Datasets:

CShorten

/

Last-Week-on-ML-ArXiv

like 0

Online vs. Offline Adaptive Domain Randomization Benchmark

Physics simulators have shown great promise for conveniently learning reinforcement learning policies in safe, unconstrained environments. However, transferring the acquired knowledge to the real world can be challenging due to the reality gap. To this end, several methods have been recently proposed to automatically tune simulator parameters with posterior distributions given real data, for use with domain randomization at training time. These approaches have been shown to work for various robotic tasks under different settings and assumptions. Nevertheless, existing literature lacks a thorough comparison of existing adaptive domain randomization methods with respect to transfer performance and real-data efficiency. In this work, we present an open benchmark for both offline and online methods (SimOpt, BayRn, DROID, DROPO), to shed light on which are most suitable for each setting and task at hand. We found that online methods are limited by the quality of the currently learned policy for the next iteration, while offline methods may sometimes fail when replaying trajectories in simulation with open-loop commands. The code used will be released at https://github.com/gabrieletiboni/adr-benchmark.

Cut Inner Layers: A Structured Pruning Strategy for Efficient U-Net GANs

Pruning effectively compresses overparameterized models. Despite the success of pruning methods for discriminative models, applying them for generative models has been relatively rarely approached. This study conducts structured pruning on U-Net generators of conditional GANs. A per-layer sensitivity analysis confirms that many unnecessary filters exist in the innermost layers near the bottleneck and can be substantially pruned. Based on this observation, we prune these filters from multiple inner layers or suggest alternative architectures by completely eliminating the layers. We evaluate our approach with Pix2Pix for image-to-image translation and Wav2Lip for speech-driven talking face generation. Our method outperforms global pruning baselines, demonstrating the importance of properly considering where to prune for U-Net generators.

DDKtor: Automatic Diadochokinetic Speech Analysis

Diadochokinetic speech tasks (DDK), in which participants repeatedly produce syllables, are commonly used as part of the assessment of speech motor impairments. These studies rely on manual analyses that are time-intensive, subjective, and provide only a coarse-grained picture of speech. This paper presents two deep neural network models that automatically segment consonants and vowels from unannotated, untranscribed speech. Both models work on the raw waveform and use convolutional layers for feature extraction. The first model is based on an LSTM classifier followed by fully connected layers, while the second model adds more convolutional layers followed by fully connected layers. These segmentations predicted by the models are used to obtain measures of speech rate and sound duration. Results on a young healthy individuals dataset show that our LSTM model outperforms the current state-of-the-art systems and performs comparably to trained human annotators. Moreover, the LSTM model also presents comparable results to trained human annotators when evaluated on unseen older individuals with Parkinson's Disease dataset.

From Kernel Methods to Neural Networks: A Unifying Variational Formulation

The minimization of a data-fidelity term and an additive regularization functional gives rise to a powerful framework for supervised learning. In this paper, we present a unifying regularization functional that depends on an operator and on a generic Radon-domain norm. We establish the existence of a minimizer and give the parametric form of the solution(s) under very mild assumptions. When the norm is Hilbertian, the proposed formulation yields a solution that involves radial-basis functions and is compatible with the classical methods of machine learning. By contrast, for the total-variation norm, the solution takes the form of a two-layer neural network with an activation function that is determined by the regularization operator. In particular, we retrieve the popular ReLU networks by letting the operator be the Laplacian. We also characterize the solution for the intermediate regularization norms $\|\cdot\|=\|\cdot\|_{L_p}$ with $p\in(1,2]$. Our framework offers guarantees of universal approximation for a broad family of regularization operators or, equivalently, for a wide variety of shallow neural networks, including the cases (such as ReLU) where the activation function is increasing polynomially. It also explains the favorable role of bias and skip connections in neural architectures.

SALO: An Efficient Spatial Accelerator Enabling Hybrid Sparse Attention Mechanisms for Long Sequences

The attention mechanisms of transformers effectively extract pertinent information from the input sequence. However, the quadratic complexity of self-attention w.r.t the sequence length incurs heavy computational and memory burdens, especially for tasks with long sequences. Existing accelerators face performance degradation in these tasks. To this end, we propose SALO to enable hybrid sparse attention mechanisms for long sequences. SALO contains a data scheduler to map hybrid sparse attention patterns onto hardware and a spatial accelerator to perform the efficient attention computation. We show that SALO achieves 17.66x and 89.33x speedup on average compared to GPU and CPU implementations, respectively, on typical workloads, i.e., Longformer and ViL.

Comparative Study of Inference Methods for Interpolative Decomposition

In this paper, we propose a probabilistic model with automatic relevance determination (ARD) for learning interpolative decomposition (ID), which is commonly used for low-rank approximation, feature selection, and identifying hidden patterns in data, where the matrix factors are latent variables associated with each data dimension. Prior densities with support on the specified subspace are used to address the constraint for the magnitude of the factored component of the observed matrix. Bayesian inference procedure based on Gibbs sampling is employed. We evaluate the model on a variety of real-world datasets including CCLE $EC50$, CCLE $IC50$, Gene Body Methylation, and Promoter Methylation datasets with different sizes, and dimensions, and show that the proposed Bayesian ID algorithms with automatic relevance determination lead to smaller reconstructive errors even compared to vanilla Bayesian ID algorithms with fixed latent dimension set to matrix rank.

Why patient data cannot be easily forgotten?

Rights provisioned within data protection regulations, permit patients to request that knowledge about their information be eliminated by data holders. With the advent of AI learned on data, one can imagine that such rights can extent to requests for forgetting knowledge of patient's data within AI models. However, forgetting patients' imaging data from AI models, is still an under-explored problem. In this paper, we study the influence of patient data on model performance and formulate two hypotheses for a patient's data: either they are common and similar to other patients or form edge cases, i.e. unique and rare cases. We show that it is not possible to easily forget patient data. We propose a targeted forgetting approach to perform patient-wise forgetting. Extensive experiments on the benchmark Automated Cardiac Diagnosis Challenge dataset showcase the improved performance of the proposed targeted forgetting approach as opposed to a state-of-the-art method.

When Does Group Invariant Learning Survive Spurious Correlations?

By inferring latent groups in the training data, recent works introduce invariant learning to the case where environment annotations are unavailable. Typically, learning group invariance under a majority/minority split is empirically shown to be effective in improving out-of-distribution generalization on many datasets. However, theoretical guarantee for these methods on learning invariant mechanisms is lacking. In this paper, we reveal the insufficiency of existing group invariant learning methods in preventing classifiers from depending on spurious correlations in the training set. Specifically, we propose two criteria on judging such sufficiency. Theoretically and empirically, we show that existing methods can violate both criteria and thus fail in generalizing to spurious correlation shifts. Motivated by this, we design a new group invariant learning method, which constructs groups with statistical independence tests, and reweights samples by group label proportion to meet the criteria. Experiments on both synthetic and real data demonstrate that the new method significantly outperforms existing group invariant learning methods in generalizing to spurious correlation shifts.

Revisiting Label Smoothing and Knowledge Distillation Compatibility: What was Missing?

This work investigates the compatibility between label smoothing (LS) and knowledge distillation (KD). Contemporary findings addressing this thesis statement take dichotomous standpoints: Muller et al. (2019) and Shen et al. (2021b). Critically, there is no effort to understand and resolve these contradictory findings, leaving the primal question -- to smooth or not to smooth a teacher network? -- unanswered. The main contributions of our work are the discovery, analysis and validation of systematic diffusion as the missing concept which is instrumental in understanding and resolving these contradictory findings. This systematic diffusion essentially curtails the benefits of distilling from an LS-trained teacher, thereby rendering KD at increased temperatures ineffective. Our discovery is comprehensively supported by large-scale experiments, analyses and case studies including image classification, neural machine translation and compact student distillation tasks spanning across multiple datasets and teacher-student architectures. Based on our analysis, we suggest practitioners to use an LS-trained teacher with a low-temperature transfer to achieve high performance students. Code and models are available at https://keshik6.github.io/revisiting-ls-kd-compatibility/

Imaging the time series of one single referenced EEG electrode for Epileptic Seizures Risk Analysis

The time series captured by a single scalp electrode (plus the reference electrode) of refractory epileptic patients is used to forecast seizures susceptibility. The time series is preprocessed, segmented, and each segment transformed into an image, using three different known methods: Recurrence Plot, Gramian Angular Field, Markov Transition Field. The likelihood of the occurrence of a seizure in a future predefined time window is computed by averaging the output of the softmax layer of a CNN, differently from the usual consideration of the output of the classification layer. By thresholding this likelihood, seizure forecasting has better performance. Interestingly, for almost every patient, the best threshold was different from 50%. The results show that this technique can predict with good results for some seizures and patients. However, more tests, namely more patients and more seizures, are needed to better understand the real potential of this technique.

Variational Quantum Approximate Support Vector Machine With Inference Transfer

A kernel-based quantum classifier is the most interesting and powerful quantum machine learning technique for hyperlinear classification of complex data, which can be easily realized in shallow-depth quantum circuits such as a SWAP test classifier. Surprisingly, a support vector machine can be realized inherently and explicitly on these circuits by introduction of a variational scheme to map the quadratic optimization problem of the SVM theory to a quantum-classical variational optimization problem. This scheme is realized with parameterized quantum circuits (PQC) to create a nonuniform weight vector to index qubits that can evaluate training loss and classification score in a linear time. We train the classical parameters of this Variational Quantum Approximate Support Vector Machine (VQASVM), which can be transferred to many copies of other VQASVM decision inference circuits for classification of new query data. Our VQASVM algorithm is experimented with toy example data sets on cloud-based quantum machines for feasibility evaluation, and numerically investigated to evaluate its performance on a standard iris flower data set. The accuracy of iris data classification reached 98.8%.

GERNERMED++: Transfer Learning in German Medical NLP

We present a statistical model for German medical natural language processing trained for named entity recognition (NER) as an open, publicly available model. The work serves as a refined successor to our first GERNERMED model which is substantially outperformed by our work. We demonstrate the effectiveness of combining multiple techniques in order to achieve strong results in entity recognition performance by the means of transfer-learning on pretrained deep language models (LM), word-alignment and neural machine translation. Due to the sparse situation on open, public medical entity recognition models for German texts, this work offers benefits to the German research community on medical NLP as a baseline model. Since our model is based on public English data, its weights are provided without legal restrictions on usage and distribution. The sample code and the statistical model is available at: https://github.com/frankkramer-lab/GERNERMED-pp

RegMixup: Mixup as a Regularizer Can Surprisingly Improve Accuracy and Out Distribution Robustness

We show that the effectiveness of the well celebrated Mixup [Zhang et al., 2018] can be further improved if instead of using it as the sole learning objective, it is utilized as an additional regularizer to the standard cross-entropy loss. This simple change not only provides much improved accuracy but also significantly improves the quality of the predictive uncertainty estimation of Mixup in most cases under various forms of covariate shifts and out-of-distribution detection experiments. In fact, we observe that Mixup yields much degraded performance on detecting out-of-distribution samples possibly, as we show empirically, because of its tendency to learn models that exhibit high-entropy throughout; making it difficult to differentiate in-distribution samples from out-distribution ones. To show the efficacy of our approach (RegMixup), we provide thorough analyses and experiments on vision datasets (ImageNet & CIFAR-10/100) and compare it with a suite of recent approaches for reliable uncertainty estimation.

Auto-Encoder-Extreme Learning Machine Model for Boiler NOx Emission Concentration Prediction

An automatic encoder (AE) extreme learning machine (ELM)-AE-ELM model is proposed to predict the NOx emission concentration based on the combination of mutual information algorithm (MI), AE, and ELM. First, the importance of practical variables is computed by the MI algorithm, and the mechanism is analyzed to determine the variables related to the NOx emission concentration. Then, the time delay correlations between the selected variables and NOx emission concentration are further analyzed to reconstruct the modeling data. Subsequently, the AE is applied to extract hidden features within the input variables. Finally, an ELM algorithm establishes the relationship between the NOx emission concentration and deep features. The experimental results on practical data indicate that the proposed model shows promising performance compared to state-of-art models.

Beyond neural scaling laws: beating power law scaling via data pruning

Widely observed neural scaling laws, in which error falls off as a power of the training set size, model size, or both, have driven substantial performance improvements in deep learning. However, these improvements through scaling alone require considerable costs in compute and energy. Here we focus on the scaling of error with dataset size and show how both in theory and practice we can break beyond power law scaling and reduce it to exponential scaling instead if we have access to a high-quality data pruning metric that ranks the order in which training examples should be discarded to achieve any pruned dataset size. We then test this new exponential scaling prediction with pruned dataset size empirically, and indeed observe better than power law scaling performance on ResNets trained on CIFAR-10, SVHN, and ImageNet. Given the importance of finding high-quality pruning metrics, we perform the first large-scale benchmarking study of ten different data pruning metrics on ImageNet. We find most existing high performing metrics scale poorly to ImageNet, while the best are computationally intensive and require labels for every image. We therefore developed a new simple, cheap and scalable self-supervised pruning metric that demonstrates comparable performance to the best supervised metrics. Overall, our work suggests that the discovery of good data-pruning metrics may provide a viable path forward to substantially improved neural scaling laws, thereby reducing the resource costs of modern deep learning.

Data augmentation for learning predictive models on EEG: a systematic comparison

The use of deep learning for electroencephalography (EEG) classification tasks has been rapidly growing in the last years, yet its application has been limited by the relatively small size of EEG datasets. Data augmentation, which consists in artificially increasing the size of the dataset during training, has been a key ingredient to obtain state-of-the-art performances across applications such as computer vision or speech. While a few augmentation transformations for EEG data have been proposed in the literature, their positive impact on performance across tasks remains elusive. In this work, we propose a unified and exhaustive analysis of the main existing EEG augmentations, which are compared in a common experimental setting. Our results highlight the best data augmentations to consider for sleep stage classification and motor imagery brain computer interfaces, showing predictive power improvements greater than 10% in some cases.

Adversarial Ensemble Training by Jointly Learning Label Dependencies and Member Models

Training an ensemble of different sub-models has empirically proven to be an effective strategy to improve deep neural networks' adversarial robustness. Current ensemble training methods for image recognition usually encode the image labels by one-hot vectors, which neglect dependency relationships between the labels. Here we propose a novel adversarial ensemble training approach to jointly learn the label dependencies and the member models. Our approach adaptively exploits the learned label dependencies to promote the diversity of the member models. We test our approach on widely used datasets MNIST, FasionMNIST, and CIFAR-10. Results show that our approach is more robust against black-box attacks compared with the state-of-the-art methods. Our code is available at https://github.com/ZJLAB-AMMI/LSD.

Can Push-forward Generative Models Fit Multimodal Distributions?

Many generative models synthesize data by transforming a standard Gaussian random variable using a deterministic neural network. Among these models are the Variational Autoencoders and the Generative Adversarial Networks. In this work, we call them "push-forward" models and study their expressivity. We show that the Lipschitz constant of these generative networks has to be large in order to fit multimodal distributions. More precisely, we show that the total variation distance and the Kullback-Leibler divergence between the generated and the data distribution are bounded from below by a constant depending on the mode separation and the Lipschitz constant. Since constraining the Lipschitz constants of neural networks is a common way to stabilize generative models, there is a provable trade-off between the ability of push-forward models to approximate multimodal distributions and the stability of their training. We validate our findings on one-dimensional and image datasets and empirically show that generative models consisting of stacked networks with stochastic input at each step, such as diffusion models do not suffer of such limitations.

SPI-GAN: Distilling Score-based Generative Models with Straight-Path Interpolations

Score-based generative models (SGMs) are a recently proposed paradigm for deep generative tasks and now show the state-of-the-art sampling performance. It is known that the original SGM design solves the two problems of the generative trilemma: i) sampling quality, and ii) sampling diversity. However, the last problem of the trilemma was not solved, i.e., their training/sampling complexity is notoriously high. To this end, distilling SGMs into simpler models, e.g., generative adversarial networks (GANs), is gathering much attention currently. We present an enhanced distillation method, called straight-path interpolation GAN (SPI-GAN), which can be compared to the state-of-the-art shortcut-based distillation method, called denoising diffusion GAN (DD-GAN). However, our method corresponds to an extreme method that does not use any intermediate shortcut information of the reverse SDE path, in which case DD-GAN fails to obtain good results. Nevertheless, our straight-path interpolation method greatly stabilizes the overall training process. As a result, SPI-GAN is one of the best models in terms of the sampling quality/diversity/time for CIFAR-10, CelebA-HQ-256, and LSUN-Church-256.

Deep Multiple Instance Learning For Forecasting Stock Trends Using Financial News

A major source of information can be taken from financial news articles, which have some correlations about the fluctuation of stock trends. In this paper, we investigate the influences of financial news on the stock trends, from a multi-instance view. The intuition behind this is based on the news uncertainty of varying intervals of news occurrences and the lack of annotation in every single financial news. Under the scenario of Multiple Instance Learning (MIL) where training instances are arranged in bags, and a label is assigned for the entire bag instead of instances, we develop a flexible and adaptive multi-instance learning model and evaluate its ability in directional movement forecast of Standard & Poors 500 index on financial news dataset. Specifically, we treat each trading day as one bag, with certain amounts of news happening on each trading day as instances in each bag. Experiment results demonstrate that our proposed multi-instance-based framework gains outstanding results in terms of the accuracy of trend prediction, compared with other state-of-art approaches and baselines.

Off-the-grid learning of sparse mixtures from a continuous dictionary

We consider a general non-linear model where the signal is a finite mixture of an unknown, possibly increasing, number of features issued from a continuous dictionary parameterized by a real nonlinear parameter. The signal is observed with Gaussian (possibly correlated) noise in either a continuous or a discrete setup. We propose an off-the-grid optimization method, that is, a method which does not use any discretization scheme on the parameter space, to estimate both the non-linear parameters of the features and the linear parameters of the mixture. We use recent results on the geometry of off-the-grid methods to give minimal separation on the true underlying non-linear parameters such that interpolating certificate functions can be constructed. Using also tail bounds for suprema of Gaussian processes we bound the prediction error with high probability. Assuming that the certificate functions can be constructed, our prediction error bound is up to log --factors similar to the rates attained by the Lasso predictor in the linear regression model. We also establish convergence rates that quantify with high probability the quality of estimation for both the linear and the non-linear parameters.

Approximate Data Deletion in Generative Models

Users have the right to have their data deleted by third-party learned systems, as codified by recent legislation such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA). Such data deletion can be accomplished by full re-training, but this incurs a high computational cost for modern machine learning models. To avoid this cost, many approximate data deletion methods have been developed for supervised learning. Unsupervised learning, in contrast, remains largely an open problem when it comes to (approximate or exact) efficient data deletion. In this paper, we propose a density-ratio-based framework for generative models. Using this framework, we introduce a fast method for approximate data deletion and a statistical test for estimating whether or not training points have been deleted. We provide theoretical guarantees under various learner assumptions and empirically demonstrate our methods across a variety of generative methods.

Open Problem: Properly learning decision trees in polynomial time?

The authors recently gave an $n^{O(\log\log n)}$ time membership query algorithm for properly learning decision trees under the uniform distribution (Blanc et al., 2021). The previous fastest algorithm for this problem ran in $n^{O(\log n)}$ time, a consequence of Ehrenfeucht and Haussler (1989)'s classic algorithm for the distribution-free setting. In this article we highlight the natural open problem of obtaining a polynomial-time algorithm, discuss possible avenues towards obtaining it, and state intermediate milestones that we believe are of independent interest.

Cyclical Kernel Adaptive Metropolis

We propose cKAM, cyclical Kernel Adaptive Metropolis, which incorporates a cyclical stepsize scheme to allow control for exploration and sampling. We show that on a crafted bimodal distribution, existing Adaptive Metropolis type algorithms would fail to converge to the true posterior distribution. We point out that this is because adaptive samplers estimates the local/global covariance structure using past history of the chain, which will lead to adaptive algorithms be trapped in a local mode. We demonstrate that cKAM encourages exploration of the posterior distribution and allows the sampler to escape from a local mode, while maintaining the high performance of adaptive methods.

Optimization-Induced Graph Implicit Nonlinear Diffusion

Due to the over-smoothing issue, most existing graph neural networks can only capture limited dependencies with their inherently finite aggregation layers. To overcome this limitation, we propose a new kind of graph convolution, called Graph Implicit Nonlinear Diffusion (GIND), which implicitly has access to infinite hops of neighbors while adaptively aggregating features with nonlinear diffusion to prevent over-smoothing. Notably, we show that the learned representation can be formalized as the minimizer of an explicit convex optimization objective. With this property, we can theoretically characterize the equilibrium of our GIND from an optimization perspective. More interestingly, we can induce new structural variants by modifying the corresponding optimization objective. To be specific, we can embed prior properties to the equilibrium, as well as introducing skip connections to promote training stability. Extensive experiments show that GIND is good at capturing long-range dependencies, and performs well on both homophilic and heterophilic graphs with nonlinear diffusion. Moreover, we show that the optimization-induced variants of our models can boost the performance and improve training stability and efficiency as well. As a result, our GIND obtains significant improvements on both node-level and graph-level tasks.

Fair Machine Learning in Healthcare: A Review

Benefiting from the digitization of healthcare data and the development of computing power, machine learning methods are increasingly used in the healthcare domain. Fairness problems have been identified in machine learning for healthcare, resulting in an unfair allocation of limited healthcare resources or excessive health risks for certain groups. Therefore, addressing the fairness problems has recently attracted increasing attention from the healthcare community. However, the intersection of machine learning for healthcare and fairness in machine learning remains understudied. In this review, we build the bridge by exposing fairness problems, summarizing possible biases, sorting out mitigation methods and pointing out challenges along with opportunities for the future.

Forgetting Data from Pre-trained GANs

Large pre-trained generative models are known to occasionally provide samples that may be undesirable for various reasons. The standard way to mitigate this is to re-train the models differently. In this work, we take a different, more compute-friendly approach and investigate how to post-edit a model after training so that it forgets certain kinds of samples. We provide three different algorithms for GANs that differ on how the samples to be forgotten are described. Extensive evaluations on real-world image datasets show that our algorithms are capable of forgetting data while retaining high generation quality at a fraction of the cost of full re-training.

Framing Algorithmic Recourse for Anomaly Detection

The problem of algorithmic recourse has been explored for supervised machine learning models, to provide more interpretable, transparent and robust outcomes from decision support systems. An unexplored area is that of algorithmic recourse for anomaly detection, specifically for tabular data with only discrete feature values. Here the problem is to present a set of counterfactuals that are deemed normal by the underlying anomaly detection model so that applications can utilize this information for explanation purposes or to recommend countermeasures. We present an approach -- Context preserving Algorithmic Recourse for Anomalies in Tabular data (CARAT), that is effective, scalable, and agnostic to the underlying anomaly detection model. CARAT uses a transformer based encoder-decoder model to explain an anomaly by finding features with low likelihood. Subsequently semantically coherent counterfactuals are generated by modifying the highlighted features, using the overall context of features in the anomalous instance(s). Extensive experiments help demonstrate the efficacy of CARAT.

Overview of Deep Learning-based CSI Feedback in Massive MIMO Systems

Many performance gains achieved by massive multiple-input and multiple-output depend on the accuracy of the downlink channel state information (CSI) at the transmitter (base station), which is usually obtained by estimating at the receiver (user terminal) and feeding back to the transmitter. The overhead of CSI feedback occupies substantial uplink bandwidth resources, especially when the number of the transmit antennas is large. Deep learning (DL)-based CSI feedback refers to CSI compression and reconstruction by a DL-based autoencoder and can greatly reduce feedback overhead. In this paper, a comprehensive overview of state-of-the-art research on this topic is provided, beginning with basic DL concepts widely used in CSI feedback and then categorizing and describing some existing DL-based feedback works. The focus is on novel neural network architectures and utilization of communication expert knowledge to improve CSI feedback accuracy. Works on bit-level CSI feedback and joint design of CSI feedback with other communication modules are also introduced, and some practical issues, including training dataset collection, online training, complexity, generalization, and standardization effect, are discussed. At the end of the paper, some challenges and potential research directions associated with DL-based CSI feedback in future wireless communication systems are identified.

Exploiting Semantic Role Contextualized Video Features for Multi-Instance Text-Video Retrieval EPIC-KITCHENS-100 Multi-Instance Retrieval Challenge 2022

In this report, we present our approach for EPIC-KITCHENS-100 Multi-Instance Retrieval Challenge 2022. We first parse sentences into semantic roles corresponding to verbs and nouns; then utilize self-attentions to exploit semantic role contextualized video features along with textual features via triplet losses in multiple embedding spaces. Our method overpasses the strong baseline in normalized Discounted Cumulative Gain (nDCG), which is more valuable for semantic similarity. Our submission is ranked 3rd for nDCG and ranked 4th for mAP.

Theoretical Perspectives on Deep Learning Methods in Inverse Problems

In recent years, there have been significant advances in the use of deep learning methods in inverse problems such as denoising, compressive sensing, inpainting, and super-resolution. While this line of works has predominantly been driven by practical algorithms and experiments, it has also given rise to a variety of intriguing theoretical problems. In this paper, we survey some of the prominent theoretical developments in this line of works, focusing in particular on generative priors, untrained neural network priors, and unfolding algorithms. In addition to summarizing existing results in these topics, we highlight several ongoing challenges and open problems.

Matryoshka: Stealing Functionality of Private ML Data by Hiding Models in Model

In this paper, we present a novel insider attack called Matryoshka, which employs an irrelevant scheduled-to-publish DNN model as a carrier model for covert transmission of multiple secret models which memorize the functionality of private ML data stored in local data centers. Instead of treating the parameters of the carrier model as bit strings and applying conventional steganography, we devise a novel parameter sharing approach which exploits the learning capacity of the carrier model for information hiding. Matryoshka simultaneously achieves: (i) High Capacity -- With almost no utility loss of the carrier model, Matryoshka can hide a 26x larger secret model or 8 secret models of diverse architectures spanning different application domains in the carrier model, neither of which can be done with existing steganography techniques; (ii) Decoding Efficiency -- once downloading the published carrier model, an outside colluder can exclusively decode the hidden models from the carrier model with only several integer secrets and the knowledge of the hidden model architecture; (iii) Effectiveness -- Moreover, almost all the recovered models have similar performance as if it were trained independently on the private data; (iv) Robustness -- Information redundancy is naturally implemented to achieve resilience against common post-processing techniques on the carrier before its publishing; (v) Covertness -- A model inspector with different levels of prior knowledge could hardly differentiate a carrier model from a normal model.

TE2Rules: Extracting Rule Lists from Tree Ensembles

Tree Ensemble (TE) models (e.g. Gradient Boosted Trees and Random Forests) often provide higher prediction performance compared to single decision trees. However, TE models generally lack transparency and interpretability, as humans have difficulty understanding their decision logic. This paper presents a novel approach to convert a TE trained for a binary classification task, to a rule list (RL) that is a global equivalent to the TE and is comprehensible for a human. This RL captures all necessary and sufficient conditions for decision making by the TE. Experiments on benchmark datasets demonstrate that, compared to state-of-the-art methods, (i) predictions from the RL generated by TE2Rules have high fidelity with respect to the original TE, (ii) the RL from TE2Rules has high interpretability measured by the number and the length of the decision rules, (iii) the run-time of TE2Rules algorithm can be reduced significantly at the cost of a slightly lower fidelity, and (iv) the RL is a fast alternative to the state-of-the-art rule-based instance-level outcome explanation techniques.

Using Twitter Data to Understand Public Perceptions of Approved versus Off-label Use for COVID-19-related Medications

Understanding public discourse on emergency use of unproven therapeutics is essential to monitor safe use and combat misinformation. We developed a natural language processing (NLP)-based pipeline to understand public perceptions of and stances on COVID-19-related drugs on Twitter across time. This retrospective study included 609,189 US-based tweets between January 29th, 2020 and November 30th, 2021 on four drugs that gained wide public attention during the COVID-19 pandemic: 1) Hydroxychloroquine and Ivermectin, drug therapies with anecdotal evidence; and 2) Molnupiravir and Remdesivir, FDA-approved treatment options for eligible patients. Time-trend analysis was used to understand the popularity and related events. Content and demographic analyses were conducted to explore potential rationales of people's stances on each drug. Time-trend analysis revealed that Hydroxychloroquine and Ivermectin received much more discussion than Molnupiravir and Remdesivir, particularly during COVID-19 surges. Hydroxychloroquine and Ivermectin were highly politicized, related to conspiracy theories, hearsay, celebrity effects, etc. The distribution of stance between the two major US political parties was significantly different (p<0.001); Republicans were much more likely to support Hydroxychloroquine (+55%) and Ivermectin (+30%) than Democrats. People with healthcare backgrounds tended to oppose Hydroxychloroquine (+7%) more than the general population; in contrast, the general population was more likely to support Ivermectin (+14%). We make all the data, code, and models available at https://github.com/ningkko/COVID-drug.

EBMs vs. CL: Exploring Self-Supervised Visual Pretraining for Visual Question Answering

The availability of clean and diverse labeled data is a major roadblock for training models on complex tasks such as visual question answering (VQA). The extensive work on large vision-and-language models has shown that self-supervised learning is effective for pretraining multimodal interactions. In this technical report, we focus on visual representations. We review and evaluate self-supervised methods to leverage unlabeled images and pretrain a model, which we then fine-tune on a custom VQA task that allows controlled evaluation and diagnosis. We compare energy-based models (EBMs) with contrastive learning (CL). While EBMs are growing in popularity, they lack an evaluation on downstream tasks. We find that both EBMs and CL can learn representations from unlabeled images that enable training a VQA model on very little annotated data. In a simple setting similar to CLEVR, we find that CL representations also improve systematic generalization, and even match the performance of representations from a larger, supervised, ImageNet-pretrained model. However, we find EBMs to be difficult to train because of instabilities and high variability in their results. Although EBMs prove useful for OOD detection, other results on supervised energy-based training and uncertainty calibration are largely negative. Overall, CL currently seems a preferable option over EBMs.

Hardness and Algorithms for Robust and Sparse Optimization

We explore algorithms and limitations for sparse optimization problems such as sparse linear regression and robust linear regression. The goal of the sparse linear regression problem is to identify a small number of key features, while the goal of the robust linear regression problem is to identify a small number of erroneous measurements. Specifically, the sparse linear regression problem seeks a $k$-sparse vector $x\in\mathbb{R}^d$ to minimize $\|Ax-b\|_2$, given an input matrix $A\in\mathbb{R}^{n\times d}$ and a target vector $b\in\mathbb{R}^n$, while the robust linear regression problem seeks a set $S$ that ignores at most $k$ rows and a vector $x$ to minimize $\|(Ax-b)_S\|_2$. We first show bicriteria, NP-hardness of approximation for robust regression building on the work of [OWZ15] which implies a similar result for sparse regression. We further show fine-grained hardness of robust regression through a reduction from the minimum-weight $k$-clique conjecture. On the positive side, we give an algorithm for robust regression that achieves arbitrarily accurate additive error and uses runtime that closely matches the lower bound from the fine-grained hardness result, as well as an algorithm for sparse regression with similar runtime. Both our upper and lower bounds rely on a general reduction from robust linear regression to sparse regression that we introduce. Our algorithms, inspired by the 3SUM problem, use approximate nearest neighbor data structures and may be of independent interest for solving sparse optimization problems. For instance, we demonstrate that our techniques can also be used for the well-studied sparse PCA problem.

Convolutional Neural Network Based Partial Face Detection

Due to the massive explanation of artificial intelligence, machine learning technology is being used in various areas of our day-to-day life. In the world, there are a lot of scenarios where a simple crime can be prevented before it may even happen or find the person responsible for it. A face is one distinctive feature that we have and can differentiate easily among many other species. But not just different species, it also plays a significant role in determining someone from the same species as us, humans. Regarding this critical feature, a single problem occurs most often nowadays. When the camera is pointed, it cannot detect a person's face, and it becomes a poor image. On the other hand, where there was a robbery and a security camera installed, the robber's identity is almost indistinguishable due to the low-quality camera. But just making an excellent algorithm to work and detecting a face reduces the cost of hardware, and it doesn't cost that much to focus on that area. Facial recognition, widget control, and such can be done by detecting the face correctly. This study aims to create and enhance a machine learning model that correctly recognizes faces. Total 627 Data have been collected from different Bangladeshi people's faces on four angels. In this work, CNN, Harr Cascade, Cascaded CNN, Deep CNN & MTCNN are these five machine learning approaches implemented to get the best accuracy of our dataset. After creating and running the model, Multi-Task Convolutional Neural Network (MTCNN) achieved 96.2% best model accuracy with training data rather than other machine learning models.

What Can Secondary Predictions Tell Us? An Exploration on Question-Answering with SQuAD-v2.0

Performance in natural language processing, and specifically for the question-answer task, is typically measured by comparing a model\'s most confident (primary) prediction to golden answers (the ground truth). We are making the case that it is also useful to quantify how close a model came to predicting a correct answer even for examples that failed. We define the Golden Rank (GR) of an example as the rank of its most confident prediction that exactly matches a ground truth, and show why such a match always exists. For the 16 transformer models we analyzed, the majority of exactly matched golden answers in secondary prediction space hover very close to the top rank. We refer to secondary predictions as those ranking above 0 in descending confidence probability order. We demonstrate how the GR can be used to classify questions and visualize their spectrum of difficulty, from persistent near successes to persistent extreme failures. We derive a new aggregate statistic over entire test sets, named the Golden Rank Interpolated Median (GRIM) that quantifies the proximity of failed predictions to the top choice made by the model. To develop some intuition and explore the applicability of these metrics we use the Stanford Question Answering Dataset (SQuAD-2) and a few popular transformer models from the Hugging Face hub. We first demonstrate that the GRIM is not directly correlated with the F1 and exact match (EM) scores. We then calculate and visualize these scores for various transformer architectures, probe their applicability in error analysis by clustering failed predictions, and compare how they relate to other training diagnostics such as the EM and F1 scores. We finally suggest various research goals, such as broadening data collection for these metrics and their possible use in adversarial training.

Intrinsic Anomaly Detection for Multi-Variate Time Series

We introduce a novel, practically relevant variation of the anomaly detection problem in multi-variate time series: intrinsic anomaly detection. It appears in diverse practical scenarios ranging from DevOps to IoT, where we want to recognize failures of a system that operates under the influence of a surrounding environment. Intrinsic anomalies are changes in the functional dependency structure between time series that represent an environment and time series that represent the internal state of a system that is placed in said environment. We formalize this problem, provide under-studied public and new purpose-built data sets for it, and present methods that handle intrinsic anomaly detection. These address the short-coming of existing anomaly detection methods that cannot differentiate between expected changes in the system's state and unexpected ones, i.e., changes in the system that deviate from the environment's influence. Our most promising approach is fully unsupervised and combines adversarial learning and time series representation learning, thereby addressing problems such as label sparsity and subjectivity, while allowing to navigate and improve notoriously problematic anomaly detection data sets.

Deformable Graph Transformer

Transformer-based models have been widely used and achieved state-of-the-art performance in various domains such as natural language processing and computer vision. Recent works show that Transformers can also be generalized to graph-structured data. However, the success is limited to small-scale graphs due to technical challenges such as the quadratic complexity in regards to the number of nodes and non-local aggregation that often leads to inferior generalization performance to conventional graph neural networks. In this paper, to address these issues, we propose Deformable Graph Transformer (DGT) that performs sparse attention with dynamically sampled key and value pairs. Specifically, our framework first constructs multiple node sequences with various criteria to consider both structural and semantic proximity. Then, the sparse attention is applied to the node sequences for learning node representations with a reduced computational cost. We also design simple and effective positional encodings to capture structural similarity and distance between nodes. Experiments demonstrate that our novel graph Transformer consistently outperforms existing Transformer-based models and shows competitive performance compared to state-of-the-art models on 8 graph benchmark datasets including large-scale graphs.

Active Exploration via Experiment Design in Markov Chains

A key challenge in science and engineering is to design experiments to learn about some unknown quantity of interest. Classical experimental design optimally allocates the experimental budget to maximize a notion of utility (e.g., reduction in uncertainty about the unknown quantity). We consider a rich setting, where the experiments are associated with states in a {\em Markov chain}, and we can only choose them by selecting a {\em policy} controlling the state transitions. This problem captures important applications, from exploration in reinforcement learning to spatial monitoring tasks. We propose an algorithm -- \textsc{markov-design} -- that efficiently selects policies whose measurement allocation \emph{provably converges to the optimal one}. The algorithm is sequential in nature, adapting its choice of policies (experiments) informed by past measurements. In addition to our theoretical analysis, we showcase our framework on applications in ecological surveillance and pharmacology.

Spherical Channels for Modeling Atomic Interactions

Modeling the energy and forces of atomic systems is a fundamental problem in computational chemistry with the potential to help address many of the world's most pressing problems, including those related to energy scarcity and climate change. These calculations are traditionally performed using Density Functional Theory, which is computationally very expensive. Machine learning has the potential to dramatically improve the efficiency of these calculations from days or hours to seconds. We propose the Spherical Channel Network (SCN) to model atomic energies and forces. The SCN is a graph neural network where nodes represent atoms and edges their neighboring atoms. The atom embeddings are a set of spherical functions, called spherical channels, represented using spherical harmonics. We demonstrate, that by rotating the embeddings based on the 3D edge orientation, more information may be utilized while maintaining the rotational equivariance of the messages. While equivariance is a desirable property, we find that by relaxing this constraint in both message passing and aggregation, improved accuracy may be achieved. We demonstrate state-of-the-art results on the large-scale Open Catalyst 2020 dataset in both energy and force prediction for numerous tasks and metrics.

On the Rényi Cross-Entropy

The R\'{e}nyi cross-entropy measure between two distributions, a generalization of the Shannon cross-entropy, was recently used as a loss function for the improved design of deep learning generative adversarial networks. In this work, we examine the properties of this measure and derive closed-form expressions for it when one of the distributions is fixed and when both distributions belong to the exponential family. We also analytically determine a formula for the cross-entropy rate for stationary Gaussian processes and for finite-alphabet Markov sources.

An Empirical Study of Challenges in Converting Deep Learning Models

There is an increase in deploying Deep Learning (DL)-based software systems in real-world applications. Usually DL models are developed and trained using DL frameworks that have their own internal mechanisms/formats to represent and train DL models, and usually those formats cannot be recognized by other frameworks. Moreover, trained models are usually deployed in environments different from where they were developed. To solve the interoperability issue and make DL models compatible with different frameworks/environments, some exchange formats are introduced for DL models, like ONNX and CoreML. However, ONNX and CoreML were never empirically evaluated by the community to reveal their prediction accuracy, performance, and robustness after conversion. Poor accuracy or non-robust behavior of converted models may lead to poor quality of deployed DL-based software systems. We conduct, in this paper, the first empirical study to assess ONNX and CoreML for converting trained DL models. In our systematic approach, two popular DL frameworks, Keras and PyTorch, are used to train five widely used DL models on three popular datasets. The trained models are then converted to ONNX and CoreML and transferred to two runtime environments designated for such formats, to be evaluated. We investigate the prediction accuracy before and after conversion. Our results unveil that the prediction accuracy of converted models are at the same level of originals. The performance (time cost and memory consumption) of converted models are studied as well. The size of models are reduced after conversion, which can result in optimized DL-based software deployment. Converted models are generally assessed as robust at the same level of originals. However, obtained results show that CoreML models are more vulnerable to adversarial attacks compared to ONNX.

Bottleneck Low-rank Transformers for Low-resource Spoken Language Understanding

End-to-end spoken language understanding (SLU) systems benefit from pretraining on large corpora, followed by fine-tuning on application-specific data. The resulting models are too large for on-edge applications. For instance, BERT-based systems contain over 110M parameters. Observing the model is overparameterized, we propose lean transformer structure where the dimension of the attention mechanism is automatically reduced using group sparsity. We propose a variant where the learned attention subspace is transferred to an attention bottleneck layer. In a low-resource setting and without pre-training, the resulting compact SLU model achieves accuracies competitive with pre-trained large models.

Multistep Automated Data Labelling Procedure (MADLaP) for Thyroid Nodules on Ultrasound: An Artificial Intelligence Approach for Automating Image Annotation

Machine learning (ML) for diagnosis of thyroid nodules on ultrasound is an active area of research. However, ML tools require large, well-labelled datasets, the curation of which is time-consuming and labor-intensive. The purpose of our study was to develop and test a deep-learning-based tool to facilitate and automate the data annotation process for thyroid nodules; we named our tool Multistep Automated Data Labelling Procedure (MADLaP). MADLaP was designed to take multiple inputs included pathology reports, ultrasound images, and radiology reports. Using multiple step-wise modules including rule-based natural language processing, deep-learning-based imaging segmentation, and optical character recognition, MADLaP automatically identified images of a specific thyroid nodule and correctly assigned a pathology label. The model was developed using a training set of 378 patients across our health system and tested on a separate set of 93 patients. Ground truths for both sets were selected by an experienced radiologist. Performance metrics including yield (how many labeled images the model produced) and accuracy (percentage correct) were measured using the test set. MADLaP achieved a yield of 63% and an accuracy of 83%. The yield progressively increased as the input data moved through each module, while accuracy peaked part way through. Error analysis showed that inputs from certain examination sites had lower accuracy (40%) than the other sites (90%, 100%). MADLaP successfully created curated datasets of labeled ultrasound images of thyroid nodules. While accurate, the relatively suboptimal yield of MADLaP exposed some challenges when trying to automatically label radiology images from heterogeneous sources. The complex task of image curation and annotation could be automated, allowing for enrichment of larger datasets for use in machine learning development.

Reinforcement Learning in Medical Image Analysis: Concepts, Applications, Challenges, and Future Directions

Motivation: Medical image analysis involves tasks to assist physicians in qualitative and quantitative analysis of lesions or anatomical structures, significantly improving the accuracy and reliability of diagnosis and prognosis. Traditionally, these tasks are finished by physicians or medical physicists and lead to two major problems: (i) low efficiency; (ii) biased by personal experience. In the past decade, many machine learning methods have been applied to accelerate and automate the image analysis process. Compared to the enormous deployments of supervised and unsupervised learning models, attempts to use reinforcement learning in medical image analysis are scarce. This review article could serve as the stepping-stone for related research. Significance: From our observation, though reinforcement learning has gradually gained momentum in recent years, many researchers in the medical analysis field find it hard to understand and deploy in clinics. One cause is lacking well-organized review articles targeting readers lacking professional computer science backgrounds. Rather than providing a comprehensive list of all reinforcement learning models in medical image analysis, this paper may help the readers to learn how to formulate and solve their medical image analysis research as reinforcement learning problems. Approach & Results: We selected published articles from Google Scholar and PubMed. Considering the scarcity of related articles, we also included some outstanding newest preprints. The papers are carefully reviewed and categorized according to the type of image analysis task. We first review the basic concepts and popular models of reinforcement learning. Then we explore the applications of reinforcement learning models in landmark detection. Finally, we conclude the article by discussing the reviewed reinforcement learning approaches' limitations and possible improvements.

Learning Time Delay Systems with Neural Ordinary Differential Equations

A novel way of using neural networks to learn the dynamics of time delay systems from sequential data is proposed. A neural network with trainable delays is used to approximate the right hand side of a delay differential equation. We relate the delay differential equation to an ordinary differential equation by discretizing the time history and train the corresponding neural ordinary differential equation (NODE) to learn the dynamics. An example on learning the dynamics of the Mackey-Glass equation using data from chaotic behavior is given. After learning both the nonlinearity and the time delay, we demonstrate that the bifurcation diagram of the neural network matches that of the original system.

TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s

This paper presents a novel nearest neighbor search algorithm achieving TPU (Google Tensor Processing Unit) peak performance, outperforming state-of-the-art GPU algorithms with similar level of recall. The design of the proposed algorithm is motivated by an accurate accelerator performance model that takes into account both the memory and instruction bottlenecks. Our algorithm comes with an analytical guarantee of recall in expectation and does not require maintaining sophisticated index data structure or tuning, making it suitable for applications with frequent updates. Our work is available in the open-source package of Jax and Tensorflow on TPU.

Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs

This paper studies the problem of forecasting general stochastic processes using an extension of the Neural Jump ODE (NJ-ODE) framework. While NJ-ODE was the first framework to establish convergence guarantees for the prediction of irregularly observed time-series, these results were limited to data stemming from It\^o-diffusions with complete observations, in particular Markov processes where all coordinates are observed simultaneously. In this work, we generalise these results to generic, possibly non-Markovian or discontinuous, stochastic processes with incomplete observations, by utilising the reconstruction properties of the signature transform. These theoretical results are supported by empirical studies, where it is shown that the path-dependent NJ-ODE outperforms the original NJ-ODE framework in the case of non-Markovian data.

Neural Integro-Differential Equations

Modeling continuous dynamical systems from discretely sampled observations is a fundamental problem in data science. Often, such dynamics are the result of non-local processes that present an integral over time. As such, these systems are modeled with Integro-Differential Equations (IDEs); generalizations of differential equations that comprise both an integral and a differential component. For example, brain dynamics are not accurately modeled by differential equations since their behavior is non-Markovian, i.e. dynamics are in part dictated by history. Here, we introduce the Neural IDE (NIDE), a framework that models ordinary and integral components of IDEs using neural networks. We test NIDE on several toy and brain activity datasets and demonstrate that NIDE outperforms other models, including Neural ODE. These tasks include time extrapolation as well as predicting dynamics from unseen initial conditions, which we test on whole-cortex activity recordings in freely behaving mice. Further, we show that NIDE can decompose dynamics into its Markovian and non-Markovian constituents, via the learned integral operator, which we test on fMRI brain activity recordings of people on ketamine. Finally, the integrand of the integral operator provides a latent space that gives insight into the underlying dynamics, which we demonstrate on wide-field brain imaging recordings. Altogether, NIDE is a novel approach that enables modeling of complex non-local dynamics with neural networks.

A Perturbation Bound on the Subspace Estimator from Canonical Projections

This paper derives a perturbation bound on the optimal subspace estimator obtained from a subset of its canonical projections contaminated by noise. This fundamental result has important implications in matrix completion, subspace clustering, and related problems.

NumS: Scalable Array Programming for the Cloud

Scientists increasingly rely on Python tools to perform scalable distributed memory array operations using rich, NumPy-like expressions. However, many of these tools rely on dynamic schedulers optimized for abstract task graphs, which often encounter memory and network bandwidth-related bottlenecks due to sub-optimal data and operator placement decisions. Tools built on the message passing interface (MPI), such as ScaLAPACK and SLATE, have better scaling properties, but these solutions require specialized knowledge to use. In this work, we present NumS, an array programming library which optimizes NumPy-like expressions on task-based distributed systems. This is achieved through a novel scheduler called Load Simulated Hierarchical Scheduling (LSHS). LSHS is a local search method which optimizes operator placement by minimizing maximum memory and network load on any given node within a distributed system. Coupled with a heuristic for load balanced data layouts, our approach is capable of attaining communication lower bounds on some common numerical operations, and our empirical study shows that LSHS enhances performance on Ray by decreasing network load by a factor of 2x, requiring 4x less memory, and reducing execution time by 10x on the logistic regression problem. On terabyte-scale data, NumS achieves competitive performance to SLATE on DGEMM, up to 20x speedup over Dask on a key operation for tensor factorization, and a 2x speedup on logistic regression compared to Dask ML and Spark's MLlib.

Collecting high-quality adversarial data for machine reading comprehension tasks with humans and models in the loop

We present our experience as annotators in the creation of high-quality, adversarial machine-reading-comprehension data for extractive QA for Task 1 of the First Workshop on Dynamic Adversarial Data Collection (DADC). DADC is an emergent data collection paradigm with both models and humans in the loop. We set up a quasi-experimental annotation design and perform quantitative analyses across groups with different numbers of annotators focusing on successful adversarial attacks, cost analysis, and annotator confidence correlation. We further perform a qualitative analysis of our perceived difficulty of the task given the different topics of the passages in our dataset and conclude with recommendations and suggestions that might be of value to people working on future DADC tasks and related annotation interfaces.

Applications of Reinforcement Learning in Finance -- Trading with a Double Deep Q-Network

This paper presents a Double Deep Q-Network algorithm for trading single assets, namely the E-mini S&P 500 continuous futures contract. We use a proven setup as the foundation for our environment with multiple extensions. The features of our trading agent are constantly being expanded to include additional assets such as commodities, resulting in four models. We also respond to environmental conditions, including costs and crises. Our trading agent is first trained for a specific time period and tested on new data and compared with the long-and-hold strategy as a benchmark (market). We analyze the differences between the various models and the in-sample/out-of-sample performance with respect to the environment. The experimental results show that the trading agent follows an appropriate behavior. It can adjust its policy to different circumstances, such as more extensive use of the neutral position when trading costs are present. Furthermore, the net asset value exceeded that of the benchmark, and the agent outperformed the market in the test set. We provide initial insights into the behavior of an agent in a financial domain using a DDQN algorithm. The results of this study can be used for further development.

Online Anomaly Detection Based On Reservoir Sampling and LOF for IoT devices

The growing number of IoT devices and their use to monitor the operation of machines and equipment increases interest in anomaly detection algorithms running on devices. However, the difficulty is the limitations of the available computational and memory resources on the devices. In the case of microcontrollers (MCUs), these are single megabytes of program and several hundred kilobytes of working memory. Consequently, algorithms must be appropriately matched to the capabilities of the devices. In the paper, we analyse the processing pipeline for anomaly detection and implementation of the Local Outliner Factor (LOF) algorithm on a MCU. We also show that it is possible to train such an algorithm directly on the device, which gives great potential to use the solution in real devices.

Supervised Training of Conditional Monge Maps

Optimal transport (OT) theory describes general principles to define and select, among many possible choices, the most efficient way to map a probability measure onto another. That theory has been mostly used to estimate, given a pair of source and target probability measures $(\mu,\nu)$, a parameterized map $T_\theta$ that can efficiently map $\mu$ onto $\nu$. In many applications, such as predicting cell responses to treatments, the data measures $\mu,\nu$ (features of untreated/treated cells) that define optimal transport problems do not arise in isolation but are associated with a context $c$ (the treatment). To account for and incorporate that context in OT estimation, we introduce CondOT, an approach to estimate OT maps conditioned on a context variable, using several pairs of measures $(\mu_i, \nu_i)$ tagged with a context label $c_i$. Our goal is to % extract from a dataset of labeled pairs $\{(c_i, (\mu_i, \nu_i))\}$ learn a global map $\mathcal{T}_{\theta}$ which is not only expected to fit em all pairs in the dataset $\{(c_i, (\mu_i, \nu_i))\}$, i.e., $\mathcal{T}_{\theta}(c_i) \sharp\mu_i \approx \nu_i$, but should generalize to produce meaningful maps $\mathcal{T}_{\theta}(c_{\text{new}})$ conditioned on unseen contexts $c_{\text{new}}$. Our approach harnesses and provides a novel usage for partially input convex neural networks, for which we introduce a robust and efficient initialization strategy inspired by Gaussian approximations. We demonstrate the ability of CondOT to infer the effect of an arbitrary combination of genetic or therapeutic perturbations on single cells, using only observations of the effects of said perturbations separately.

Semi-supervised Contrastive Outlier removal for Pseudo Expectation Maximization (SCOPE)

Semi-supervised learning is the problem of training an accurate predictive model by combining a small labeled dataset with a presumably much larger unlabeled dataset. Many methods for semi-supervised deep learning have been developed, including pseudolabeling, consistency regularization, and contrastive learning techniques. Pseudolabeling methods however are highly susceptible to confounding, in which erroneous pseudolabels are assumed to be true labels in early iterations, thereby causing the model to reinforce its prior biases and thereby fail to generalize to strong predictive performance. We present a new approach to suppress confounding errors through a method we describe as Semi-supervised Contrastive Outlier removal for Pseudo Expectation Maximization (SCOPE). Like basic pseudolabeling, SCOPE is related to Expectation Maximization (EM), a latent variable framework which can be extended toward understanding cluster-assumption deep semi-supervised algorithms. However, unlike basic pseudolabeling which fails to adequately take into account the probability of the unlabeled samples given the model, SCOPE introduces an outlier suppression term designed to improve the behavior of EM iteration given a discrimination DNN backbone in the presence of outliers. Our results show that SCOPE greatly improves semi-supervised classification accuracy over a baseline, and furthermore when combined with consistency regularization achieves the highest reported accuracy for the semi-supervised CIFAR-10 classification task using 250 and 4000 labeled samples. Moreover, we show that SCOPE reduces the prevalence of confounding errors during pseudolabeling iterations by pruning erroneous high-confidence pseudolabeled samples that would otherwise contaminate the labeled set in subsequent retraining iterations.

GAN-based Intrinsic Exploration For Sample Efficient Reinforcement Learning

In this study, we address the problem of efficient exploration in reinforcement learning. Most common exploration approaches depend on random action selection, however these approaches do not work well in environments with sparse or no rewards. We propose Generative Adversarial Network-based Intrinsic Reward Module that learns the distribution of the observed states and sends an intrinsic reward that is computed as high for states that are out of distribution, in order to lead agent to unexplored states. We evaluate our approach in Super Mario Bros for a no reward setting and in Montezuma's Revenge for a sparse reward setting and show that our approach is indeed capable of exploring efficiently. We discuss a few weaknesses and conclude by discussing future works.

Target alignment in truncated kernel ridge regression

Kernel ridge regression (KRR) has recently attracted renewed interest due to its potential for explaining the transient effects, such as double descent, that emerge during neural network training. In this work, we study how the alignment between the target function and the kernel affects the performance of the KRR. We focus on the truncated KRR (TKRR) which utilizes an additional parameter that controls the spectral truncation of the kernel matrix. We show that for polynomial alignment, there is an \emph{over-aligned} regime, in which TKRR can achieve a faster rate than what is achievable by full KRR. The rate of TKRR can improve all the way to the parametric rate, while that of full KRR is capped at a sub-optimal value. This shows that target alignemnt can be better leveraged by utilizing spectral truncation in kernel methods. We also consider the bandlimited alignment setting and show that the regularization surface of TKRR can exhibit transient effects including multiple descent and non-monotonic behavior. Our results show that there is a strong and quantifable relation between the shape of the \emph{alignment spectrum} and the generalization performance of kernel methods, both in terms of rates and in finite samples.

No imputation without representation

By filling in missing values in datasets, imputation allows these datasets to be used with algorithms that cannot handle missing values by themselves. However, missing values may in principle contribute useful information that is lost through imputation. The missing-indicator approach can be used in combination with imputation to instead represent this information as a part of the dataset. There are several theoretical considerations why missing-indicators may or may not be beneficial, but there has not been any large-scale practical experiment on real-life datasets to test this question for machine learning predictions. We perform this experiment for three imputation strategies and a range of different classification algorithms, on the basis of twenty real-life datasets. We find that on these datasets, missing-indicators generally increase classification performance. In addition, we find no evidence for most algorithms that nearest neighbour and iterative imputation lead to better performance than simple mean/mode imputation. Therefore, we recommend the use of missing-indicators with mean/mode imputation as a safe default, with the caveat that for decision trees, pruning is necessary to prevent overfitting. In a follow-up experiment, we determine attribute-specific missingness thresholds for each classifier above which missing-indicators are more likely than not to increase classification performance, and observe that these thresholds are much lower for categorical than for numerical attributes. Finally, we argue that mean imputation of numerical attributes may preserve some of the information from missing values, and we show that in the absence of missing-indicators, it can similarly be useful to apply mean imputation to one-hot encoded categorical attributes instead of mode imputation.

Masked World Models for Visual Control

Visual model-based reinforcement learning (RL) has the potential to enable sample-efficient robot learning from visual observations. Yet the current approaches typically train a single model end-to-end for learning both visual representations and dynamics, making it difficult to accurately model the interaction between robots and small objects. In this work, we introduce a visual model-based RL framework that decouples visual representation learning and dynamics learning. Specifically, we train an autoencoder with convolutional layers and vision transformers (ViT) to reconstruct pixels given masked convolutional features, and learn a latent dynamics model that operates on the representations from the autoencoder. Moreover, to encode task-relevant information, we introduce an auxiliary reward prediction objective for the autoencoder. We continually update both autoencoder and dynamics model using online samples collected from environment interaction. We demonstrate that our decoupling approach achieves state-of-the-art performance on a variety of visual robotic tasks from Meta-world and RLBench, e.g., we achieve 81.7% success rate on 50 visual robotic manipulation tasks from Meta-world, while the baseline achieves 67.9%. Code is available on the project website: https://sites.google.com/view/mwm-rl.

PyEPO: A PyTorch-based End-to-End Predict-then-Optimize Library for Linear and Integer Programming

In deterministic optimization, it is typically assumed that all parameters of the problem are fixed and known. In practice, however, some parameters may be a priori unknown but can be estimated from historical data. A typical predict-then-optimize approach separates predictions and optimization into two stages. Recently, end-to-end predict-then-optimize has become an attractive alternative. In this work, we present the PyEPO package, a PyTorch-based end-to-end predict-then-optimize library in Python. To the best of our knowledge, PyEPO (pronounced like "pineapple" with a silent "n") is the first such generic tool for linear and integer programming with predicted objective function coefficients. It provides two base algorithms: the first is based on the convex surrogate loss function from the seminal work of Elmachtoub & Grigas (2021), and the second is based on the differentiable black-box solver approach of Vlastelica et al. (2019). PyEPO provides a simple interface for the definition of new optimization problems, the implementation of state-of-the-art predict-then-optimize training algorithms, the use of custom neural network architectures, and the comparison of end-to-end approaches with the two-stage approach. PyEPO enables us to conduct a comprehensive set of experiments comparing a number of end-to-end and two-stage approaches along axes such as prediction accuracy, decision quality, and running time on problems such as Shortest Path, Multiple Knapsack, and the Traveling Salesperson Problem. We discuss some empirical insights from these experiments which could guide future research. PyEPO and its documentation are available at https://github.com/khalil-research/PyEPO.

Gaussian Latent Dirichlet Allocation for Discrete Human State Discovery

In this article we propose and validate an unsupervised probabilistic model, Gaussian Latent Dirichlet Allocation (GLDA), for the problem of discrete state discovery from repeated, multivariate psychophysiological samples collected from multiple, inherently distinct, individuals. Psychology and medical research heavily involves measuring potentially related but individually inconclusive variables from a cohort of participants to derive diagnosis, necessitating clustering analysis. Traditional probabilistic clustering models such as Gaussian Mixture Model (GMM) assume a global mixture of component distributions, which may not be realistic for observations from different patients. The GLDA model borrows the individual-specific mixture structure from a popular topic model Latent Dirichlet Allocation (LDA) in Natural Language Processing and merges it with the Gaussian component distributions of GMM to suit continuous type data. We implemented GLDA using STAN (a probabilistic modeling language) and applied it on two datasets, one containing Ecological Momentary Assessments (EMA) and the other heart measures from electrocardiogram and impedance cardiograph. We found that in both datasets the GLDA-learned class weights achieved significantly higher correlations with clinically assessed depression, anxiety, and stress scores than those produced by the baseline GMM. Our findings demonstrate the advantage of GLDA over conventional finite mixture models for human state discovery from repeated multivariate data, likely due to better characterization of potential underlying between-participant differences. Future work is required to validate the utility of this model on a broader range of applications.

Zero-Shot Building Control

Heating and cooling systems in buildings account for 31% of global energy use, much of which are regulated by Rule Based Controllers (RBCs) that neither maximise energy efficiency nor minimise emissions by interacting optimally with the grid. Control via Reinforcement Learning (RL) has been shown to significantly improve building energy efficiency, but existing solutions require pre-training in simulators that are prohibitively expensive to obtain for every building in the world. In response, we show it is possible to perform safe, zero-shot control of buildings by combining ideas from system identification and model-based RL. We call this combination PEARL (Probabilistic Emission-Abating Reinforcement Learning) and show it reduces emissions without pre-training, needing only a three hour commissioning period. In experiments across three varied building energy simulations, we show PEARL outperforms an existing RBC once, and popular RL baselines in all cases, reducing building emissions by as much as 31% whilst maintaining thermal comfort.

Latent Combinational Game Design

We present an approach for generating playable games that blend a given set of games in a desired combination using deep generative latent variable models. We refer to this approach as latent combinational game design -- latent since we use learned latent representations to perform blending, combinational since game blending is a combinational creativity process and game design since the approach generates novel, playable games. We use Gaussian Mixture Variational Autoencoders (GMVAEs), which use a mixture of Gaussians to model the VAE latent space. Through supervised training, each component learns to encode levels from one game and lets us define new, blended games as linear combinations of these learned components. This enables generating new games that blend the input games as well as control the relative proportions of each game in the blend. We also extend prior work using conditional VAEs to perform blending and compare against the GMVAE. Our results show that both models can generate playable blended games that blend the input games in the desired proportions.

Evaluating Understanding on Conceptual Abstraction Benchmarks

A long-held objective in AI is to build systems that understand concepts in a humanlike way. Setting aside the difficulty of building such a system, even trying to evaluate one is a challenge, due to present-day AI's relative opacity and its proclivity for finding shortcut solutions. This is exacerbated by humans' tendency to anthropomorphize, assuming that a system that can recognize one instance of a concept must also understand other instances, as a human would. In this paper, we argue that understanding a concept requires the ability to use it in varied contexts. Accordingly, we propose systematic evaluations centered around concepts, by probing a system's ability to use a given concept in many different instantiations. We present case studies of such an evaluations on two domains -- RAVEN (inspired by Raven's Progressive Matrices) and the Abstraction and Reasoning Corpus (ARC) -- that have been used to develop and assess abstraction abilities in AI systems. Our concept-based approach to evaluation reveals information about AI systems that conventional test sets would have left hidden.

Integral Transforms in a Physics-Informed (Quantum) Neural Network setting: Applications & Use-Cases

In many computational problems in engineering and science, function or model differentiation is essential, but also integration is needed. An important class of computational problems include so-called integro-differential equations which include both integrals and derivatives of a function. In another example, stochastic differential equations can be written in terms of a partial differential equation of a probability density function of the stochastic variable. To learn characteristics of the stochastic variable based on the density function, specific integral transforms, namely moments, of the density function need to be calculated. Recently, the machine learning paradigm of Physics-Informed Neural Networks emerged with increasing popularity as a method to solve differential equations by leveraging automatic differentiation. In this work, we propose to augment the paradigm of Physics-Informed Neural Networks with automatic integration in order to compute complex integral transforms on trained solutions, and to solve integro-differential equations where integrals are computed on-the-fly during training. Furthermore, we showcase the techniques in various application settings, numerically simulating quantum computer-based neural networks as well as classical neural networks.

DayDreamer: World Models for Physical Robot Learning

To solve tasks in complex environments, robots need to learn from experience. Deep reinforcement learning is a common approach to robot learning but requires a large amount of trial and error to learn, limiting its deployment in the physical world. As a consequence, many advances in robot learning rely on simulators. On the other hand, learning inside of simulators fails to capture the complexity of the real world, is prone to simulator inaccuracies, and the resulting behaviors do not adapt to changes in the world. The Dreamer algorithm has recently shown great promise for learning from small amounts of interaction by planning within a learned world model, outperforming pure reinforcement learning in video games. Learning a world model to predict the outcomes of potential actions enables planning in imagination, reducing the amount of trial and error needed in the real environment. However, it is unknown whether Dreamer can facilitate faster learning on physical robots. In this paper, we apply Dreamer to 4 robots to learn online and directly in the real world, without simulators. Dreamer trains a quadruped robot to roll off its back, stand up, and walk from scratch and without resets in only 1 hour. We then push the robot and find that Dreamer adapts within 10 minutes to withstand perturbations or quickly roll over and stand back up. On two different robotic arms, Dreamer learns to pick and place multiple objects directly from camera images and sparse rewards, approaching human performance. On a wheeled robot, Dreamer learns to navigate to a goal position purely from camera images, automatically resolving ambiguity about the robot orientation. Using the same hyperparameters across all experiments, we find that Dreamer is capable of online learning in the real world, establishing a strong baseline. We release our infrastructure for future applications of world models to robot learning.

Risk Perspective Exploration in Distributional Reinforcement Learning

Distributional reinforcement learning demonstrates state-of-the-art performance in continuous and discrete control settings with the features of variance and risk, which can be used to explore. However, the exploration method employing the risk property is hard to find, although numerous exploration methods in Distributional RL employ the variance of return distribution per action. In this paper, we present risk scheduling approaches that explore risk levels and optimistic behaviors from a risk perspective. We demonstrate the performance enhancement of the DMIX algorithm using risk scheduling in a multi-agent setting with comprehensive experiments.

Verifiable Goal Recognition for Autonomous Driving with Occlusions

When used in autonomous driving, goal recognition allows the future behaviour of other vehicles to be more accurately predicted. A recent goal recognition method for autonomous vehicles, GRIT, has been shown to be fast, accurate, interpretable and verifiable. In autonomous driving, vehicles can encounter novel scenarios that were unseen during training, and the environment is partially observable due to occlusions. However, GRIT can only operate in fixed frame scenarios, with full observability. We present a novel goal recognition method named Goal Recognition with Interpretable Trees under Occlusion (OGRIT), which solves these shortcomings of GRIT. We demonstrate that OGRIT can generalise between different scenarios and handle missing data due to occlusions, while still being fast, accurate, interpretable and verifiable.

Rethinking Optimization with Differentiable Simulation from a Global Perspective

Differentiable simulation is a promising toolkit for fast gradient-based policy optimization and system identification. However, existing approaches to differentiable simulation have largely tackled scenarios where obtaining smooth gradients has been relatively easy, such as systems with mostly smooth dynamics. In this work, we study the challenges that differentiable simulation presents when it is not feasible to expect that a single descent reaches a global optimum, which is often a problem in contact-rich scenarios. We analyze the optimization landscapes of diverse scenarios that contain both rigid bodies and deformable objects. In dynamic environments with highly deformable objects and fluids, differentiable simulators produce rugged landscapes with nonetheless useful gradients in some parts of the space. We propose a method that combines Bayesian optimization with semi-local 'leaps' to obtain a global search method that can use gradients effectively, while also maintaining robust performance in regions with noisy gradients. We show that our approach outperforms several gradient-based and gradient-free baselines on an extensive set of experiments in simulation, and also validate the method using experiments with a real robot and deformables. Videos and supplementary materials are available at https://tinyurl.com/globdiff

Generative Anomaly Detection for Time Series Datasets

Traffic congestion anomaly detection is of paramount importance in intelligent traffic systems. The goals of transportation agencies are two-fold: to monitor the general traffic conditions in the area of interest and to locate road segments under abnormal congestion states. Modeling congestion patterns can achieve these goals for citywide roadways, which amounts to learning the distribution of multivariate time series (MTS). However, existing works are either not scalable or unable to capture the spatial-temporal information in MTS simultaneously. To this end, we propose a principled and comprehensive framework consisting of a data-driven generative approach that can perform tractable density estimation for detecting traffic anomalies. Our approach first clusters segments in the feature space and then uses conditional normalizing flow to identify anomalous temporal snapshots at the cluster level in an unsupervised setting. Then, we identify anomalies at the segment level by using a kernel density estimator on the anomalous cluster. Extensive experiments on synthetic datasets show that our approach significantly outperforms several state-of-the-art congestion anomaly detection and diagnosis methods in terms of Recall and F1-Score. We also use the generative model to sample labeled data, which can train classifiers in a supervised setting, alleviating the lack of labeled data for anomaly detection in sparse settings.

Building Matters: Spatial Variability in Machine Learning Based Thermal Comfort Prediction in Winters

Thermal comfort in indoor environments has an enormous impact on the health, well-being, and performance of occupants. Given the focus on energy efficiency and Internet-of-Things enabled smart buildings, machine learning (ML) is being increasingly used for data-driven thermal comfort (TC) prediction. Generally, ML-based solutions are proposed for air-conditioned or HVAC ventilated buildings and the models are primarily designed for adults. On the other hand, naturally ventilated (NV) buildings are the norm in most countries. They are also ideal for energy conservation and long-term sustainability goals. However, the indoor environment of NV buildings lacks thermal regulation and varies significantly across spatial contexts. These factors make TC prediction extremely challenging. Thus, determining the impact of the building environment on the performance of TC models is important. Further, the generalization capability of TC prediction models across different NV indoor spaces needs to be studied. This work addresses these problems. Data is gathered through month-long field experiments conducted in 5 naturally ventilated school buildings, involving 512 primary school students. The impact of spatial variability on student comfort is demonstrated through variation in prediction accuracy (by as much as 71%). The influence of building environment on TC prediction is also demonstrated through variation in feature importance. Further, a comparative analysis of spatial variability in model performance is done for children (our dataset) and adults (ASHRAE-II database). Finally, the generalization capability of thermal comfort models in NV classrooms is assessed and major challenges are highlighted.

How to Steer Your Adversary: Targeted and Efficient Model Stealing Defenses with Gradient Redirection

Model stealing attacks present a dilemma for public machine learning APIs. To protect financial investments, companies may be forced to withhold important information about their models that could facilitate theft, including uncertainty estimates and prediction explanations. This compromise is harmful not only to users but also to external transparency. Model stealing defenses seek to resolve this dilemma by making models harder to steal while preserving utility for benign users. However, existing defenses have poor performance in practice, either requiring enormous computational overheads or severe utility trade-offs. To meet these challenges, we present a new approach to model stealing defenses called gradient redirection. At the core of our approach is a provably optimal, efficient algorithm for steering an adversary's training updates in a targeted manner. Combined with improvements to surrogate networks and a novel coordinated defense strategy, our gradient redirection defense, called GRAD${}^2$, achieves small utility trade-offs and low computational overhead, outperforming the best prior defenses. Moreover, we demonstrate how gradient redirection enables reprogramming the adversary with arbitrary behavior, which we hope will foster work on new avenues of defense.

Memory Safe Computations with XLA Compiler

Software packages like TensorFlow and PyTorch are designed to support linear algebra operations, and their speed and usability determine their success. However, by prioritising speed, they often neglect memory requirements. As a consequence, the implementations of memory-intensive algorithms that are convenient in terms of software design can often not be run for large problems due to memory overflows. Memory-efficient solutions require complex programming approaches with significant logic outside the computational framework. This impairs the adoption and use of such algorithms. To address this, we developed an XLA compiler extension that adjusts the computational data-flow representation of an algorithm according to a user-specified memory limit. We show that k-nearest neighbour and sparse Gaussian process regression methods can be run at a much larger scale on a single device, where standard implementations would have failed. Our approach leads to better use of hardware resources. We believe that further focus on removing memory constraints at a compiler level will widen the range of machine learning methods that can be developed in the future.

Learning Variable Impedance Control for Aerial Sliding on Uneven Heterogeneous Surfaces by Proprioceptive and Tactile Sensing

The recent development of novel aerial vehicles capable of physically interacting with the environment leads to new applications such as contact-based inspection. These tasks require the robotic system to exchange forces with partially-known environments, which may contain uncertainties including unknown spatially-varying friction properties and discontinuous variations of the surface geometry. Finding a control strategy that is robust against these environmental uncertainties remains an open challenge. This paper presents a learning-based adaptive control strategy for aerial sliding tasks. In particular, the gains of a standard impedance controller are adjusted in real-time by a policy based on the current control signals, proprioceptive measurements, and tactile sensing. This policy is trained in simulation with simplified actuator dynamics in a student-teacher learning setup. The real-world performance of the proposed approach is verified using a tilt-arm omnidirectional flying vehicle. The proposed controller structure combines data-driven and model-based control methods, enabling our approach to successfully transfer directly and without adaptation from simulation to the real platform. Compared to fine-tuned state of the art interaction control methods we achieve reduced tracking error and improved disturbance rejection.

Quantum Neural Architecture Search with Quantum Circuits Metric and Bayesian Optimization

Quantum neural networks are promising for a wide range of applications in the Noisy Intermediate-Scale Quantum era. As such, there is an increasing demand for automatic quantum neural architecture search. We tackle this challenge by designing a quantum circuits metric for Bayesian optimization with Gaussian process. To this goal, we propose a new quantum gates distance that characterizes the gates' action over every quantum state and provide a theoretical perspective on its geometrical properties. Our approach significantly outperforms the benchmark on three empirical quantum machine learning problems including training a quantum generative adversarial network, solving combinatorial optimization in the MaxCut problem, and simulating quantum Fourier transform. Our method can be extended to characterize behaviors of various quantum machine learning models.

On the universality of the volatility formation process: when machine learning and rough volatility agree

We train an LSTM network based on a pooled dataset made of hundreds of liquid stocks aiming to forecast the next daily realized volatility for all stocks. Showing the consistent outperformance of this universal LSTM relative to other asset-specific parametric models, we uncover nonparametric evidences of a universal volatility formation mechanism across assets relating past market realizations, including daily returns and volatilities, to current volatilities. A parsimonious parametric forecasting device combining the rough fractional stochastic volatility and quadratic rough Heston models with fixed parameters results in the same level of performance as the universal LSTM, which confirms the universality of the volatility formation process from a parametric perspective.

RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network

This work introduces the RevSilo, the first reversible module for bidirectional multi-scale feature fusion. Like other reversible methods, RevSilo eliminates the need to store hidden activations by recomputing them. Existing reversible methods, however, do not apply to multi-scale feature fusion and are therefore not applicable to a large class of networks. Bidirectional multi-scale feature fusion promotes local and global coherence and has become a de facto design principle for networks targeting spatially sensitive tasks e.g. HRNet and EfficientDet. When paired with high-resolution inputs, these networks achieve state-of-the-art results across various computer vision tasks, but training them requires substantial accelerator memory for saving large, multi-resolution activations. These memory requirements cap network size and limit progress. Using reversible recomputation, the RevSilo alleviates memory issues while still operating across resolution scales. Stacking RevSilos, we create RevBiFPN, a fully reversible bidirectional feature pyramid network. For classification, RevBiFPN is competitive with networks such as EfficientNet while using up to 19.8x lesser training memory. When fine-tuned on COCO, RevBiFPN provides up to a 2.5% boost in AP over HRNet using fewer MACs and a 2.4x reduction in training-time memory.

Learning the Solution Operator of Boundary Value Problems using Graph Neural Networks

As an alternative to classical numerical solvers for partial differential equations (PDEs) subject to boundary value constraints, there has been a surge of interest in investigating neural networks that can solve such problems efficiently. In this work, we design a general solution operator for two different time-independent PDEs using graph neural networks (GNNs) and spectral graph convolutions. We train the networks on simulated data from a finite elements solver on a variety of shapes and inhomogeneities. In contrast to previous works, we focus on the ability of the trained operator to generalize to previously unseen scenarios. Specifically, we test generalization to meshes with different shapes and superposition of solutions for a different number of inhomogeneities. We find that training on a diverse dataset with lots of variation in the finite element meshes is a key ingredient for achieving good generalization results in all cases. With this, we believe that GNNs can be used to learn solution operators that generalize over a range of properties and produce solutions much faster than a generic solver. Our dataset, which we make publicly available, can be used and extended to verify the robustness of these models under varying conditions.

Continual Learning with Transformers for Image Classification

In many real-world scenarios, data to train machine learning models become available over time. However, neural network models struggle to continually learn new concepts without forgetting what has been learnt in the past. This phenomenon is known as catastrophic forgetting and it is often difficult to prevent due to practical constraints, such as the amount of data that can be stored or the limited computation sources that can be used. Moreover, training large neural networks, such as Transformers, from scratch is very costly and requires a vast amount of training data, which might not be available in the application domain of interest. A recent trend indicates that dynamic architectures based on an expansion of the parameters can reduce catastrophic forgetting efficiently in continual learning, but this needs complex tuning to balance the growing number of parameters and barely share any information across tasks. As a result, they struggle to scale to a large number of tasks without significant overhead. In this paper, we validate in the computer vision domain a recent solution called Adaptive Distillation of Adapters (ADA), which is developed to perform continual learning using pre-trained Transformers and Adapters on text classification tasks. We empirically demonstrate on different classification tasks that this method maintains a good predictive performance without retraining the model or increasing the number of model parameters over the time. Besides it is significantly faster at inference time compared to the state-of-the-art methods.

Equivariant Priors for Compressed Sensing with Unknown Orientation

In compressed sensing, the goal is to reconstruct the signal from an underdetermined system of linear measurements. Thus, prior knowledge about the signal of interest and its structure is required. Additionally, in many scenarios, the signal has an unknown orientation prior to measurements. To address such recovery problems, we propose using equivariant generative models as a prior, which encapsulate orientation information in their latent space. Thereby, we show that signals with unknown orientations can be recovered with iterative gradient descent on the latent space of these models and provide additional theoretical recovery guarantees. We construct an equivariant variational autoencoder and use the decoder as generative prior for compressed sensing. We discuss additional potential gains of the proposed approach in terms of convergence and latency.

Safe Exploration Incurs Nearly No Additional Sample Complexity for Reward-free RL

While the primary goal of the exploration phase in reward-free reinforcement learning (RF-RL) is to reduce the uncertainty in the estimated model with minimum number of trajectories, in practice, the agent often needs to abide by certain safety constraint at the same time. It remains unclear how such safe exploration requirement would affect the corresponding sample complexity to achieve the desired optimality of the obtained policy in planning. In this work, we make a first attempt to answer this question. In particular, we consider the scenario where a safe baseline policy is known beforehand, and propose a unified Safe reWard-frEe ExploraTion (SWEET) framework. We then particularize the SWEET framework to the tabular and the low-rank MDP settings, and develop algorithms coined Tabular-SWEET and Low-rank-SWEET, respectively. Both algorithms leverage the concavity and continuity of the newly introduced truncated value functions, and are guaranteed to achieve zero constraint violation during exploration with high probability. Furthermore, both algorithms can provably find a near-optimal policy subject to any constraint in the planning phase. Remarkably, the sample complexities under both algorithms match or even outperform the state of the art in their constraint-free counterparts up to some constant factors, proving that safety constraint hardly increases the sample complexity for RF-RL.

Deep Neural Networks pruning via the Structured Perspective Regularization

In Machine Learning, Artificial Neural Networks (ANNs) are a very powerful tool, broadly used in many applications. Often, the selected (deep) architectures include many layers, and therefore a large amount of parameters, which makes training, storage and inference expensive. This motivated a stream of research about compressing the original networks into smaller ones without excessively sacrificing performances. Among the many proposed compression approaches, one of the most popular is \emph{pruning}, whereby entire elements of the ANN (links, nodes, channels, \ldots) and the corresponding weights are deleted. Since the nature of the problem is inherently combinatorial (what elements to prune and what not), we propose a new pruning method based on Operational Research tools. We start from a natural Mixed-Integer-Programming model for the problem, and we use the Perspective Reformulation technique to strengthen its continuous relaxation. Projecting away the indicator variables from this reformulation yields a new regularization term, which we call the Structured Perspective Regularization, that leads to structured pruning of the initial architecture. We test our method on some ResNet architectures applied to CIFAR-10, CIFAR-100 and ImageNet datasets, obtaining competitive performances w.r.t.~the state of the art for structured pruning.

Modeling Extraneous Activity Delays in Business Process Simulation

Business Process Simulation (BPS) is a common approach to estimate the impact of changes to a business process on its performance measures. For example, BPS allows us to estimate what would be the cycle time of a process if we automated one of its activities. The starting point of BPS is a business process model annotated with simulation parameters (a BPS model). Several studies have proposed methods to automatically discover BPS models from event logs via process mining. However, current techniques in this space discover BPS models that only capture waiting times caused by resource contention or resource unavailability. Oftentimes, a considerable portion of the waiting time in a business process is caused by extraneous delays, e.g. a resource waits for the customer to return a phone call. This paper proposes a method that discovers extraneous delays from input data, and injects timer events into a BPS model to capture the discovered delays. An empirical evaluation involving synthetic and real-life logs shows that the approach produces BPS models that better reflect the temporal dynamics of the process, relative to BPS models that do not capture extraneous delays.

Short-Term Plasticity Neurons Learning to Learn and Forget

Short-term plasticity (STP) is a mechanism that stores decaying memories in synapses of the cerebral cortex. In computing practice, STP has been used, but mostly in the niche of spiking neurons, even though theory predicts that it is the optimal solution to certain dynamic tasks. Here we present a new type of recurrent neural unit, the STP Neuron (STPN), which indeed turns out strikingly powerful. Its key mechanism is that synapses have a state, propagated through time by a self-recurrent connection-within-the-synapse. This formulation enables training the plasticity with backpropagation through time, resulting in a form of learning to learn and forget in the short term. The STPN outperforms all tested alternatives, i.e. RNNs, LSTMs, other models with fast weights, and differentiable plasticity. We confirm this in both supervised and reinforcement learning (RL), and in tasks such as Associative Retrieval, Maze Exploration, Atari video games, and MuJoCo robotics. Moreover, we calculate that, in neuromorphic or biological circuits, the STPN minimizes energy consumption across models, as it depresses individual synapses dynamically. Based on these, biological STP may have been a strong evolutionary attractor that maximizes both efficiency and computational power. The STPN now brings these neuromorphic advantages also to a broad spectrum of machine learning practice. Code is available at https://github.com/NeuromorphicComputing/stpn

On the amplification of security and privacy risks by post-hoc explanations in machine learning models

A variety of explanation methods have been proposed in recent years to help users gain insights into the results returned by neural networks, which are otherwise complex and opaque black-boxes. However, explanations give rise to potential side-channels that can be leveraged by an adversary for mounting attacks on the system. In particular, post-hoc explanation methods that highlight input dimensions according to their importance or relevance to the result also leak information that weakens security and privacy. In this work, we perform the first systematic characterization of the privacy and security risks arising from various popular explanation techniques. First, we propose novel explanation-guided black-box evasion attacks that lead to 10 times reduction in query count for the same success rate. We show that the adversarial advantage from explanations can be quantified as a reduction in the total variance of the estimated gradient. Second, we revisit the membership information leaked by common explanations. Contrary to observations in prior studies, via our modified attacks we show significant leakage of membership information (above 100% improvement over prior results), even in a much stricter black-box setting. Finally, we study explanation-guided model extraction attacks and demonstrate adversarial gains through a large reduction in query count.

Learning Symmetric Rules with SATNet

SATNet is a differentiable constraint solver with a custom backpropagation algorithm, which can be used as a layer in a deep-learning system. It is a promising proposal for bridging deep learning and logical reasoning. In fact, SATNet has been successfully applied to learn, among others, the rules of a complex logical puzzle, such as Sudoku, just from input and output pairs where inputs are given as images. In this paper, we show how to improve the learning of SATNet by exploiting symmetries in the target rules of a given but unknown logical puzzle or more generally a logical formula. We present SymSATNet, a variant of SATNet that translates the given symmetries of the target rules to a condition on the parameters of SATNet and requires that the parameters should have a particular parametric form that guarantees the condition. The requirement dramatically reduces the number of parameters to learn for the rules with enough symmetries, and makes the parameter learning of SymSATNet much easier than that of SATNet. We also describe a technique for automatically discovering symmetries of the target rules from examples. Our experiments with Sudoku and Rubik's cube show the substantial improvement of SymSATNet over the baseline SATNet.

Stain Isolation-based Guidance for Improved Stain Translation

Unsupervised and unpaired domain translation using generative adversarial neural networks, and more precisely CycleGAN, is state of the art for the stain translation of histopathology images. It often, however, suffers from the presence of cycle-consistent but non structure-preserving errors. We propose an alternative approach to the set of methods which, relying on segmentation consistency, enable the preservation of pathology structures. Focusing on immunohistochemistry (IHC) and multiplexed immunofluorescence (mIF), we introduce a simple yet effective guidance scheme as a loss function that leverages the consistency of stain translation with stain isolation. Qualitative and quantitative experiments show the ability of the proposed approach to improve translation between the two domains.

Increasing Confidence in Adversarial Robustness Evaluations

Hundreds of defenses have been proposed to make deep neural networks robust against minimal (adversarial) input perturbations. However, only a handful of these defenses held up their claims because correctly evaluating robustness is extremely challenging: Weak attacks often fail to find adversarial examples even if they unknowingly exist, thereby making a vulnerable network look robust. In this paper, we propose a test to identify weak attacks, and thus weak defense evaluations. Our test slightly modifies a neural network to guarantee the existence of an adversarial example for every sample. Consequentially, any correct attack must succeed in breaking this modified network. For eleven out of thirteen previously-published defenses, the original evaluation of the defense fails our test, while stronger attacks that break these defenses pass it. We hope that attack unit tests - such as ours - will be a major component in future robustness evaluations and increase confidence in an empirical field that is currently riddled with skepticism.

Fundamental Limits of Communication Efficiency for Model Aggregation in Distributed Learning: A Rate-Distortion Approach

One of the main focuses in distributed learning is communication efficiency, since model aggregation at each round of training can consist of millions to billions of parameters. Several model compression methods, such as gradient quantization and sparsification, have been proposed to improve the communication efficiency of model aggregation. However, the information-theoretic minimum communication cost for a given distortion of gradient estimators is still unknown. In this paper, we study the fundamental limit of communication cost of model aggregation in distributed learning from a rate-distortion perspective. By formulating the model aggregation as a vector Gaussian CEO problem, we derive the rate region bound and sum-rate-distortion function for the model aggregation problem, which reveals the minimum communication rate at a particular gradient distortion upper bound. We also analyze the communication cost at each iteration and total communication cost based on the sum-rate-distortion function with the gradient statistics of real-world datasets. It is found that the communication gain by exploiting the correlation between worker nodes is significant for SignSGD, and a high distortion of gradient estimator can achieve low total communication cost in gradient compression.

BAGEL: A Benchmark for Assessing Graph Neural Network Explanations

The problem of interpreting the decisions of machine learning is a well-researched and important. We are interested in a specific type of machine learning model that deals with graph data called graph neural networks. Evaluating interpretability approaches for graph neural networks (GNN) specifically are known to be challenging due to the lack of a commonly accepted benchmark. Given a GNN model, several interpretability approaches exist to explain GNN models with diverse (sometimes conflicting) evaluation methodologies. In this paper, we propose a benchmark for evaluating the explainability approaches for GNNs called Bagel. In Bagel, we firstly propose four diverse GNN explanation evaluation regimes -- 1) faithfulness, 2) sparsity, 3) correctness. and 4) plausibility. We reconcile multiple evaluation metrics in the existing literature and cover diverse notions for a holistic evaluation. Our graph datasets range from citation networks, document graphs, to graphs from molecules and proteins. We conduct an extensive empirical study on four GNN models and nine post-hoc explanation approaches for node and graph classification tasks. We open both the benchmarks and reference implementations and make them available at https://github.com/Mandeep-Rathee/Bagel-benchmark.

Improving Disease Classification Performance and Explainability of Deep Learning Models in Radiology with Heatmap Generators

As deep learning is widely used in the radiology field, the explainability of such models is increasingly becoming essential to gain clinicians' trust when using the models for diagnosis. In this research, three experiment sets were conducted with a U-Net architecture to improve the classification performance while enhancing the heatmaps corresponding to the model's focus through incorporating heatmap generators during training. All of the experiments used the dataset that contained chest radiographs, associated labels from one of the three conditions ("normal", "congestive heart failure (CHF)", and "pneumonia"), and numerical information regarding a radiologist's eye-gaze coordinates on the images. The paper (A. Karargyris and Moradi, 2021) that introduced this dataset developed a U-Net model, which was treated as the baseline model for this research, to show how the eye-gaze data can be used in multi-modal training for explainability improvement. To compare the classification performances, the 95% confidence intervals (CI) of the area under the receiver operating characteristic curve (AUC) were measured. The best method achieved an AUC of 0.913 (CI: 0.860-0.966). The greatest improvements were for the "pneumonia" and "CHF" classes, which the baseline model struggled most to classify, resulting in AUCs of 0.859 (CI: 0.732-0.957) and 0.962 (CI: 0.933-0.989), respectively. The proposed method's decoder was also able to produce probability masks that highlight the determining image parts in model classifications, similarly as the radiologist's eye-gaze data. Hence, this work showed that incorporating heatmap generators and eye-gaze information into training can simultaneously improve disease classification and provide explainable visuals that align well with how the radiologist viewed the chest radiographs when making diagnosis.

Towards a Grounded Theory of Causation for Embodied AI

There exist well-developed frameworks for causal modelling, but these require rather a lot of human domain expertise to define causal variables and perform interventions. In order to enable autonomous agents to learn abstract causal models through interactive experience, the existing theoretical foundations need to be extended and clarified. Existing frameworks give no guidance regarding variable choice / representation, and more importantly, give no indication as to which behaviour policies or physical transformations of state space shall count as interventions. The framework sketched in this paper describes actions as transformations of state space, for instance induced by an agent running a policy. This makes it possible to describe in a uniform way both transformations of the micro-state space and abstract models thereof, and say when the latter is veridical / grounded / natural. We then introduce (causal) variables, define a mechanism as an invariant predictor, and say when an action can be viewed as a ``surgical intervention'', thus bringing the objective of causal representation & intervention skill learning into clearer focus.

Smart Application for Fall Detection Using Wearable ECG & Accelerometer Sensors

Timely and reliable detection of falls is a large and rapidly growing field of research due to the medical and financial demand of caring for a constantly growing elderly population. Within the past 2 decades, the availability of high-quality hardware (high-quality sensors and AI microchips) and software (machine learning algorithms) technologies has served as a catalyst for this research by giving developers the capabilities to develop such systems. This study developed multiple application components in order to investigate the development challenges and choices for fall detection systems, and provide materials for future research. The smart application developed using this methodology was validated by the results from fall detection modelling experiments and model mobile deployment. The best performing model overall was the ResNet152 on a standardised, and shuffled dataset with a 2s window size which achieved 92.8% AUC, 7.28% sensitivity, and 98.33% specificity. Given these results it is evident that accelerometer and ECG sensors are beneficial for fall detection, and allow for the discrimination between falls and other activities. This study leaves a significant amount of room for improvement due to weaknesses identified in the resultant dataset. These improvements include using a labelling protocol for the critical phase of a fall, increasing the number of dataset samples, improving the test subject representation, and experimenting with frequency domain preprocessing.

Information Entropy Initialized Concrete Autoencoder for Optimal Sensor Placement and Reconstruction of Geophysical Fields

We propose a new approach to the optimal placement of sensors for the problem of reconstructing geophysical fields from sparse measurements. Our method consists of two stages. In the first stage, we estimate the variability of the physical field as a function of spatial coordinates by approximating its information entropy through the Conditional PixelCNN network. To calculate the entropy, a new ordering of a two-dimensional data array (spiral ordering) is proposed, which makes it possible to obtain the entropy of a physical field simultaneously for several spatial scales. In the second stage, the entropy of the physical field is used to initialize the distribution of optimal sensor locations. This distribution is further optimized with the Concrete Autoencoder architecture with the straight-through gradient estimator and adversarial loss to simultaneously minimize the number of sensors and maximize reconstruction accuracy. Our method scales linearly with data size, unlike commonly used Principal Component Analysis. We demonstrate our method on the two examples: (a) temperature and (b) salinity fields around the Barents Sea and the Svalbard group of islands. For these examples, we compute the reconstruction error of our method and a few baselines. We test our approach against two baselines (1) PCA with QR factorization and (2) climatology. We find out that the obtained optimal sensor locations have clear physical interpretation and correspond to the boundaries between sea currents.

Dynamic Memory for Interpretable Sequential Optimisation

Real-world applications of reinforcement learning for recommendation and experimentation faces a practical challenge: the relative reward of different bandit arms can evolve over the lifetime of the learning agent. To deal with these non-stationary cases, the agent must forget some historical knowledge, as it may no longer be relevant to minimise regret. We present a solution to handling non-stationarity that is suitable for deployment at scale, to provide business operators with automated adaptive optimisation. Our solution aims to provide interpretable learning that can be trusted by humans, whilst responding to non-stationarity to minimise regret. To this end, we develop an adaptive Bayesian learning agent that employs a novel form of dynamic memory. It enables interpretability through statistical hypothesis testing, by targeting a set point of statistical power when comparing rewards and adjusting its memory dynamically to achieve this power. By design, the agent is agnostic to different kinds of non-stationarity. Using numerical simulations, we compare its performance against an existing proposal and show that, under multiple non-stationary scenarios, our agent correctly adapts to real changes in the true rewards. In all bandit solutions, there is an explicit trade-off between learning and achieving maximal performance. Our solution sits on a different point on this trade-off when compared to another similarly robust approach: we prioritise interpretability, which relies on more learning, at the cost of some regret. We describe the architecture of a large-scale deployment of automatic optimisation-as-a-service where our agent achieves interpretability whilst adapting to changing circumstances.

RAW-GNN: RAndom Walk Aggregation based Graph Neural Network

Graph-Convolution-based methods have been successfully applied to representation learning on homophily graphs where nodes with the same label or similar attributes tend to connect with one another. Due to the homophily assumption of Graph Convolutional Networks (GCNs) that these methods use, they are not suitable for heterophily graphs where nodes with different labels or dissimilar attributes tend to be adjacent. Several methods have attempted to address this heterophily problem, but they do not change the fundamental aggregation mechanism of GCNs because they rely on summation operators to aggregate information from neighboring nodes, which is implicitly subject to the homophily assumption. Here, we introduce a novel aggregation mechanism and develop a RAndom Walk Aggregation-based Graph Neural Network (called RAW-GNN) method. The proposed approach integrates the random walk strategy with graph neural networks. The new method utilizes breadth-first random walk search to capture homophily information and depth-first search to collect heterophily information. It replaces the conventional neighborhoods with path-based neighborhoods and introduces a new path-based aggregator based on Recurrent Neural Networks. These designs make RAW-GNN suitable for both homophily and heterophily graphs. Extensive experimental results showed that the new method achieved state-of-the-art performance on a variety of homophily and heterophily graphs.

Robustifying Vision Transformer without Retraining from Scratch by Test-Time Class-Conditional Feature Alignment

Vision Transformer (ViT) is becoming more popular in image processing. Specifically, we investigate the effectiveness of test-time adaptation (TTA) on ViT, a technique that has emerged to correct its prediction during test-time by itself. First, we benchmark various test-time adaptation approaches on ViT-B16 and ViT-L16. It is shown that the TTA is effective on ViT and the prior-convention (sensibly selecting modulation parameters) is not necessary when using proper loss function. Based on the observation, we propose a new test-time adaptation method called class-conditional feature alignment (CFA), which minimizes both the class-conditional distribution differences and the whole distribution differences of the hidden representation between the source and target in an online manner. Experiments of image classification tasks on common corruption (CIFAR-10-C, CIFAR-100-C, and ImageNet-C) and domain adaptation (digits datasets and ImageNet-Sketch) show that CFA stably outperforms the existing baselines on various datasets. We also verify that CFA is model agnostic by experimenting on ResNet, MLP-Mixer, and several ViT variants (ViT-AugReg, DeiT, and BeiT). Using BeiT backbone, CFA achieves 19.8% top-1 error rate on ImageNet-C, outperforming the existing test-time adaptation baseline 44.0%. This is a state-of-the-art result among TTA methods that do not need to alter training phase.