[
  {
    "id": 0,
    "page": 4,
    "bounding_box": [
      139.5268300374349,
      72.198974609375,
      472.4731699625651,
      255.51202392578125
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | c | p{3cm} | p{3cm} | p{3cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2.5cm} | >{\\centering\\arraybackslash}m{2.5cm} | >{\\centering\\arraybackslash}m{2.5cm} | >{\\centering\\arraybackslash}m{2.5cm} |}\n\\hline\nName & Number of classes & Total number of sample & \\#Input Features\\\\\n\\hline\nExtended Yale Face Dataset (Frontal Pose) & 38 & 2432 & 900 \\\\ \\hline\nExtended Yale Face Dataset (All Poses) & 28 & 11482 & 900 \\\\ \\hline\nNCKU Taiwan Face Dataset & 90 & 3330 & 768 \\\\ \\hline\nMNIST Dataset & 10 & 70000 & 784 \\\\ \\hline\n\\end{tabular}\n \\\\\n\\end{center}\n\\caption{List of datasets used for benchmarking}\n\\label{table:dataset-summary}\n\\end{table}",
    "extracted_content": [
      [
        "Name",
        "Number of\nclasses",
        "Total number\nof sample",
        "#Input\nFeatures"
      ],
      [
        "Extended Yale\nFace Dataset\n(Frontal Pose)",
        "38",
        "2432",
        "900"
      ],
      [
        "Extended Yale\nFace Dataset\n(All Poses)",
        "28",
        "11482",
        "900"
      ],
      [
        "NCKU\nTaiwan Face\nDataset",
        "90",
        "3330",
        "768"
      ],
      [
        "MNIST\nDataset",
        "10",
        "70000",
        "784"
      ]
    ],
    "similarity_score": 0.8833652007648184,
    "table_image": "images/1607.01354v1/table_0.png",
    "page_image": "pages/1607.01354v1/page_4.png"
  },
  {
    "id": 1,
    "page": 5,
    "bounding_box": [
      167.8785683768136,
      334.4599914550781,
      444.1215558733259,
      504.2229919433594
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | c | c | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} |}\n\\hline\nClassifier & Setting & Accuracy in $R^{900}$ & Accuracy in $R^{64}$\\\\\n\\hline\nNeural Network & 75-50-38 & & 98.3\\% \\\\ \\hline\nk-Nearest Neighbor & k=3 & 60.6\\% & 97.3\\% \\\\ \\hline\nk-Nearest Neighbor & k=5 & 60.3\\% & 97.5\\% \\\\ \\hline\nk-Nearest Neighbor & k=7 & 58.5\\% & 97.5\\% \\\\ \\hline\nk-Nearest Neighbor & k=9 & 56.7\\% & 97.5\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Results on Extended Yale B (Frontal) Face data set}\n\\label{table:cropped-yale-res}\n\\end{table}",
    "extracted_content": [
      [
        "Classifier",
        "Setting",
        "Accuracy\nin R900",
        "Accuracy\nin R64"
      ],
      [
        "Neural\nNetwork",
        "75-50-38",
        "",
        "98.3%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=3",
        "60.6%",
        "97.3%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=5",
        "60.3%",
        "97.5%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=7",
        "58.5%",
        "97.5%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=9",
        "56.7%",
        "97.5%"
      ]
    ],
    "similarity_score": 0.4260355029585799,
    "table_image": "images/1607.01354v1/table_1.png",
    "page_image": "pages/1607.01354v1/page_5.png"
  },
  {
    "id": 2,
    "page": 6,
    "bounding_box": [
      133.33400344848633,
      72.198974609375,
      478.6657485961914,
      171.02899169921875
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | p{1cm} | p{1cm} | p{1cm} | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | }\n\\hline\nStudy & \\#Subjects & \\#Train Images per Subject & \\#Model Params (million) & Accuracy\\\\\n\\hline\nCurrent Paper & 38 & 48  & 0.5 & 98.3\\% \\\\ \\hline\nHinton et. al. \\cite{conf/icml/TangSH12a} & 10 & 7 & 1.3 & 97\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Comparison of results on Extended Yale B (Frontal) data set}\n\\label{table:yale-frontal-compare}\n\\end{table}",
    "extracted_content": [
      [
        "Study",
        "#Subjects",
        "#Train\nImages per\nSubject",
        "#Model\nParams\n(million)",
        "Accuracy"
      ],
      [
        "Current\nPaper",
        "38",
        "48",
        "0.5",
        "98.3%"
      ],
      [
        "Hinton et.\nal. [22]",
        "10",
        "7",
        "1.3",
        "97%"
      ]
    ],
    "similarity_score": 0.7058823529411765,
    "table_image": "images/1607.01354v1/table_2.png",
    "page_image": "pages/1607.01354v1/page_6.png"
  },
  {
    "id": 3,
    "page": 7,
    "bounding_box": [
      167.8785683768136,
      72.198974609375,
      444.1215558733259,
      241.9630126953125
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | c | c | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} |}\n\\hline\nClassifier & Setting & Accuracy in $R^{900}$ & Accuracy in $R^{64}$\\\\\n\\hline\nNeural Network & 75-50-38 & & 95.7\\% \\\\ \\hline\nk-Nearest Neighbor & k=3 & 81.6\\% & 95.4\\% \\\\ \\hline\nk-Nearest Neighbor & k=5 & 81.3\\% & 95.4\\% \\\\ \\hline\nk-Nearest Neighbor & k=7 & 81.0\\% & 95.4\\% \\\\ \\hline\nk-Nearest Neighbor & k=9 & 80.5\\% & 95.3\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Results on Extended Yale B (All Pose) face data set}\n\\label{table:yale-all-pose-res}\n\\end{table}",
    "extracted_content": [
      [
        "Classifier",
        "Setting",
        "Accuracy\nin R900",
        "Accuracy\nin R64"
      ],
      [
        "Neural\nNetwork",
        "75-50-38",
        "",
        "95.7%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=3",
        "81.6%",
        "95.4%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=5",
        "81.3%",
        "95.4%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=7",
        "81.0%",
        "95.4%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=9",
        "80.5%",
        "95.3%"
      ]
    ],
    "similarity_score": 0.6444007858546169,
    "table_image": "images/1607.01354v1/table_3.png",
    "page_image": "pages/1607.01354v1/page_7.png"
  },
  {
    "id": 4,
    "page": 7,
    "bounding_box": [
      167.8785683768136,
      277.57000732421875,
      444.1215558733259,
      447.3340148925781
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | c | c | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} |}\n\\hline\nClassifier & Setting & Accuracy in $R^{768}$ & Accuracy in $R^{25}$\\\\\n\\hline\nNeural Network & 25-50-90 & & 99.5\\% \\\\ \\hline\nk-Nearest Neighbor & k=3 & 97.171\\% & 99.6\\% \\\\ \\hline\nk-Nearest Neighbor & k=5 & 94.44\\% & 99.6\\% \\\\ \\hline\nk-Nearest Neighbor & k=7 & 91.81\\% & 99.6\\% \\\\ \\hline\nk-Nearest Neighbor & k=9 & 89.09\\% & 99.6\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Results on Taiwan Face data set}\n\\label{table:taiwan-res}\n\\end{table}",
    "extracted_content": [
      [
        "Classifier",
        "Setting",
        "Accuracy\nin R768",
        "Accuracy\nin R25"
      ],
      [
        "Neural\nNetwork",
        "25-50-90",
        "",
        "99.5%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=3",
        "97.171%",
        "99.6%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=5",
        "94.44%",
        "99.6%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=7",
        "91.81%",
        "99.6%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=9",
        "89.09%",
        "99.6%"
      ]
    ],
    "similarity_score": 0.8008130081300813,
    "table_image": "images/1607.01354v1/table_4.png",
    "page_image": "pages/1607.01354v1/page_7.png"
  },
  {
    "id": 5,
    "page": 8,
    "bounding_box": [
      167.8785683768136,
      96.70501708984375,
      444.1215558733259,
      266.468994140625
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | c | c | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} |}\n\\hline\nClassifier & Setting & Accuracy in $R^{784}$ & Accuracy in $R^{36}$\\\\\n\\hline\nNeural Network & 36-5-10 & & 98.08\\% \\\\ \\hline\nk-Nearest Neighbor & k=3 & 97.05\\% & 97.5\\% \\\\ \\hline\nk-Nearest Neighbor & k=5 & 96.88\\% & 97.5\\% \\\\ \\hline\nk-Nearest Neighbor & k=7 & 96.94\\% & 97.6\\% \\\\ \\hline\nk-Nearest Neighbor & k=9 & 96.59\\% & 97.6\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Results on MNIST data set}\n\\label{table:mnist-res}\n\\end{table}",
    "extracted_content": [
      [
        "Classifier",
        "Setting",
        "Accuracy\nin R784",
        "Accuracy\nin R36"
      ],
      [
        "Neural\nNetwork",
        "36-5-10",
        "",
        "98.08%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=3",
        "97.05%",
        "97.5%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=5",
        "96.88%",
        "97.5%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=7",
        "96.94%",
        "97.6%"
      ],
      [
        "k-Nearest\nNeighbor",
        "k=9",
        "96.59%",
        "97.6%"
      ]
    ],
    "similarity_score": 0.5714285714285714,
    "table_image": "images/1607.01354v1/table_5.png",
    "page_image": "pages/1607.01354v1/page_8.png"
  },
  {
    "id": 6,
    "page": 8,
    "bounding_box": [
      160.78020629882812,
      368.5159912109375,
      451.21979370117185,
      523.5339965820312
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | p{1.2cm} | p{1.8cm} | p{1cm} | p{1cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2.5cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} |}\n\\hline\nStudy & Method & \\#Model Params (million) & Accuracy\\\\\n\\hline\nThis Paper & Discriminative Encoder  & 0.23 & 98.08\\% \\\\ \\hline\nHinton et. al. \\cite{journals/jmlr/SalakhutdinovH07} & Autoencoder & 1.7 & 99\\% \\\\ \\hline\nSchmidhuber et. al. \\cite{journals/corr/abs-1003-0358} & Simple Deep Neural Nets + Elastic Distortions & 11.9 mil & 99.65\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Comparison of results on MNIST data set}\n\\label{table:mnist-compare}\n\\end{table}",
    "extracted_content": [
      [
        "Study",
        "Method",
        "#Model\nParams\n(million)",
        "Accuracy"
      ],
      [
        "This Paper",
        "Discriminative\nEncoder",
        "0.23",
        "98.08%"
      ],
      [
        "Hinton et.\nal. [16]",
        "Autoencoder",
        "1.7",
        "99%"
      ],
      [
        "Schmidhuber\net. al. [2]",
        "Simple Deep\nNeural Nets +\nElastic\nDistortions",
        "11.9 mil",
        "99.65%"
      ]
    ],
    "similarity_score": 0.6913123844731978,
    "table_image": "images/1607.01354v1/table_6.png",
    "page_image": "pages/1607.01354v1/page_8.png"
  },
  {
    "id": 7,
    "page": 9,
    "bounding_box": [
      72.1658312479655,
      97.7020263671875,
      551.9191487630209,
      294.9620056152344
    ],
    "latex_content": "\\begin{table}[h]\n\\begin{center}\n%\\begin{tabular}{ | p{1cm} | p{0.5cm}| p{0.65cm}| p{1.05cm}| p{1.05cm} | p{0.45cm} | p{0.45cm} |p{0.45cm}| p{0.45cm} | }\n\\begin{tabular}{| >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{1.5cm} | >{\\centering\\arraybackslash}m{1.5cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{2cm} | >{\\centering\\arraybackslash}m{1cm} | >{\\centering\\arraybackslash}m{1cm} | >{\\centering\\arraybackslash}m{1cm} | >{\\centering\\arraybackslash}m{1cm} |}\n\\hline\nDataset & Input Space Size & Reduced Space Size & Network (AE) & Network (DE) &  IS & PCA & AE & DE\\\\ \\hline\nYale (Frontal Pose) & 900 & 64 & 400-200-64-200-400-900 & 400-200-64-900 & 60.6\\% & 51.4\\% & 82.4\\% & 97.3\\% \\\\ \\hline\nYale (All Poses) & 900 & 64 & 400-200-64-200-400-900 & 400-200-64-900 & 81.6\\% & 74.6\\% & 89.1\\% & 95.4\\% \\\\ \\hline\nTaiwan Face Db & 768 & 25 & 196-64-25-64-196-768 & 196-64-25-768 & 97.1\\% & 96.9\\% & 96.8\\% & 99.6\\% \\\\ \\hline\nMNIST & 784 & 36 & 225-100-36-100-225-784 & 225-100-36-784 & 97.0\\% & 97.3\\% & 97.0\\% & 97.5\\% \\\\ \\hline\n\\end{tabular}\n\\\\\n\\end{center}\n\\caption{Results of 3-NN classifier on all datasets using various dimensionality reduction approaches: IS (original input space), PCA (principal component analysis), AE (autoencoder), DE (discriminative encoder)}\n\\label{table:summary-res}\n\\end{table}",
    "extracted_content": [
      [
        "Dataset",
        "Input\nSpace\nSize",
        "Reduced\nSpace\nSize",
        "Network\n(AE)",
        "Network\n(DE)",
        "IS",
        "PCA",
        "AE",
        "DE"
      ],
      [
        "Yale\n(Frontal\nPose)",
        "900",
        "64",
        "400-200-\n64-200-\n400-900",
        "400-200-\n64-900",
        "60.6%",
        "51.4%",
        "82.4%",
        "97.3%"
      ],
      [
        "Yale (All\nPoses)",
        "900",
        "64",
        "400-200-\n64-200-\n400-900",
        "400-200-\n64-900",
        "81.6%",
        "74.6%",
        "89.1%",
        "95.4%"
      ],
      [
        "Taiwan\nFace Db",
        "768",
        "25",
        "196-64-25-\n64-196-768",
        "196-64-25-\n768",
        "97.1%",
        "96.9%",
        "96.8%",
        "99.6%"
      ],
      [
        "MNIST",
        "784",
        "36",
        "225-100-\n36-100-\n225-784",
        "225-100-\n36-784",
        "97.0%",
        "97.3%",
        "97.0%",
        "97.5%"
      ]
    ],
    "similarity_score": 0.4491362763915547,
    "table_image": "images/1607.01354v1/table_7.png",
    "page_image": "pages/1607.01354v1/page_9.png"
  }
]