e-01 -2.82148987e-01 -1.07623942e-01 3.80713016e-01 -2.45947227e-01 6.06565416e-01 -8.13655078e-01 -5.22774458e-01 4.31909949e-01 -2.57119555e-02 -2.63538808e-01 -2.71411479e-01 -5.57976425e-01 -1.29788387e+00 -4.42169458e-01 -5.04618466e-01 5.74608207e-01 3.44460577e-01 6.13691568e-01 6.48104250e-01 5.38121104e-01 7.65348434e-01 -1.19642210e+00 -3.05978775e-01 -6.23661816e-01 -5.29040098e-01 1.09599337e-01 2.20481455e-01 -6.05486035e-01 -3.24794464e-02 4.93728481e-02]
[4.994266033172607, 5.704833507537842]
6b524437-4d70-4fc9-b943-7907b207279c
predicting-human-scanpaths-in-visual-question
null
null
http://openaccess.thecvf.com//content/CVPR2021/html/Chen_Predicting_Human_Scanpaths_in_Visual_Question_Answering_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_Predicting_Human_Scanpaths_in_Visual_Question_Answering_CVPR_2021_paper.pdf
Predicting Human Scanpaths in Visual Question Answering
Attention has been an important mechanism for both humans and computer vision systems. While state-of-the-art models to predict attention focus on estimating a static probabilistic saliency map with free-viewing behavior, real-life scenarios are filled with tasks of varying types and complexities, and visual exploration is a temporal process that contributes to task performance. To bridge the gap, we conduct a first study to understand and predict the temporal sequences of eye fixations (a.k.a. scanpaths) during performing general tasks, and examine how scanpaths affect task performance. We present a new deep reinforcement learning method to predict scanpaths leading to different performances in visual question answering. Conditioned on a task guidance map, the proposed model learns question-specific attention patterns to generate scanpaths. It addresses the exposure bias in scanpath prediction with self-critical sequence training and designs a Consistency-Divergence loss to generate distinguishable scanpaths between correct and incorrect answers. The proposed model not only accurately predicts the spatio-temporal patterns of human behavior in visual question answering, such as fixation position, duration, and order, but also generalizes to free-viewing and visual search tasks, achieving human-level performance in all tasks and significantly outperforming the state of the art.
['Qi Zhao', 'Ming Jiang', 'Xianyu Chen']
2021-06-19
null
null
null
cvpr-2021-1
['scanpath-prediction']
['computer-vision']
[ 3.97553444e-01 -1.55534491e-01 6.58924878e-02 -4.26285386e-01 -2.04266682e-01 -4.55695093e-01 4.30219352e-01 1.25361502e-01 -3.43696207e-01 2.98963666e-01 -5.47128078e-03 -4.88722235e-01 -2.01438546e-01 -3.24700028e-01 -6.96913719e-01 -2.74906665e-01 1.96989119e-01 3.88777852e-01 8.05768430e-01 -3.31960678e-01 1.02385902e+00 1.54560491e-01 -1.95553243e+00 4.08906102e-01 1.08406377e+00 1.03569508e+00 9.02216792e-01 9.45332527e-01 -1.98111117e-01 8.18547964e-01 -7.00677514e-01 -2.18628988e-01 1.34843767e-01 -6.89648807e-01 -8.79503727e-01 -7.81034902e-02 6.20960593e-01 -1.34790555e-01 -3.89462829e-01 9.48624492e-01 3.22179615e-01 4.30368870e-01 6.45647287e-01 -1.23440385e+00 -1.42246747e+00 9.62055847e-02 -6.24149561e-01 9.65807676e-01 5.62329590e-01 8.33779156e-01 1.10736179e+00 -6.79888427e-01 2.35955983e-01 1.24614024e+00 1.52909547e-01 4.81577456e-01 -1.05815792e+00 -5.39663494e-01 4.62202728e-01 1.05440688e+00 -9.82278645e-01 -6.00302070e-02 6.92701399e-01 -6.88841045e-01 9.41448927e-01 2.78190643e-01 8.53538394e-01 1.15945995e+00 7.19444811e-01 9.17714417e-01 1.33314574e+00 -4.12438780e-01 3.08906108e-01 4.91598472e-02 2.93873459e-01 5.68434954e-01 -1.18500426e-01 3.86025757e-01 -8.80484581e-01 1.96916491e-01 7.90505230e-01 5.88077903e-02 -3.05512428e-01 -4.02612180e-01 -1.29032135e+00 6.24492586e-01 7.65524030e-01 8.53700340e-02 -5.43733239e-01 -1.68051943e-02 -6.35878146e-02 3.55991051e-02 -7.77345449e-02 8.61605704e-01 -3.16057980e-01 -4.75774072e-02 -9.07491565e-01 5.73695838e-01 2.81919509e-01 1.07423556e+00 7.52911329e-01 3.81346308e-02 -1.29492903e+00 6.29891455e-01 1.22421861e-01 5.89141190e-01 7.28925467e-01 -9.97903883e-01 3.24108154e-01 5.72474599e-01 4.30387467e-01 -1.04493546e+00 -5.36086500e-01 -4.67349470e-01 -2.67334163e-01 1.42813429e-01 5.69527686e-01 3.59772027e-01 -1.13056707e+00 1.76932347e+00 8.83583650e-02 -1.45183221e-01 -6.80788040e-01 1.31913531e+00 4.42766547e-01 4.61157650e-01 3.33994657e-01 -7.20338374e-02 1.65038800e+00 -1.32036650e+00 -9.71082747e-01 -6.49253309e-01 3.59067582e-02 -5.87224245e-01 1.90955782e+00 2.53399700e-01 -1.25007737e+00 -1.05937672e+00 -8.65457714e-01 -2.44816035e-01 -1.29713416e-01 -2.05493987e-01 3.14284563e-01 3.41129661e-01 -1.24437952e+00 4.27036434e-01 -4.48877931e-01 -2.41298780e-01 5.33333123e-01 -4.53964844e-02 3.60105872e-01 3.76751311e-02 -1.33718777e+00 1.19704068e+00 2.23381504e-01 -1.94251630e-02 -1.04380596e+00 -9.64580476e-01 -5.52303851e-01 4.00812119e-01 5.11142671e-01 -8.31373572e-01 1.56781685e+00 -1.10516906e+00 -1.31534255e+00 7.68165886e-01 -6.21219873e-01 -4.04100388e-01 1.66124910e-01 -2.00679943e-01 -3.36686015e-01 1.68544158e-01 2.33136848e-01 9.77906942e-01 1.28914452e+00 -1.09003091e+00 -6.87117040e-01 -2.07350671e-01 8.82426128e-02 3.64421964e-01 -1.35451248e-02 -1.61976904e-01 -1.29844189e-01 -4.63449836e-01 -2.06218138e-01 -8.45576704e-01 -7.99792036e-02 7.05005154e-02 -2.23079279e-01 -4.95400250e-01 4.55214262e-01 -7.37479031e-01 1.43004918e+00 -1.95598054e+00 1.20527528e-01 -4.94522676e-02 3.66597086e-01 2.49245435e-01 -2.72917956e-01 1.44162551e-01 3.49795334e-02 -1.59996301e-01 -1.95223734e-01 9.02926698e-02 3.25176045e-02 -1.60505772e-01 -4.69810635e-01 2.08090127e-01 2.43489489e-01 1.39757657e+00 -1.17982638e+00 -4.15162712e-01 -3.08051463e-02 -6.30022585e-02 -6.87756360e-01 5.87432206e-01 -6.04045808e-01 6.92768872e-01 -1.16605423e-01 6.48216546e-01 4.21274304e-01 -7.35376000e-01 -1.69153884e-01 1.07515976e-01 -1.21708557e-01 2.42783010e-01 -3.81308943e-01 1.56580412e+00 -3.37596148e-01 1.01182830e+00 -4.93361801e-01 -7.10271478e-01 8.40809584e-01 -2.97991306e-01 -2.52437085e-01 -1.51384521e+00 9.84326005e-02 -1.86875597e-01 5.23313522e-01 -8.73916507e-01 6.97856843e-01 2.37461075e-01 2.51399428e-01 5.47402859e-01 4.35551032e-02 -1.50024053e-02 2.24058926e-01 1.00276740e-02 8.96093488e-01 1.85662404e-01 3.99461359e-01 -4.14850324e-01 4.37397391e-01 8.80598426e-02 1.78013578e-01 1.06341100e+00 -7.89815426e-01 6.81629598e-01 6.41520858e-01 -4.83409971e-01 -1.03401828e+00 -1.22443748e+00 1.80029944e-01 1.70320880e+00 6.43312216e-01 3.79618369e-02 -9.34693456e-01 -5.68227828e-01 -2.37101823e-01 1.29350007e+00 -1.07063401e+00 -5.69123685e-01 -4.43759173e-01 -2.05078244e-01 3.22902277e-02 4.36778456e-01 5.03224909e-01 -1.90066135e+00 -1.36058807e+00 -2.49646798e-01 -2.75054365e-01 -8.16645980e-01 -1.05064511e+00 -2.61774629e-01 -5.73103607e-01 -1.36675501e+00 -7.88209498e-01 -7.71713674e-01 7.28871524e-01 5.90130389e-01 1.42340827e+00 1.50229931e-01 -3.83355707e-01 4.24368531e-01 -3.94967288e-01 -6.41663551e-01 3.84319462e-02 2.05674041e-02 -3.01215798e-01 -5.84337935e-02 5.71963608e-01 -1.78717285e-01 -1.37953150e+00 6.22819364e-01 -7.61248767e-01 1.12334155e-01 7.13603318e-01 7.55039573e-01 5.76402485e-01 -5.49437225e-01 3.58553797e-01 -5.87500215e-01 9.96755958e-01 -6.61744714e-01 -5.99584103e-01 6.00292981e-01 -8.88763309e-01 2.14881614e-01 3.00038785e-01 -6.27400815e-01 -1.11044300e+00 -4.45805758e-01 2.64673501e-01 -6.13787591e-01 -2.17970818e-01 1.09369673e-01 3.42720687e-01 1.94843039e-01 9.31201637e-01 5.45830369e-01 -4.91624810e-02 -2.13196129e-02 1.47866994e-01 1.49058625e-01 3.23894531e-01 -1.92425892e-01 4.91324455e-01 2.44048417e-01 -1.36971742e-01 -5.93106091e-01 -1.25539434e+00 -4.32766199e-01 -5.74155390e-01 -4.17894125e-01 1.02372122e+00 -4.15130377e-01 -9.81608808e-01 3.54616553e-01 -1.33743751e+00 -6.54462278e-01 -2.77102470e-01 4.14773114e-02 -7.31761992e-01 1.13380872e-01 -1.08575523e-01 -8.05215657e-01 -2.19537169e-01 -1.17492163e+00 8.47733498e-01 6.81836665e-01 -4.48835373e-01 -7.41237879e-01 1.40420243e-01 3.85839671e-01 6.57211781e-01 -3.24510515e-01 1.02940476e+00 -5.05111039e-01 -9.39628780e-01 5.41137457e-01 -5.18356860e-01 -1.41258180e-01 -1.34487942e-01 -3.46050382e-01 -8.09154093e-01 -1.40514106e-01 2.21270338e-01 -1.08848125e-01 8.19196224e-01 7.37370968e-01 1.64358854e+00 -2.81812340e-01 -1.75540283e-01 5.90895951e-01 9.83419478e-01 4.77735758e-01 7.78367460e-01 3.73729289e-01 5.02834737e-01 9.80646610e-01 9.45930779e-01 1.91071242e-01 6.16773188e-01 8.04567099e-01 7.88300693e-01 2.15456322e-01 -1.55493587e-01 -4.57479268e-01 1.21269241e-01 1.46919280e-01 2.03311190e-01 -1.66679919e-01 -9.85301435e-01 7.48167932e-01 -1.79224908e+00 -1.16210222e+00 -2.53609627e-01 2.09181213e+00 7.36504018e-01 1.63971201e-01 2.20365942e-01 -2.68161595e-01 8.51320088e-01 3.66254807e-01 -1.16714668e+00 -4.71869349e-01 2.12452397e-01 2.87277341e-01 2.37049237e-01 3.62396449e-01 -5.79354584e-01 1.01729095e+00 6.91205549e+00 5.94962537e-01 -9.70064282e-01 2.61550903e-01 4.84875917e-01 -2.88532346e-01 -4.57177907e-01 -3.56639549e-02 -6.41026497e-01 7.09001243e-01 9.95814562e-01 -6.60800636e-02 6.88448846e-01 6.42982543e-01 3.32899392e-01 -4.37248975e-01 -1.02925158e+00 1.01570868e+00 3.47199321e-01 -1.15491879e+00 3.93354967e-02 -2.69638181e-01 8.54593635e-01 -3.04808855e-01 6.03429794e-01 4.27624464e-01 -2.47386638e-02 -1.22766280e+00 1.06122565e+00 8.88637185e-01 6.43659413e-01 -3.26109290e-01 2.65226424e-01 5.04737914e-01 -7.60131598e-01 -5.44841170e-01 -4.66350287e-01 -2.38459468e-01 2.03813925e-01 2.14491218e-01 -7.91243255e-01 -1.66059569e-01 9.52581763e-01 4.08267111e-01 -1.11197603e+00 1.30939770e+00 -4.65046942e-01 4.90490735e-01 5.25584757e-01 -5.51736534e-01 1.39106810e-01 1.07928522e-01 4.69605237e-01 8.04022312e-01 3.89113069e-01 8.36800188e-02 -3.36274266e-01 1.38976169e+00 1.89077914e-01 -1.18271649e-01 -1.88463733e-01 1.79658532e-01 6.01466298e-01 8.26552093e-01 -6.56120121e-01 -1.76503882e-01 -1.01863772e-01 1.00814509e+00 5.22222400e-01 6.89562857e-01 -1.15301538e+00 -3.00022930e-01 5.11568189e-01 6.09967589e-01 5.93658090e-01 -1.08788364e-01 -5.57295322e-01 -7.27945507e-01 -5.80981281e-03 -8.00344050e-01 2.05511838e-01 -1.24231386e+00 -1.06163752e+00 7.30306447e-01 -1.79369878e-02 -1.15294814e+00 -9.09876525e-02 -5.69017947e-01 -8.04570735e-01 1.13736045e+00 -1.85500109e+00 -6.57384813e-01 -7.15946198e-01 6.44879282e-01 1.07488108e+00 -1.73863351e-01 3.15266758e-01 -3.33493531e-01 -3.22192371e-01 6.24486685e-01 -4.22567576e-01 -3.80479246e-01 7.17712641e-01 -1.22005928e+00 5.87599814e-01 6.91310763e-01 2.00290605e-01 6.56973124e-01 9.31653380e-01 -4.49734449e-01 -9.32239473e-01 -6.70948744e-01 6.94608212e-01 -7.68388987e-01 4.31883991e-01 -2.60090917e-01 -1.17423427e+00 4.72226083e-01 5.51281631e-01 -1.29831836e-01 4.67870533e-01 7.67326951e-02 -3.65674347e-01 1.58201337e-01 -7.53569722e-01 7.52735615e-01 1.24578857e+00 -5.95215797e-01 -9.89343643e-01 2.42334098e-01 8.34239781e-01 -5.85922539e-01 7.18121678e-02 -1.72941305e-03 4.91520375e-01 -1.44610810e+00 8.53985310e-01 -7.10603654e-01 6.15506589e-01 -3.13165933e-01 4.07129735e-01 -1.40818131e+00 -8.81667197e-01 -3.55132848e-01 -4.69091058e-01 4.86164421e-01 2.92333275e-01 -4.19090331e-01 6.61391675e-01 4.45858896e-01 -1.56901985e-01 -8.43888104e-01 -6.47893786e-01 -4.12211657e-01 -2.47705668e-01 -1.26729757e-01 5.62839150e-01 3.23063284e-01 -2.34769091e-01 3.75264853e-01 -2.58954197e-01 1.53143913e-01 5.19638300e-01 3.14586073e-01 4.48859930e-01 -1.13091707e+00 -2.10838243e-01 -7.31589317e-01 5.94421551e-02 -1.43369043e+00 1.23397015e-01 -5.55131495e-01 3.12578142e-01 -1.27682304e+00 2.44393140e-01 5.78461662e-02 -5.88282108e-01 2.30272356e-02 -8.15736234e-01 -6.29359931e-02 3.25012207e-01 4.56080675e-01 -1.00493550e+00 6.58602238e-01 1.75399387e+00 -1.00582406e-01 -3.42103451e-01 3.08139157e-02 -8.92476857e-01 3.21650982e-01 7.12276280e-01 -1.93287984e-01 -8.40489268e-01 -5.73414147e-01 4.95183975e-01 1.66022137e-01 6.57211304e-01 -9.51647580e-01 3.30388337e-01 -5.58516204e-01 4.25806612e-01 -7.82454610e-01 4.77551185e-02 -4.91877526e-01 -4.69019592e-01 5.41082501e-01 -8.81825805e-01 4.18692619e-01 2.10829109e-01 8.35226774e-01 -3.51809785e-02 -2.80886769e-01 7.74841309e-01 -2.00395375e-01 -9.39085066e-01 3.76139134e-01 -2.50253826e-01 4.14498270e-01 1.15917039e+00 -5.27563691e-01 -4.83599961e-01 -4.18088347e-01 -4.92142439e-01 5.06248653e-01 1.82584003e-01 8.61109018e-01 8.60606253e-01 -1.06292737e+00 -4.99987334e-01 3.20309699e-01 2.34218076e-01 -2.32870981e-01 7.79922426e-01 8.67020309e-01 -4.26832438e-01 7.51438975e-01 -6.35504246e-01 -7.70072877e-01 -7.26403773e-01 9.49191511e-01 2.06547484e-01 -1.16552614e-01 -1.56753421e-01 1.04352617e+00 6.71994925e-01 1.46574974e-01 8.82090926e-02 -1.81846604e-01 -6.01771712e-01 -1.06175840e-01 7.27880418e-01 3.04659903e-01 -1.74532548e-01 -2.57871389e-01 -1.51967451e-01 4.31240320e-01 -2.05446094e-01 4.71351249e-03 6.46636069e-01 -4.40827847e-01 2.86169797e-01 6.24154866e-01 5.66612303e-01 -4.82509822e-01 -1.84307671e+00 -7.51654878e-02 -8.76976177e-02 -8.91759634e-01 -2.70820588e-01 -9.83637810e-01 -6.91473722e-01 1.24111164e+00 6.58028007e-01 3.00265014e-01 9.75670934e-01 1.50660321e-01 7.70728588e-01 1.44255683e-01 2.61745572e-01 -1.16967797e+00 9.09824848e-01 6.63737118e-01 1.13218570e+00 -1.47843921e+00 -3.25842738e-01 8.32914263e-02 -1.08409035e+00 6.45064771e-01 1.23963690e+00 -1.25811890e-01 4.10818189e-01 -6.06517613e-01 -9.55772921e-02 -3.98229629e-01 -9.77303922e-01 -3.00401807e-01 5.01677930e-01 8.67221355e-01 2.03950092e-01 -1.81814432e-01 -2.40821481e-01 4.40039009e-01 -3.23015988e-01 -2.07539454e-01 3.26952189e-01 6.34465456e-01 -6.09417856e-01 -4.72204357e-01 -4.28175747e-01 7.13135183e-01 -9.32678655e-02 -2.71272868e-01 -1.64544433e-01 4.55004632e-01 -6.96946494e-03 7.04549968e-01 4.64456350e-01 -2.54239410e-01 5.07645428e-01 2.01628670e-01 4.97325271e-01 -6.64844513e-01 -4.99851316e-01 -5.88483751e-01 -6.53260529e-01 -7.04567492e-01 -1.26350895e-01 -6.26119196e-01 -1.00526750e+00 -2.64008008e-02 -2.09741816e-01 -8.52119774e-02 3.78371119e-01 9.32368398e-01 5.68573296e-01 9.07287240e-01 4.24597114e-01 -7.76785672e-01 -7.07339704e-01 -9.98378515e-01 -3.81005615e-01 6.04196131e-01 5.15128970e-01 -1.08303487e+00 -3.43582004e-01 3.67150679e-02]
[10.1679105758667, 1.232889175415039]
17bf4c3a-58de-4598-b85c-2ee8e7c042f0
nested-named-entity-recognition-revisited
null
null
https://aclanthology.org/N18-1079
https://aclanthology.org/N18-1079.pdf
Nested Named Entity Recognition Revisited
We propose a novel recurrent neural network-based approach to simultaneously handle nested named entity recognition and nested entity mention detection. The model learns a hypergraph representation for nested entities using features extracted from a recurrent neural network. In evaluations on three standard data sets, we show that our approach significantly outperforms existing state-of-the-art methods, which are feature-based. The approach is also efficient: it operates linearly in the number of tokens and the number of possible output labels at any token. Finally, we present an extension of our model that jointly learns the head of each entity mention.
['Arzoo Katiyar', 'Claire Cardie']
2018-06-01
null
null
null
naacl-2018-6
['nested-named-entity-recognition']
['natural-language-processing']
[-1.54681608e-01 6.43495321e-01 -3.91120195e-01 -3.37889701e-01 -8.06257486e-01 -5.08031368e-01 4.25806314e-01 5.29474854e-01 -5.03134906e-01 6.09107137e-01 5.19251525e-01 -3.73161733e-01 1.64747119e-01 -1.12113762e+00 -7.52383053e-01 -1.18125565e-01 -8.13415766e-01 6.47004008e-01 2.86105424e-01 -1.80571586e-01 5.91851100e-02 5.31907499e-01 -1.15825856e+00 2.05696926e-01 7.06624568e-01 5.76197445e-01 -2.62036175e-01 5.86888731e-01 -4.77261186e-01 1.35341632e+00 -8.06485653e-01 -6.96003079e-01 -7.73600563e-02 3.17426361e-02 -1.15095341e+00 -3.37858111e-01 5.56426108e-01 -4.71110903e-02 -8.37951362e-01 8.25350821e-01 4.08490539e-01 4.08217609e-01 5.70509851e-01 -1.03563619e+00 -1.00718343e+00 1.36034334e+00 -3.44474018e-01 4.17388171e-01 2.74383485e-01 -4.17468935e-01 1.64677930e+00 -9.49681342e-01 1.18033254e+00 1.04292512e+00 9.22792315e-01 2.31834188e-01 -1.12558377e+00 -5.47911286e-01 4.43916440e-01 1.02275707e-01 -1.42072070e+00 -4.35509145e-01 4.19387639e-01 -1.70702666e-01 1.91974568e+00 -7.21651092e-02 3.56549472e-01 6.64867342e-01 -1.97110698e-02 1.05335557e+00 2.21644163e-01 -3.63161772e-01 -3.66402864e-02 -1.68254271e-01 6.76782966e-01 1.26872194e+00 3.43945414e-01 -1.26707971e-01 -3.64737540e-01 -5.83492637e-01 6.74387038e-01 -2.23841891e-01 4.88916300e-02 -1.52784213e-01 -1.03283060e+00 9.95583415e-01 9.31790292e-01 5.12796342e-01 -5.76924860e-01 3.28234673e-01 6.59994662e-01 2.88779348e-01 3.63082051e-01 6.12598181e-01 -7.30083585e-01 1.32005543e-01 -8.54024053e-01 4.55505624e-02 1.38323581e+00 1.21924436e+00 6.58980966e-01 1.63120002e-01 -3.95961165e-01 6.24618530e-01 3.57552260e-01 1.30188182e-01 2.80420333e-01 -5.35102308e-01 6.08011842e-01 1.04859054e+00 -1.79353282e-01 -1.08831275e+00 -8.76189113e-01 -3.36457580e-01 -7.70610213e-01 -6.12786293e-01 -3.17257233e-02 -4.74519610e-01 -1.10198307e+00 1.69977784e+00 3.61110866e-01 4.36667889e-01 3.75182271e-01 2.47090623e-01 1.55113101e+00 6.89674079e-01 3.13397229e-01 -1.49346720e-02 1.41305327e+00 -1.14369321e+00 -7.70998955e-01 -1.00282036e-01 1.00151944e+00 -1.93203345e-01 2.56565750e-01 -3.76310647e-01 -1.03216159e+00 8.14889669e-02 -7.16226041e-01 -3.90808016e-01 -8.62113416e-01 4.29370552e-01 1.01967871e+00 5.84042192e-01 -1.27263737e+00 4.01475936e-01 -8.28434050e-01 -1.51071027e-01 -1.67046875e-01 7.07196832e-01 -5.38705885e-01 4.91055101e-01 -1.69760191e+00 9.09057498e-01 8.74495029e-01 3.93476129e-01 -5.54399133e-01 -5.66397846e-01 -1.46325636e+00 4.39775795e-01 3.02765995e-01 -5.92580616e-01 1.44842494e+00 -1.92889661e-01 -1.06847596e+00 7.23901272e-01 -4.82582241e-01 -7.84178674e-01 -1.59240812e-01 -2.76781376e-02 -6.54550135e-01 -4.54753675e-02 1.67849466e-01 5.06703615e-01 8.10526386e-02 -1.13715780e+00 -6.20189786e-01 -1.14078246e-01 1.96306944e-01 -5.56437001e-02 -2.53277868e-01 1.47413239e-01 -4.17584151e-01 -5.33915818e-01 2.09864713e-02 -8.60049903e-01 -5.90981364e-01 -7.64362395e-01 -1.15631783e+00 -8.04392338e-01 6.66564882e-01 -4.35893476e-01 1.52528894e+00 -1.86116982e+00 -1.00194082e-01 3.80167931e-01 6.22842371e-01 1.89909354e-01 -1.68654293e-01 7.69915402e-01 -2.52848864e-01 5.12133181e-01 -2.79288515e-02 -4.94087756e-01 2.03970298e-01 -4.59972173e-02 -3.46368641e-01 3.37571919e-01 3.24958980e-01 1.27243412e+00 -8.24053705e-01 -5.74703634e-01 -2.89507926e-01 6.27997756e-01 -2.93337971e-01 2.52178878e-01 -1.46253958e-01 -4.65961277e-01 -4.22096819e-01 4.54966962e-01 2.53751546e-01 -7.61344850e-01 6.30873501e-01 -1.30990252e-01 -1.21534251e-01 9.09354985e-01 -1.35763192e+00 1.19571185e+00 -4.44810599e-01 5.97949862e-01 -3.85279745e-01 -4.92282569e-01 1.06542909e+00 7.10568964e-01 4.19211462e-02 -2.91985750e-01 -1.42141610e-01 1.44935042e-01 -3.87829334e-01 -4.21939462e-01 9.97153044e-01 2.95366615e-01 -5.18884242e-01 4.68041241e-01 4.51058537e-01 7.17868686e-01 5.02733052e-01 7.53872573e-01 1.49962533e+00 -4.00168806e-01 7.30991542e-01 2.28069380e-01 3.18622738e-01 -2.69553185e-01 8.21686089e-01 1.04935217e+00 4.16423157e-02 2.30389647e-02 7.48408854e-01 -5.91534674e-01 -7.84842849e-01 -1.01948941e+00 1.02580093e-01 1.59666204e+00 -2.06349060e-01 -7.64385045e-01 -3.89191836e-01 -9.21749353e-01 1.08156996e-02 6.12516522e-01 -9.59117174e-01 1.93268970e-01 -9.71391320e-01 -3.60039026e-01 1.15850568e+00 9.61929858e-01 2.43805632e-01 -1.62842083e+00 -3.49039793e-01 4.94456053e-01 -1.12817667e-01 -1.24132228e+00 -3.68756920e-01 6.29933953e-01 -7.41458356e-01 -7.85043240e-01 -2.94405073e-01 -1.42392123e+00 5.10614216e-01 -2.64151573e-01 1.61304235e+00 4.14594769e-01 -1.42193839e-01 9.87064615e-02 -1.69679895e-01 -1.41540477e-02 -2.99916863e-01 1.14681375e+00 -1.55488312e-01 -5.28746068e-01 5.40281773e-01 -4.65495706e-01 -6.96767271e-02 -2.29864046e-01 -7.13299096e-01 -5.82666993e-01 6.20973587e-01 7.91420996e-01 2.57132620e-01 -3.14953774e-01 6.61128104e-01 -1.56931746e+00 6.49602830e-01 -7.90170133e-01 -5.40899813e-01 6.50013387e-01 -3.08658957e-01 4.46810454e-01 5.08466363e-01 -1.77487865e-01 -1.05335021e+00 3.20467323e-01 -2.29143783e-01 -1.45261154e-01 -3.19941491e-01 1.08529258e+00 1.21759281e-01 1.34788767e-01 5.09988725e-01 5.34688868e-02 -7.52438068e-01 -3.57792765e-01 7.77443886e-01 4.15423036e-01 6.11197293e-01 -1.80798605e-01 7.08579123e-01 2.98635662e-02 -2.16854021e-01 -8.17348719e-01 -7.25339353e-01 -7.05102205e-01 -9.18036461e-01 1.02652691e-01 6.34217858e-01 -1.01768315e+00 -1.07665122e+00 1.30328014e-01 -1.59464419e+00 -1.00859683e-02 -1.09256759e-01 2.39775330e-01 -3.99786010e-02 1.58285096e-01 -1.39222860e+00 -8.50366652e-01 -6.37218654e-01 -7.36713707e-01 1.00111759e+00 3.55188400e-01 -2.33815461e-01 -1.31263947e+00 4.51276243e-01 -3.88204217e-01 2.45932415e-01 1.22954242e-01 1.10003483e+00 -1.49539661e+00 -7.38094389e-01 -4.48631942e-01 -2.61178166e-01 -6.33592904e-01 -5.44614419e-02 1.95400670e-01 -6.20282471e-01 -3.44368182e-02 -8.02765787e-01 -4.14232194e-01 1.21850204e+00 1.69183105e-01 3.67037475e-01 -7.67785847e-01 -7.19017804e-01 3.56016785e-01 1.30662346e+00 -4.47277203e-02 5.31736970e-01 5.59937656e-01 8.31494510e-01 5.48032284e-01 4.98123020e-02 3.78124297e-01 7.91868925e-01 2.21534982e-01 3.64202708e-01 -2.65833825e-01 2.83758968e-01 -4.28229362e-01 2.08293885e-01 8.99183214e-01 2.09665701e-01 -4.66250241e-01 -1.19155478e+00 1.01841855e+00 -1.90410125e+00 -1.06836772e+00 -2.90688038e-01 1.46445346e+00 7.77640581e-01 -1.85242516e-03 2.62911201e-01 -3.18388075e-01 9.96379077e-01 3.97108644e-01 -4.67095137e-01 -5.53356290e-01 -1.39756113e-01 1.18233792e-01 6.42733812e-01 4.22674537e-01 -1.45501816e+00 1.34353900e+00 7.82170105e+00 2.31381819e-01 -6.95801377e-01 -1.65565029e-01 1.68276638e-01 3.37161899e-01 -4.44221348e-01 -3.17210346e-01 -1.24659336e+00 -2.94945985e-01 1.30376065e+00 -5.02884507e-01 -1.08913377e-01 1.01601756e+00 -5.82333624e-01 4.28775102e-01 -9.83746290e-01 2.73729861e-01 1.31317288e-01 -1.55131805e+00 4.47881669e-02 -1.06197819e-01 6.31807506e-01 4.99124974e-01 -1.52022755e-02 6.11329854e-01 9.99032795e-01 -9.02770340e-01 2.90598065e-01 2.61384249e-01 3.83019686e-01 -9.59908903e-01 8.61274123e-01 7.29231164e-02 -1.82222784e+00 -1.15477778e-01 -1.91287935e-01 2.03876853e-01 3.41101438e-01 6.02919400e-01 -1.25160253e+00 6.37312472e-01 4.70438093e-01 5.86818576e-01 -6.40348017e-01 1.13245761e+00 -6.44174099e-01 5.50622106e-01 -3.27542394e-01 -3.31170470e-01 4.49429065e-01 4.32799131e-01 4.47355241e-01 1.90762150e+00 -2.87137479e-01 2.78439879e-01 4.36183631e-01 7.13000774e-01 -8.43472064e-01 1.47833049e-01 -8.30691874e-01 -9.89316255e-02 9.20367956e-01 1.40530300e+00 -7.40413487e-01 -5.17587483e-01 -3.16235572e-01 7.42531300e-01 1.14147747e+00 3.14255655e-01 -5.59563339e-01 -1.00873542e+00 2.50654936e-01 -6.35185361e-01 8.67537677e-01 -2.20150873e-01 1.24790691e-01 -1.19612646e+00 -1.30466580e-01 -4.58312422e-01 8.92693698e-01 -1.41176909e-01 -1.50469971e+00 9.57296371e-01 -2.87281036e-01 -5.05408764e-01 -7.31546760e-01 -3.86215180e-01 -5.92923999e-01 4.89796072e-01 -1.39410138e+00 -1.01985013e+00 3.41046631e-01 3.74604583e-01 3.40380669e-02 -5.26467860e-02 1.20533168e+00 3.32870454e-01 -8.01528931e-01 8.92187476e-01 -5.04290126e-02 1.18031549e+00 2.61896580e-01 -1.52133703e+00 1.33342886e+00 8.71454418e-01 5.64215064e-01 1.15079772e+00 4.74120289e-01 -6.90882444e-01 -1.16989970e+00 -1.24886191e+00 1.70849383e+00 -2.26393819e-01 8.67722213e-01 -4.12584066e-01 -1.04009593e+00 1.48509967e+00 4.69773352e-01 7.73771554e-02 9.72220242e-01 9.16507781e-01 -7.74603009e-01 5.38575888e-01 -9.08096492e-01 4.59044248e-01 9.73817408e-01 -7.87205935e-01 -9.31989670e-01 2.66912252e-01 1.23822069e+00 -6.57580376e-01 -1.17286193e+00 4.30038065e-01 3.96619141e-01 -4.68547225e-01 7.67572939e-01 -1.01358378e+00 2.44772792e-01 -8.04128572e-02 1.24395385e-01 -1.07575643e+00 -5.29020429e-01 -4.22172487e-01 -7.49746144e-01 1.32903337e+00 1.21703720e+00 -5.75549483e-01 8.96750629e-01 7.63298392e-01 5.35927340e-02 -7.87582338e-01 -9.24383581e-01 -6.33506775e-01 -1.37121245e-01 1.07084326e-02 5.66348433e-01 1.17649138e+00 4.72403467e-01 7.74949431e-01 -1.88368142e-01 5.64807415e-01 3.65825593e-01 2.30068654e-01 3.22733402e-01 -1.34098291e+00 2.69731469e-02 -1.82156920e-01 -5.91437578e-01 -1.03977489e+00 8.15497041e-01 -1.06049931e+00 1.31848589e-01 -1.77579987e+00 1.89985320e-01 -3.44181299e-01 -4.44267601e-01 1.00238454e+00 -2.25983858e-01 -7.50565454e-02 2.63415784e-01 1.41511252e-02 -9.92447138e-01 3.52217466e-01 2.48419791e-01 -2.96367884e-01 -5.32545507e-01 -2.23867491e-01 -4.12557781e-01 5.36489606e-01 5.27717829e-01 -8.27770174e-01 1.64464369e-01 -4.35032845e-01 5.90051472e-01 2.72725552e-01 -6.77607879e-02 -5.46558559e-01 9.17129278e-01 4.22999322e-01 2.09860533e-01 -1.02232111e+00 6.61198348e-02 -3.25113952e-01 -1.93492115e-01 2.72320718e-01 -8.44727635e-01 4.79007125e-01 1.15097567e-01 6.00115478e-01 -3.24833572e-01 -2.43812621e-01 1.88788757e-01 -2.60531843e-01 -7.67717004e-01 3.50992739e-01 -5.09696960e-01 2.95493543e-01 7.49260545e-01 3.60877007e-01 -5.67832351e-01 -3.53293657e-01 -9.20582592e-01 4.60169494e-01 -1.47424206e-01 4.46045637e-01 6.16829872e-01 -1.24636066e+00 -6.27946615e-01 -1.95798025e-01 4.61606272e-02 -2.77971923e-01 -4.29142751e-02 3.25107127e-01 -2.42771149e-01 6.84145987e-01 3.94344240e-01 -1.71508148e-01 -1.37762582e+00 7.18228877e-01 3.83294851e-01 -9.42841768e-01 -7.08607733e-01 8.89031529e-01 -2.55273491e-01 -1.09239852e+00 6.26596451e-01 -1.30920872e-01 -6.51772916e-01 3.08109194e-01 6.05731308e-01 3.54975224e-01 1.92189187e-01 -8.34121168e-01 -5.69952786e-01 -3.95915024e-02 -5.74597001e-01 1.58793367e-02 1.41059971e+00 1.88491434e-01 -4.43993658e-01 7.62174666e-01 1.41552806e+00 -6.67864457e-02 -2.48161674e-01 -5.37322044e-01 6.53434157e-01 2.65768230e-01 -2.79442370e-01 -5.42761147e-01 -1.19771826e+00 3.77627462e-01 1.01495549e-01 5.61424196e-01 6.22770905e-01 2.70787716e-01 7.86056519e-01 1.34637737e+00 1.46715134e-01 -7.64106512e-01 -3.75762582e-01 1.14430869e+00 3.13033342e-01 -8.63931060e-01 -1.35831818e-01 -6.26307130e-01 -6.01758063e-01 1.23603010e+00 6.49192750e-01 -2.55770057e-01 6.61882699e-01 5.34741879e-01 1.58644050e-01 -6.33169293e-01 -1.18047833e+00 -3.96198571e-01 1.74221933e-01 3.09308350e-01 8.96545947e-01 -6.46142988e-03 5.80856316e-02 4.30719048e-01 -1.68905407e-01 -2.78705299e-01 6.53209507e-01 1.06136763e+00 -4.39397216e-01 -8.86405766e-01 1.85339227e-02 5.47280312e-01 -7.18476236e-01 -5.83324075e-01 -5.90948701e-01 8.88069153e-01 -3.52029532e-01 8.65005672e-01 1.21906936e-01 -1.90319419e-01 5.61221719e-01 3.63408715e-01 -1.29862756e-01 -8.30978334e-01 -1.25363374e+00 -4.66682434e-01 4.66963321e-01 -4.67092901e-01 -5.20642817e-01 -5.46721280e-01 -1.72459507e+00 -2.71340817e-01 -8.07279646e-01 5.40027022e-01 2.61886418e-01 7.95041919e-01 6.30975664e-01 5.36997020e-01 6.17630541e-01 -5.26929379e-01 -1.69656098e-01 -8.47726524e-01 -6.82674110e-01 7.91373372e-04 5.89344561e-01 -4.35336769e-01 -1.61047757e-01 -2.28423953e-01]
[9.536746978759766, 9.376058578491211]
1233cc9f-fcf7-4dff-8778-1aff13e22dd1
document-structure-measure-for-hypernym
1811.12728
null
http://arxiv.org/abs/1811.12728v1
http://arxiv.org/pdf/1811.12728v1.pdf
Document Structure Measure for Hypernym discovery
Hypernym discovery is the problem of finding terms that have is-a relationship with a given term. We introduce a new context type, and a relatedness measure to differentiate hypernyms from other types of semantic relationships. Our Document Structure measure is based on hierarchical position of terms in a document, and their presence or otherwise in definition text. This measure quantifies the document structure using multiple attributes, and classes of weighted distance functions.
['Aswin Kannan', 'Shanmukha C Guttula', 'Hima P Karanam', 'Balaji Ganesan', 'Arun Kumar']
2018-11-30
null
null
null
null
['hypernym-discovery']
['natural-language-processing']
[ 2.56829053e-01 3.01349163e-02 -5.50479949e-01 -5.50947905e-01 3.54478389e-01 -8.45425010e-01 7.56621897e-01 8.39569628e-01 -4.87449050e-01 6.73592985e-01 4.30813581e-01 -4.10035908e-01 -9.01516676e-01 -1.32402420e+00 2.30723992e-01 -4.17041034e-01 -3.41397256e-01 7.23825037e-01 1.48111150e-01 -5.79760909e-01 4.78791595e-01 2.50576943e-01 -1.65079391e+00 1.39954388e-01 7.43540347e-01 7.41635382e-01 5.33363670e-02 9.81891900e-02 -8.60334396e-01 7.07307756e-01 -8.13325465e-01 -4.25581157e-01 -1.97555777e-02 -2.77152359e-01 -1.29522669e+00 -4.38502908e-01 2.25695759e-01 3.34011167e-01 -2.30400473e-01 1.32990539e+00 1.00289825e-02 3.60012591e-01 7.44979680e-01 -1.47164237e+00 -6.77098334e-01 7.87994444e-01 -2.53356576e-01 3.45315725e-01 9.79900181e-01 -8.88967514e-01 1.76560485e+00 -5.31224847e-01 1.04533565e+00 1.33471763e+00 5.79850674e-01 2.08675206e-01 -9.75309014e-01 -5.03683031e-01 -1.63135514e-01 4.63386029e-01 -1.54556382e+00 2.57912129e-01 2.39749059e-01 -4.10177112e-01 1.40917802e+00 5.08774817e-01 5.86090505e-01 3.07208568e-01 -7.91321024e-02 -2.15779960e-01 7.73477137e-01 -1.00081325e+00 1.48138329e-01 -1.02239229e-01 8.53020608e-01 6.00999594e-01 9.84298766e-01 -3.73027354e-01 -2.68507063e-01 -6.68420315e-01 2.01323092e-01 -7.42677227e-02 -2.48805001e-01 -1.49893165e-01 -9.17442679e-01 8.46491098e-01 1.38042778e-01 1.08980536e+00 -5.58985323e-02 1.35164365e-01 3.86686355e-01 3.80499750e-01 5.05162142e-02 1.06662428e+00 -3.41303468e-01 2.92475149e-02 -6.60805821e-01 3.09248179e-01 1.13592458e+00 1.20500207e+00 7.91726291e-01 -6.47046089e-01 6.22845255e-03 9.43148792e-01 2.84618646e-01 1.90632537e-01 8.10506999e-01 -7.96186090e-01 9.56961438e-02 1.20090973e+00 -2.32968941e-01 -1.22616386e+00 -4.02086854e-01 -8.10452998e-02 -1.78699985e-01 -2.16580585e-01 -2.11691320e-01 5.63399315e-01 -7.25240290e-01 1.80491638e+00 3.42924237e-01 -1.74446583e-01 8.98795426e-02 4.34795827e-01 1.19162130e+00 3.55955809e-01 9.29747745e-02 -5.49922764e-01 1.60166276e+00 -3.49542797e-01 -1.19397163e+00 2.70566285e-01 9.71742392e-01 -6.03092134e-01 7.95135081e-01 2.07286309e-02 -7.68514037e-01 1.08239539e-01 -1.06915295e+00 -2.49106586e-01 -1.29647410e+00 -8.16870272e-01 7.75953710e-01 6.53565943e-01 -8.80730569e-01 5.67509830e-01 1.52836561e-01 -8.07738543e-01 -1.81349844e-01 1.65191323e-01 -2.82926232e-01 -6.95691584e-03 -1.91823387e+00 1.25741601e+00 9.86208320e-01 -7.26782322e-01 4.04765308e-02 -5.24576545e-01 -1.01909518e+00 2.51209706e-01 5.24101615e-01 -8.62396002e-01 7.63773024e-01 -3.41193348e-01 -2.99454302e-01 1.34729731e+00 -1.61376089e-01 -3.46504897e-01 -5.20071328e-01 3.39221716e-01 -1.27544522e+00 1.43719897e-01 4.28888917e-01 7.61431735e-03 3.23239446e-01 -1.29557562e+00 -8.79567087e-01 -4.24030036e-01 5.19207478e-01 3.20528120e-01 -5.95087826e-01 3.20112348e-01 -2.09359735e-01 -6.52647018e-01 3.62825334e-01 -4.94859397e-01 5.93202561e-02 -4.33528244e-01 -3.16063523e-01 -6.18320048e-01 6.01467371e-01 -2.31261551e-01 1.95156789e+00 -1.66961789e+00 -2.54672229e-01 8.89952719e-01 7.54821599e-01 1.88022070e-02 5.34841679e-02 7.82842457e-01 -5.08272350e-01 6.79230750e-01 -2.85324365e-01 6.54470623e-01 1.38799772e-01 6.94265604e-01 -3.82898122e-01 -1.97509199e-01 -6.59954011e-01 6.33907557e-01 -1.36948478e+00 -6.52875960e-01 -1.56841457e-01 3.45507413e-02 4.12238250e-03 -7.19409958e-02 -1.22188762e-01 -6.88538492e-01 -1.82726860e-01 7.51625240e-01 3.45788091e-01 -1.92580476e-01 6.50352716e-01 -2.77157396e-01 -5.58584509e-03 6.93663836e-01 -9.82895136e-01 1.40873992e+00 -3.59947979e-01 4.16064948e-01 -4.34852391e-01 -7.58092344e-01 9.05750275e-01 4.81273264e-01 7.04284310e-01 -6.06245279e-01 1.83895543e-01 4.56496924e-01 3.03900599e-01 -8.47971678e-01 8.48789930e-01 -2.27880478e-01 -8.60988721e-02 5.74708223e-01 2.72196997e-02 -2.00775966e-01 9.53506827e-01 4.43383962e-01 1.59405208e+00 -4.37250733e-01 1.11600244e+00 -5.97080648e-01 4.44079041e-01 1.02410600e-01 4.09101367e-01 5.73310494e-01 2.44573921e-01 -5.80044538e-02 3.57175082e-01 -3.37759852e-01 -9.15776193e-01 -1.13244486e+00 -6.13905191e-01 1.18943679e+00 4.74701226e-01 -1.42801905e+00 -2.15261504e-01 -5.52783906e-01 5.31769574e-01 8.59802783e-01 -7.50806093e-01 -1.69010624e-01 -1.84365645e-01 -5.08437097e-01 7.39914119e-01 2.19297901e-01 -1.18702821e-01 -9.18933153e-01 -5.92789233e-01 -3.37943295e-03 -1.48923576e-01 -9.17371929e-01 -2.43270203e-01 3.26452672e-01 -6.73492610e-01 -1.32828760e+00 -2.10339110e-02 -9.56595004e-01 5.83631754e-01 3.39112431e-01 1.63660574e+00 6.13068640e-01 -5.89078546e-01 4.13945705e-01 -9.32679713e-01 -1.50615439e-01 -1.63876683e-01 -7.57391304e-02 1.55739278e-01 -8.23818028e-01 8.55272472e-01 -6.90939963e-01 -1.37705654e-01 1.23365492e-01 -1.34519756e+00 -5.86090028e-01 -8.37689564e-02 6.38986409e-01 3.49097341e-01 6.96994066e-01 4.54405434e-02 -9.67134476e-01 1.24759901e+00 -5.57012975e-01 -8.27824101e-02 8.94751310e-01 -1.42791319e+00 3.84777516e-01 -5.47539741e-02 -1.90659195e-01 -5.71890712e-01 -6.47942126e-01 2.84467846e-01 1.43555984e-01 1.68993711e-01 1.07351696e+00 -2.10186765e-01 1.20569684e-01 6.36608660e-01 -2.23154053e-01 -6.10453606e-01 -2.74824798e-01 8.52447093e-01 5.89944243e-01 4.37075734e-01 -8.41073990e-01 4.99844432e-01 3.15738469e-01 4.64470744e-01 -6.64221823e-01 -7.60345817e-01 -1.11895382e+00 -6.78782463e-01 5.99383712e-02 6.20032310e-01 -2.47633502e-01 -7.42853880e-01 -4.87186342e-01 -1.22978008e+00 7.97513425e-01 -3.65053862e-01 3.65385234e-01 4.97151352e-03 5.89211583e-01 -1.30152881e-01 -7.41370797e-01 -5.50785005e-01 -1.49311051e-01 3.96138966e-01 -6.29928410e-02 -8.52494419e-01 -1.20439231e+00 4.42169547e-01 3.58452350e-02 1.81281626e-01 3.14525694e-01 1.71691597e+00 -1.19445884e+00 4.81589764e-01 -3.79808456e-01 -1.05657555e-01 -1.06018350e-01 5.89301467e-01 1.18930884e-01 -3.83946329e-01 1.58885255e-01 2.25182716e-02 2.14569092e-01 7.17807174e-01 -1.10246785e-01 9.88828063e-01 -7.11738408e-01 -7.80447364e-01 2.41970152e-01 1.66834974e+00 7.97994077e-01 4.82887357e-01 6.89791977e-01 7.06221044e-01 8.99083734e-01 5.27363241e-01 4.13410693e-01 2.70559281e-01 5.71311712e-01 1.83999494e-01 4.41860318e-01 2.63485372e-01 -9.77291614e-02 -4.71872240e-01 8.97624254e-01 -1.57952249e-01 -3.76688391e-01 -1.24198937e+00 5.59044063e-01 -1.82696497e+00 -1.31359303e+00 -4.84470457e-01 2.19527960e+00 1.25024366e+00 7.31883943e-03 3.20404992e-02 3.42789829e-01 1.08110237e+00 -1.01430133e-01 -8.59248564e-02 -5.18229723e-01 -3.57664943e-01 4.15588468e-01 3.85652035e-01 7.88987160e-01 -6.00361407e-01 9.31644380e-01 7.53219604e+00 8.19039583e-01 -2.75359839e-01 -9.57845449e-02 -4.05263692e-01 5.54759353e-02 -8.08116913e-01 3.80702972e-01 -6.24537528e-01 3.32073092e-01 5.21269798e-01 -9.04055357e-01 3.03221673e-01 7.79142618e-01 -4.69772696e-01 2.35065222e-02 -1.16898203e+00 9.38544035e-01 3.41136336e-01 -9.59322989e-01 5.05085588e-01 1.23912066e-01 4.15110767e-01 -5.08131623e-01 -4.52750027e-01 -1.05867125e-01 4.83027726e-01 -1.13235259e+00 1.50072038e-01 2.72052377e-01 7.15132952e-01 -6.65458500e-01 5.41700304e-01 -2.04342231e-01 -1.54097438e+00 -2.92132869e-02 -3.55728120e-01 -7.94913322e-02 1.56192482e-01 9.67750728e-01 -7.83391953e-01 6.23314202e-01 6.96023822e-01 6.91721141e-01 -2.55213648e-01 1.09399092e+00 -5.57499409e-01 6.63602054e-02 -2.22410113e-01 -2.68968761e-01 8.55268538e-02 -3.95851135e-01 7.41478920e-01 1.49555278e+00 3.62336606e-01 2.28050143e-01 -2.01080859e-01 7.51468301e-01 -1.17568605e-01 3.30879778e-01 -7.93015242e-01 -1.84365019e-01 1.16380012e+00 1.43838573e+00 -7.66839623e-01 -6.16064012e-01 -3.32845300e-01 6.47988796e-01 -6.91697523e-02 6.42278641e-02 -5.08817136e-01 -1.16128230e+00 8.35788965e-01 7.03790486e-02 -2.82952338e-01 -3.94612402e-02 -3.61276895e-01 -8.42026114e-01 5.95626496e-02 -5.27673185e-01 1.02898240e+00 -7.34969079e-01 -1.61496687e+00 6.07764363e-01 3.00805956e-01 -1.18518114e+00 -2.18750522e-01 -8.13216984e-01 -4.41037744e-01 6.88458383e-01 -1.04945445e+00 -6.09930575e-01 -3.36705476e-01 6.32283568e-01 -1.31508738e-01 -8.48383233e-02 1.27352357e+00 1.86968952e-01 4.99760509e-02 4.11659598e-01 -1.77349180e-01 5.74079640e-02 5.43413877e-01 -1.61414373e+00 2.08688974e-01 4.05406684e-01 2.47321859e-01 1.38725436e+00 8.96349669e-01 -9.33462083e-01 -8.04637730e-01 -5.54103434e-01 1.71386611e+00 -4.55739290e-01 9.42566872e-01 -5.51768718e-03 -1.01856625e+00 4.23587203e-01 3.05169851e-01 -3.43247205e-01 1.34512186e+00 6.30137384e-01 -1.11795926e+00 1.62336960e-01 -1.33130574e+00 4.82946575e-01 1.69796979e+00 -7.62261450e-01 -1.62157059e+00 3.87010932e-01 9.92180049e-01 3.40144515e-01 -1.23252749e+00 6.37101948e-01 6.80715501e-01 -3.51193428e-01 1.10576546e+00 -8.66665125e-01 1.35801479e-01 -4.82591897e-01 -4.76522535e-01 -1.13604152e+00 -7.12098539e-01 -4.72132474e-01 -4.75038052e-01 1.31822872e+00 6.48786545e-01 -5.43021142e-01 2.22613111e-01 7.49648035e-01 1.15387850e-01 -3.48244220e-01 -8.74161959e-01 -1.41745126e+00 -6.57476261e-02 -1.49118751e-01 1.07334590e+00 1.85395777e+00 1.15796018e+00 6.75023139e-01 3.17282647e-01 -3.20615739e-01 5.19632220e-01 2.50277340e-01 -3.33112210e-01 -1.87398839e+00 1.49179086e-01 -9.29899812e-01 -7.83965826e-01 -2.07606316e-01 3.16478074e-01 -1.38566709e+00 -2.90294170e-01 -1.80806935e+00 2.83634186e-01 -3.02271247e-01 -4.80201840e-01 6.35390103e-01 1.27174775e-03 -2.37125695e-01 -3.03033829e-01 3.80456150e-01 -4.98015761e-01 1.76183522e-01 8.42454314e-01 -1.78809300e-01 4.62100469e-03 -6.07539117e-01 -5.56586325e-01 7.22081602e-01 6.08315289e-01 -8.32958758e-01 -5.85494757e-01 -1.22641869e-01 9.84288573e-01 -3.66510749e-01 -2.17827901e-01 -5.31840146e-01 2.97731191e-01 -4.41171765e-01 -2.61497617e-01 -2.03933567e-01 4.64124121e-02 -1.17902100e+00 1.94805101e-01 5.88366032e-01 -6.92350924e-01 3.34767908e-01 -3.35928559e-01 1.58632383e-01 -3.35926980e-01 -1.06358194e+00 2.45077521e-01 -2.42232412e-01 -1.02292895e+00 -2.79209893e-02 -1.66351706e-01 2.42825851e-01 7.87599564e-01 -2.80611336e-01 -5.69451511e-01 9.91597399e-02 -5.73108613e-01 1.39547840e-01 5.56131303e-01 7.01486945e-01 5.94892919e-01 -1.68015587e+00 -3.09236020e-01 -5.68594038e-01 8.26185763e-01 -4.73067194e-01 -8.16939950e-01 1.43940866e-01 -4.61417556e-01 4.20656770e-01 2.55281609e-02 2.04555035e-01 -1.76764858e+00 7.38055646e-01 2.06926405e-01 -1.60269678e-01 -3.90912443e-01 5.98943293e-01 -2.27677405e-01 -3.28775972e-01 2.83619881e-01 -4.37735379e-01 -7.46399283e-01 3.91830742e-01 6.11255467e-01 5.89009523e-01 4.08574231e-02 -6.56081259e-01 -8.01059544e-01 6.43000662e-01 1.26557589e-01 -3.12446862e-01 8.08394849e-01 -1.17031755e-02 -1.16027319e+00 7.51739502e-01 1.14220119e+00 -2.33263269e-01 5.63451588e-01 -4.95098144e-01 7.97139525e-01 -6.12999439e-01 -1.94801316e-01 -9.54695940e-01 -3.90263468e-01 8.05212334e-02 3.03191066e-01 9.47519720e-01 8.58548939e-01 3.04610163e-01 6.63233757e-01 7.59309411e-01 3.16756696e-01 -1.41189754e+00 -2.06582785e-01 1.01216805e+00 7.59771109e-01 -6.63255751e-01 2.86589831e-01 -8.45343769e-01 -1.90337718e-01 1.10051537e+00 3.83889377e-01 4.07578200e-01 9.88056183e-01 3.17200691e-01 -1.63841471e-01 -8.75762045e-01 -4.62295353e-01 -7.76828229e-01 6.00452483e-01 5.90245724e-01 1.00227213e+00 8.29101130e-02 -1.37684345e+00 5.61730266e-01 -5.97599268e-01 -3.44270796e-01 6.93608448e-02 1.19225860e+00 -8.42652678e-01 -1.27570200e+00 -7.61908293e-02 6.20708227e-01 -2.75685340e-01 -6.47639036e-01 -9.29634809e-01 5.63193619e-01 5.69218874e-01 1.25609303e+00 1.22681491e-01 -6.31400108e-01 2.46432006e-01 1.59739256e-01 5.50171316e-01 -9.50073659e-01 -5.30340374e-01 -5.96917987e-01 4.35752511e-01 -1.96187496e-01 -7.45504320e-01 -2.36289963e-01 -1.77824891e+00 -3.11222047e-01 -6.47204399e-01 6.57155991e-01 6.49819016e-01 1.20502961e+00 -3.50704882e-03 2.46615753e-01 4.15226847e-01 3.57636899e-01 4.86599840e-02 -9.07258272e-01 -9.99437869e-01 8.97382915e-01 -2.35260487e-01 -8.39855194e-01 -4.40010160e-01 -1.16997987e-01]
[9.87111759185791, 8.749289512634277]
b01e3dce-2107-4639-801b-d0ad94460d8e
synthesizing-a-progression-of-subtasks-for
2305.17518
null
https://arxiv.org/abs/2305.17518v1
https://arxiv.org/pdf/2305.17518v1.pdf
Synthesizing a Progression of Subtasks for Block-Based Visual Programming Tasks
Block-based visual programming environments play an increasingly important role in introducing computing concepts to K-12 students. In recent years, they have also gained popularity in neuro-symbolic AI, serving as a benchmark to evaluate general problem-solving and logical reasoning skills. The open-ended and conceptual nature of these visual programming tasks make them challenging, both for state-of-the-art AI agents as well as for novice programmers. A natural approach to providing assistance for problem-solving is breaking down a complex task into a progression of simpler subtasks; however, this is not trivial given that the solution codes are typically nested and have non-linear execution behavior. In this paper, we formalize the problem of synthesizing such a progression for a given reference block-based visual programming task. We propose a novel synthesis algorithm that generates a progression of subtasks that are high-quality, well-spaced in terms of their complexity, and solving this progression leads to solving the reference task. We show the utility of our synthesis algorithm in improving the efficacy of AI agents (in this case, neural program synthesizers) for solving tasks in the Karel programming environment. Then, we conduct a user study to demonstrate that our synthesized progression of subtasks can assist a novice programmer in solving tasks in the Hour of Code: Maze Challenge by Code-dot-org.
['Adish Singla', 'Maria Christakis', 'Hasan Ferit Eniser', 'Ahana Ghosh', 'Alperen Tercan']
2023-05-27
null
null
null
null
['logical-reasoning']
['reasoning']
[ 1.47326551e-02 1.06811523e-02 9.55139548e-02 -1.26412839e-01 -2.82697052e-01 -9.36912477e-01 3.92459184e-01 3.83897483e-01 -1.17453165e-01 2.55619854e-01 -1.63857594e-01 -8.87052536e-01 -9.47141871e-02 -8.78861725e-01 -7.92298257e-01 -2.57422894e-01 -1.22596994e-01 4.99518186e-01 2.07602903e-01 -5.52415967e-01 3.52779418e-01 5.52123249e-01 -1.72500145e+00 5.01662314e-01 1.31380844e+00 2.57012904e-01 3.55960250e-01 7.45487630e-01 -2.78927594e-01 1.15227652e+00 -6.83608174e-01 -3.03739667e-01 2.26629779e-01 -5.67950487e-01 -9.59132314e-01 -3.75122845e-01 3.93186539e-01 -3.66166234e-02 -7.69171864e-02 1.04124379e+00 4.03849222e-03 2.60198653e-01 6.34869754e-01 -1.74865496e+00 -5.21368682e-01 8.41922939e-01 -4.52568561e-01 7.91155547e-02 5.07151127e-01 4.40766156e-01 1.01702726e+00 -5.21133959e-01 5.26202679e-01 1.04702067e+00 5.94662428e-01 7.64627397e-01 -1.65724504e+00 -7.82835007e-01 2.42076546e-01 2.75355220e-01 -1.38500440e+00 -1.31469831e-01 8.02744627e-01 -8.58969152e-01 1.22353113e+00 2.79250443e-01 1.23301423e+00 5.29755175e-01 1.46734983e-01 8.93948972e-01 9.04756427e-01 -5.51858008e-01 4.71208006e-01 2.04468742e-02 3.07270288e-01 1.08067513e+00 6.08347319e-02 1.41138330e-01 -4.13449198e-01 -4.98684682e-02 8.58774126e-01 -1.75360888e-01 -7.74591863e-02 -7.03504324e-01 -1.22894907e+00 7.28657961e-01 7.84556270e-01 4.15687114e-01 -1.27400249e-01 5.25799632e-01 3.72523010e-01 4.20543253e-01 -3.12863469e-01 1.07453048e+00 8.00397247e-03 -2.29736432e-01 -9.31583047e-01 8.86657596e-01 9.01227117e-01 9.91296589e-01 4.61797178e-01 4.64803278e-01 -1.96722403e-01 4.82682437e-01 1.71142578e-01 2.32877046e-01 3.17239255e-01 -1.16406941e+00 2.98259765e-01 9.61108327e-01 -1.20261200e-01 -1.04600799e+00 -4.51488078e-01 -4.07908589e-01 -6.54148281e-01 9.69449580e-01 7.96133578e-01 7.67734181e-03 -7.57884681e-01 1.91881883e+00 2.23312959e-01 -1.47482809e-02 -1.32007487e-02 9.34282541e-01 7.64443576e-01 9.73719120e-01 2.79823929e-01 1.39910370e-01 1.54782891e+00 -1.31133413e+00 -3.01405132e-01 -4.33685571e-01 8.24962974e-01 -1.85875669e-01 1.36442173e+00 6.84687316e-01 -1.46619952e+00 -6.77696168e-01 -1.18573153e+00 -1.73285931e-01 -2.01976120e-01 -2.00368091e-01 8.46388996e-01 5.74219644e-01 -1.20421481e+00 4.54761654e-01 -7.55531013e-01 5.38035780e-02 3.83321822e-01 2.56598771e-01 -4.54974845e-02 -1.29158989e-01 -7.38826096e-01 1.13123989e+00 4.42647994e-01 -4.02187645e-01 -1.14298439e+00 -1.18183839e+00 -9.14984584e-01 6.47441864e-01 4.17984962e-01 -7.40830600e-01 1.56234992e+00 -1.03358877e+00 -1.15613842e+00 8.24814081e-01 -1.34994492e-01 -2.85856128e-01 3.12130392e-01 3.92003149e-01 1.65745527e-01 -1.13167152e-01 -9.67403874e-02 7.44973063e-01 6.79693282e-01 -1.34276044e+00 -4.98269349e-01 -8.66754875e-02 7.89891720e-01 3.97853911e-01 -1.89108998e-02 -1.63501892e-02 -2.04390034e-01 -6.27748132e-01 -5.82440495e-02 -8.96571994e-01 -4.09589529e-01 3.19854766e-01 2.25138757e-03 -4.08197492e-01 2.31920183e-01 -4.46717799e-01 1.28574884e+00 -2.16514349e+00 6.48531854e-01 3.22132975e-01 7.93927014e-01 1.07552640e-01 -1.49775460e-01 2.79437423e-01 -2.84305602e-01 1.35039315e-01 -3.78182292e-01 7.02720582e-02 2.68566936e-01 1.68544054e-02 -3.55315387e-01 -7.22920895e-02 -3.87302302e-02 1.18522894e+00 -1.23792827e+00 -3.63300323e-01 -9.71366744e-03 -5.73380757e-03 -9.27819252e-01 2.88661242e-01 -6.01452589e-01 2.85088688e-01 -1.14595130e-01 4.20647353e-01 2.33767316e-01 -2.96332002e-01 1.70865685e-01 2.40331337e-01 -7.43604228e-02 1.88494384e-01 -1.09416211e+00 1.78458607e+00 -6.45532250e-01 9.24345791e-01 9.07572731e-03 -9.21536565e-01 7.31666982e-01 1.18126392e-01 -1.94424480e-01 -7.53089070e-01 -2.28945017e-01 -5.72131574e-03 6.17668092e-01 -2.65725583e-01 5.05468905e-01 -1.89127401e-01 -9.57019106e-02 7.82350063e-01 -2.45361865e-01 -9.11453605e-01 7.62193561e-01 3.80564123e-01 1.22512567e+00 2.08748624e-01 3.43329489e-01 -3.46815228e-01 2.92019188e-01 5.90458751e-01 1.45185262e-01 8.84794593e-01 5.89348655e-03 2.26600423e-01 8.88594329e-01 -6.47491038e-01 -1.25430405e+00 -9.50730801e-01 3.69638115e-01 1.27087295e+00 -2.03096166e-01 -7.90541351e-01 -9.51611578e-01 -3.44241887e-01 -1.01011619e-01 8.38372886e-01 -4.63144720e-01 -2.69664794e-01 -7.69496024e-01 -2.18267933e-01 7.79609323e-01 6.77605510e-01 2.25036919e-01 -1.33350503e+00 -1.26248252e+00 1.71827555e-01 4.95680869e-02 -7.10556924e-01 -1.90335467e-01 3.39547843e-01 -7.21019745e-01 -8.98145378e-01 -6.84506714e-01 -1.17273545e+00 9.90679324e-01 2.59542048e-01 1.46869493e+00 7.82676101e-01 -3.48994166e-01 1.17674388e-01 -3.50792482e-02 -4.13720340e-01 -6.74472451e-01 -9.35706869e-02 -4.12134916e-01 -7.25802779e-01 4.82301340e-02 -6.84399486e-01 -1.26412824e-01 5.08944578e-02 -9.05385971e-01 8.42161179e-01 2.16069743e-01 7.15234458e-01 2.37137735e-01 2.77120173e-01 1.40296623e-01 -6.93218946e-01 1.10672462e+00 -3.34947377e-01 -1.02791643e+00 3.73268992e-01 -3.63268703e-01 2.10992739e-01 9.52956796e-01 -5.21185756e-01 -6.40646219e-01 9.83579531e-02 6.97247982e-02 -2.49555990e-01 -1.28907815e-01 1.01236391e+00 2.59223163e-01 -3.40669423e-01 1.17223525e+00 3.85330677e-01 -1.46769494e-01 1.08596526e-01 3.57556880e-01 6.37575537e-02 6.37717605e-01 -1.29017901e+00 1.13514423e+00 -1.80084586e-01 1.83461726e-01 -5.78952193e-01 -4.69130456e-01 5.01518138e-03 -1.40619546e-01 -1.78221405e-01 6.82768226e-01 -7.28021502e-01 -1.29326010e+00 3.36679399e-01 -1.34859490e+00 -9.11758840e-01 -2.32992500e-01 -1.23378769e-01 -5.74607134e-01 -3.39942016e-02 -4.39289302e-01 -5.39168179e-01 4.59036641e-02 -1.79426515e+00 4.05470908e-01 3.43428224e-01 -7.91031361e-01 -7.77524590e-01 1.23545222e-01 2.02978984e-01 5.40079594e-01 2.08507299e-01 1.68103707e+00 -2.46675789e-01 -6.60760462e-01 2.58019805e-01 -2.73983568e-01 -1.43448308e-01 -5.12465894e-01 9.99874398e-02 -6.31079257e-01 -3.93531881e-02 -3.34372282e-01 -5.44577837e-01 2.71086156e-01 1.20218761e-01 1.25114691e+00 -1.22423977e-01 -1.43539786e-01 6.09666049e-01 1.09502196e+00 3.89561385e-01 4.95451212e-01 3.99399191e-01 7.21932411e-01 6.81654513e-01 1.57636106e-01 1.78090975e-01 5.04687667e-01 9.01137948e-01 2.78803855e-01 1.55540779e-02 2.62979008e-02 -2.65771300e-01 1.28155723e-01 6.04767621e-01 -1.21773891e-01 3.04902971e-01 -1.74597323e+00 6.51865780e-01 -1.85478735e+00 -9.46657956e-01 -1.40248686e-01 1.79379547e+00 1.33476651e+00 1.70704424e-01 3.85611832e-01 2.89084196e-01 2.49717608e-01 -1.86444536e-01 -2.89192408e-01 -7.98505425e-01 6.10373020e-01 5.35438418e-01 -1.57761022e-01 5.67594945e-01 -5.31066895e-01 1.14453745e+00 5.90566444e+00 9.11402285e-01 -1.04074121e+00 -3.20781559e-01 2.92034596e-01 1.44201621e-01 -5.12881279e-01 -1.03718899e-02 -5.03319502e-01 2.06933871e-01 8.65131199e-01 -7.29313016e-01 1.10539436e+00 1.03918016e+00 -2.08328590e-01 -1.55418426e-01 -1.62573910e+00 9.91489172e-01 -1.09255783e-01 -1.73183048e+00 3.08502428e-02 -3.75920862e-01 7.35663593e-01 -6.18773103e-01 -2.46405676e-02 7.97449052e-01 7.12310016e-01 -1.60505497e+00 1.17508650e+00 2.90309638e-02 4.59822714e-01 -9.08768117e-01 -4.75559868e-02 6.77910268e-01 -1.19688320e+00 -1.23314351e-01 -1.07673883e-01 -6.00522101e-01 -1.36840150e-01 4.33481820e-02 -6.76715970e-01 1.19607717e-01 6.20683491e-01 2.59801716e-01 -6.19847894e-01 1.33859956e+00 -7.01630414e-01 1.71019152e-01 -8.66511837e-02 -3.37405294e-01 2.21135721e-01 1.23229615e-01 3.27780515e-01 9.88986075e-01 1.27121240e-01 4.83618617e-01 2.86546797e-01 1.56504142e+00 6.35454729e-02 -2.18418792e-01 -6.79448485e-01 -3.27618033e-01 4.91991580e-01 1.16359329e+00 -1.02470303e+00 -3.44641715e-01 -1.73178375e-01 5.01385629e-01 6.98973477e-01 4.54284668e-01 -8.64807904e-01 -7.20356286e-01 5.97849488e-01 1.32595086e-02 -2.03478485e-02 -6.91437244e-01 -9.80132639e-01 -1.01105809e+00 -6.13205917e-02 -1.56426835e+00 -5.88904209e-02 -1.29009831e+00 -6.04265749e-01 6.41325891e-01 2.90398628e-01 -9.14468110e-01 -3.63791019e-01 -6.61874115e-01 -7.42704213e-01 1.06217206e+00 -1.00330603e+00 -7.54638851e-01 -7.64502883e-01 4.55291718e-01 4.66619283e-01 -2.16952562e-01 9.30764616e-01 -3.20012905e-02 -2.58129925e-01 5.32448232e-01 -4.95655537e-01 3.67846936e-02 -4.43730168e-02 -1.34531069e+00 8.10198128e-01 9.14882064e-01 2.26731658e-01 9.87968147e-01 8.31129849e-01 -4.14388716e-01 -1.56752777e+00 -4.62981552e-01 6.38734341e-01 -3.49525958e-01 6.92447960e-01 -6.24731362e-01 -8.70540977e-01 6.38306677e-01 2.86158994e-02 -1.90179676e-01 3.61098707e-01 1.08348750e-01 -5.20792067e-01 1.56059638e-01 -8.08207989e-01 1.16655600e+00 1.07504022e+00 -5.85756719e-01 -1.03781676e+00 1.92029193e-01 7.37256587e-01 -9.37891185e-01 -4.88152832e-01 -9.23644230e-02 5.95871210e-01 -7.29525387e-01 1.17023993e+00 -6.68454945e-01 1.12145865e+00 -5.49874961e-01 2.42861807e-01 -1.61422038e+00 -4.40203816e-01 -5.29428661e-01 -5.68472371e-02 7.33095825e-01 2.86190510e-01 -1.42030671e-01 8.81713867e-01 7.71150827e-01 -2.95862973e-01 -7.45323956e-01 -3.04102868e-01 -6.67985976e-01 2.29061782e-01 -6.29840016e-01 4.60731566e-01 1.01134837e+00 5.60228467e-01 2.11429909e-01 1.87841102e-01 -8.98872465e-02 4.74959999e-01 4.05374438e-01 9.50023115e-01 -1.31350029e+00 -6.95503175e-01 -9.04833496e-01 -1.06648065e-01 -1.05336964e+00 4.12235171e-01 -1.30045545e+00 1.16011016e-01 -1.45937216e+00 7.69449249e-02 -7.67377079e-01 2.23294586e-01 8.39163899e-01 4.09456454e-02 1.03754707e-01 6.46455467e-01 6.15446568e-02 -3.19236428e-01 2.25198474e-02 1.34311318e+00 -3.08403045e-01 -2.82092422e-01 -2.95854926e-01 -7.62036681e-01 7.02753067e-01 4.90075827e-01 -5.66241503e-01 -7.03149855e-01 -6.49349511e-01 8.94906282e-01 3.68820846e-01 5.33468127e-01 -1.33512378e+00 5.22404790e-01 -6.25664711e-01 8.13408867e-02 -1.21066213e-01 1.18962601e-02 -9.23209727e-01 2.52379447e-01 7.45260417e-01 -6.36687696e-01 4.64696348e-01 6.86629236e-01 -2.08135337e-01 -1.22088484e-01 -6.10423386e-01 6.67507529e-01 -4.34286356e-01 -7.71223545e-01 -2.68904030e-01 -6.74827099e-01 3.30175549e-01 1.29019487e+00 -2.83167750e-01 -6.05330050e-01 -2.68376321e-01 -6.39884174e-01 4.10864115e-01 5.63829005e-01 9.52134132e-02 6.23845875e-01 -1.20269811e+00 -4.56233233e-01 1.99157745e-01 2.27034688e-01 2.09129155e-01 4.27529290e-02 6.78297579e-01 -1.24810719e+00 2.33189270e-01 -8.83669972e-01 -4.24839586e-01 -1.38691139e+00 5.87400138e-01 2.32824311e-01 -4.01232183e-01 -5.25126815e-01 9.92964566e-01 6.03963852e-01 -4.93217081e-01 6.29750133e-01 -6.82529509e-01 -1.71491206e-02 -1.29235581e-01 7.67970204e-01 7.45826215e-02 -8.12932327e-02 1.69541150e-01 -2.36947909e-01 3.58414114e-01 2.11020801e-02 -1.93215802e-01 1.33112085e+00 5.85248411e-01 -2.86813170e-01 2.67391682e-01 5.00546932e-01 -2.23935515e-01 -8.43423545e-01 4.30743992e-02 2.31741946e-02 -1.99161500e-01 -2.39973947e-01 -8.98295403e-01 -7.81992555e-01 1.16717327e+00 -1.34774715e-01 3.51261824e-01 8.61812174e-01 -2.56130368e-01 2.31864914e-01 7.77518928e-01 4.66534525e-01 -4.96404946e-01 3.37702125e-01 9.41742957e-01 1.08739221e+00 -6.92246199e-01 -8.65637138e-02 -4.40874040e-01 -5.64739525e-01 1.31752956e+00 9.77599442e-01 -3.08286458e-01 -5.01726009e-02 5.41562736e-01 -2.12636590e-01 -2.62123764e-01 -8.51022422e-01 1.27717897e-01 2.77410120e-01 7.70381391e-01 4.90983725e-01 -7.52155110e-02 -4.43783067e-02 6.07774854e-01 -6.71927154e-01 2.14294374e-01 7.74326861e-01 1.27048421e+00 -3.49157155e-01 -9.35011804e-01 -4.43264604e-01 8.90123844e-02 2.29877681e-02 -4.24649298e-01 -1.56449482e-01 7.54061699e-01 -1.03071146e-01 3.24911922e-01 -9.04173478e-02 -2.44251475e-01 2.62260139e-01 1.68978527e-01 8.78893077e-01 -8.39241385e-01 -1.10911906e+00 -7.04657793e-01 6.78357482e-02 -5.14473140e-01 1.36352673e-01 -1.92759603e-01 -1.57475030e+00 -7.08904743e-01 3.90205115e-01 2.65024006e-01 5.86909711e-01 8.54734659e-01 3.81885394e-02 8.21855187e-01 -1.28197044e-01 -1.10108674e+00 -4.91686076e-01 -2.64580995e-01 -6.95140585e-02 1.91072449e-01 1.79038495e-01 -5.57784796e-01 7.04216659e-02 2.49936298e-01]
[9.087738990783691, 7.211365699768066]