Datasets:

CShorten

/

Last-Week-on-ML-ArXiv

like 0

Query-Efficient Adversarial Attack Based on Latin Hypercube Sampling

In order to be applicable in real-world scenario, Boundary Attacks (BAs) were proposed and ensured one hundred percent attack success rate with only decision information. However, existing BA methods craft adversarial examples by leveraging a simple random sampling (SRS) to estimate the gradient, consuming a large number of model queries. To overcome the drawback of SRS, this paper proposes a Latin Hypercube Sampling based Boundary Attack (LHS-BA) to save query budget. Compared with SRS, LHS has better uniformity under the same limited number of random samples. Therefore, the average on these random samples is closer to the true gradient than that estimated by SRS. Various experiments are conducted on benchmark datasets including MNIST, CIFAR, and ImageNet-1K. Experimental results demonstrate the superiority of the proposed LHS-BA over the state-of-the-art BA methods in terms of query efficiency. The source codes are publicly available at https://github.com/GZHU-DVL/LHS-BA.

Defending against the Label-flipping Attack in Federated Learning

Federated learning (FL) provides autonomy and privacy by design to participating peers, who cooperatively build a machine learning (ML) model while keeping their private data in their devices. However, that same autonomy opens the door for malicious peers to poison the model by conducting either untargeted or targeted poisoning attacks. The label-flipping (LF) attack is a targeted poisoning attack where the attackers poison their training data by flipping the labels of some examples from one class (i.e., the source class) to another (i.e., the target class). Unfortunately, this attack is easy to perform and hard to detect and it negatively impacts on the performance of the global model. Existing defenses against LF are limited by assumptions on the distribution of the peers' data and/or do not perform well with high-dimensional models. In this paper, we deeply investigate the LF attack behavior and find that the contradicting objectives of attackers and honest peers on the source class examples are reflected in the parameter gradients corresponding to the neurons of the source and target classes in the output layer, making those gradients good discriminative features for the attack detection. Accordingly, we propose a novel defense that first dynamically extracts those gradients from the peers' local updates, and then clusters the extracted gradients, analyzes the resulting clusters and filters out potential bad updates before model aggregation. Extensive empirical analysis on three data sets shows the proposed defense's effectiveness against the LF attack regardless of the data distribution or model dimensionality. Also, the proposed defense outperforms several state-of-the-art defenses by offering lower test error, higher overall accuracy, higher source class accuracy, lower attack success rate, and higher stability of the source class accuracy.

A Safe Semi-supervised Graph Convolution Network

In the semi-supervised learning field, Graph Convolution Network (GCN), as a variant model of GNN, has achieved promising results for non-Euclidean data by introducing convolution into GNN. However, GCN and its variant models fail to safely use the information of risk unlabeled data, which will degrade the performance of semi-supervised learning. Therefore, we propose a Safe GCN framework (Safe-GCN) to improve the learning performance. In the Safe-GCN, we design an iterative process to label the unlabeled data. In each iteration, a GCN and its supervised version(S-GCN) are learned to find the unlabeled data with high confidence. The high-confidence unlabeled data and their pseudo labels are then added to the label set. Finally, both added unlabeled data and labeled ones are used to train a S-GCN which can achieve the safe exploration of the risk unlabeled data and enable safe use of large numbers of unlabeled data. The performance of Safe-GCN is evaluated on three well-known citation network datasets and the obtained results demonstrate the effectiveness of the proposed framework over several graph-based semi-supervised learning methods.

Ask-AC: An Initiative Advisor-in-the-Loop Actor-Critic Framework

Despite the promising results achieved, state-of-the-art interactive reinforcement learning schemes rely on passively receiving supervision signals from advisor experts, in the form of either continuous monitoring or pre-defined rules, which inevitably result in a cumbersome and expensive learning process. In this paper, we introduce a novel initiative advisor-in-the-loop actor-critic framework, termed as Ask-AC, that replaces the unilateral advisor-guidance mechanism with a bidirectional learner-initiative one, and thereby enables a customized and efficacious message exchange between learner and advisor. At the heart of Ask-AC are two complementary components, namely action requester and adaptive state selector, that can be readily incorporated into various discrete actor-critic architectures. The former component allows the agent to initiatively seek advisor intervention in the presence of uncertain states, while the latter identifies the unstable states potentially missed by the former especially when environment changes, and then learns to promote the ask action on such states. Experimental results on both stationary and non-stationary environments and across different actor-critic backbones demonstrate that the proposed framework significantly improves the learning efficiency of the agent, and achieves the performances on par with those obtained by continuous advisor monitoring.

Entity Linking in Tabular Data Needs the Right Attention

Understanding the semantic meaning of tabular data requires Entity Linking (EL), in order to associate each cell value to a real-world entity in a Knowledge Base (KB). In this work, we focus on end-to-end solutions for EL on tabular data that do not rely on fact lookup in the target KB. Tabular data contains heterogeneous and sparse context, including column headers, cell values and table captions. We experiment with various models to generate a vector representation for each cell value to be linked. Our results show that it is critical to apply an attention mechanism as well as an attention mask, so that the model can only attend to the most relevant context and avoid information dilution. The most relevant context includes: same-row cells, same-column cells, headers and caption. Computational complexity, however, grows quadratically with the size of tabular data for such a complex model. We achieve constant memory usage by introducing a Tabular Entity Linking Lite model (TELL ) that generates vector representation for a cell based only on its value, the table headers and the table caption. TELL achieves 80.8% accuracy on Wikipedia tables, which is only 0.1% lower than the state-of-the-art model with quadratic memory usage.

Vector Quantisation for Robust Segmentation

The reliability of segmentation models in the medical domain depends on the model's robustness to perturbations in the input space. Robustness is a particular challenge in medical imaging exhibiting various sources of image noise, corruptions, and domain shifts. Obtaining robustness is often attempted via simulating heterogeneous environments, either heuristically in the form of data augmentation or by learning to generate specific perturbations in an adversarial manner. We propose and justify that learning a discrete representation in a low dimensional embedding space improves robustness of a segmentation model. This is achieved with a dictionary learning method called vector quantisation. We use a set of experiments designed to analyse robustness in both the latent and output space under domain shift and noise perturbations in the input space. We adapt the popular UNet architecture, inserting a quantisation block in the bottleneck. We demonstrate improved segmentation accuracy and better robustness on three segmentation tasks. Code is available at \url{https://github.com/AinkaranSanthi/Vector-Quantisation-for-Robust-Segmentation}

Explainability in Deep Reinforcement Learning, a Review into Current Methods and Applications

The use of Deep Reinforcement Learning (DRL) schemes has increased dramatically since their first introduction in 2015. Though uses in many different applications are being found they still have a problem with the lack of interpretability. This has bread a lack of understanding and trust in the use of DRL solutions from researchers and the general public. To solve this problem the field of explainable artificial intelligence (XAI) has emerged. This is a variety of different methods that look to open the DRL black boxes, they range from the use of interpretable symbolic decision trees to numerical methods like Shapley Values. This review looks at which methods are being used and what applications they are being used. This is done to identify which models are the best suited to each application or if a method is being underutilised.

Multi-Scored Sleep Databases: How to Exploit the Multiple-Labels in Automated Sleep Scoring

Study Objectives: Inter-scorer variability in scoring polysomnograms is a well-known problem. Most of the existing automated sleep scoring systems are trained using labels annotated by a single scorer, whose subjective evaluation is transferred to the model. When annotations from two or more scorers are available, the scoring models are usually trained on the scorer consensus. The averaged scorer's subjectivity is transferred into the model, losing information about the internal variability among different scorers. In this study, we aim to insert the multiple-knowledge of the different physicians into the training procedure.The goal is to optimize a model training, exploiting the full information that can be extracted from the consensus of a group of scorers. Methods: We train two lightweight deep learning based models on three different multi-scored databases. We exploit the label smoothing technique together with a soft-consensus (LSSC) distribution to insert the multiple-knowledge in the training procedure of the model. We introduce the averaged cosine similarity metric (ACS) to quantify the similarity between the hypnodensity-graph generated by the models with-LSSC and the hypnodensity-graph generated by the scorer consensus. Results: The performance of the models improves on all the databases when we train the models with our LSSC. We found an increase in ACS (up to 6.4%) between the hypnodensity-graph generated by the models trained with-LSSC and the hypnodensity-graph generated by the consensus. Conclusions: Our approach definitely enables a model to better adapt to the consensus of the group of scorers. Future work will focus on further investigations on different scoring architectures.

StyleFlow For Content-Fixed Image to Image Translation

Image-to-image (I2I) translation is a challenging topic in computer vision. We divide this problem into three tasks: strongly constrained translation, normally constrained translation, and weakly constrained translation. The constraint here indicates the extent to which the content or semantic information in the original image is preserved. Although previous approaches have achieved good performance in weakly constrained tasks, they failed to fully preserve the content in both strongly and normally constrained tasks, including photo-realism synthesis, style transfer, and colorization, etc. To achieve content-preserving transfer in strongly constrained and normally constrained tasks, we propose StyleFlow, a new I2I translation model that consists of normalizing flows and a novel Style-Aware Normalization (SAN) module. With the invertible network structure, StyleFlow first projects input images into deep feature space in the forward pass, while the backward pass utilizes the SAN module to perform content-fixed feature transformation and then projects back to image space. Our model supports both image-guided translation and multi-modal synthesis. We evaluate our model in several I2I translation benchmarks, and the results show that the proposed model has advantages over previous methods in both strongly constrained and normally constrained tasks.

A Deep Learning Approach for the solution of Probability Density Evolution of Stochastic Systems

Derivation of the probability density evolution provides invaluable insight into the behavior of many stochastic systems and their performance. However, for most real-time applica-tions, numerical determination of the probability density evolution is a formidable task. The latter is due to the required temporal and spatial discretization schemes that render most computational solutions prohibitive and impractical. In this respect, the development of an efficient computational surrogate model is of paramount importance. Recent studies on the physics-constrained networks show that a suitable surrogate can be achieved by encoding the physical insight into a deep neural network. To this aim, the present work introduces DeepPDEM which utilizes the concept of physics-informed networks to solve the evolution of the probability density via proposing a deep learning method. DeepPDEM learns the General Density Evolution Equation (GDEE) of stochastic structures. This approach paves the way for a mesh-free learning method that can solve the density evolution problem with-out prior simulation data. Moreover, it can also serve as an efficient surrogate for the solu-tion at any other spatiotemporal points within optimization schemes or real-time applica-tions. To demonstrate the potential applicability of the proposed framework, two network architectures with different activation functions as well as two optimizers are investigated. Numerical implementation on three different problems verifies the accuracy and efficacy of the proposed method.

"Even if ..." -- Diverse Semifactual Explanations of Reject

Machine learning based decision making systems applied in safety critical areas require reliable high certainty predictions. For this purpose, the system can be extended by an reject option which allows the system to reject inputs where only a prediction with an unacceptably low certainty would be possible. While being able to reject uncertain samples is important, it is also of importance to be able to explain why a particular sample was rejected. With the ongoing rise of eXplainable AI (XAI), a lot of explanation methodologies for machine learning based systems have been developed -- explaining reject options, however, is still a novel field where only very little prior work exists. In this work, we propose to explain rejects by semifactual explanations, an instance of example-based explanation methods, which them self have not been widely considered in the XAI community yet. We propose a conceptual modeling of semifactual explanations for arbitrary reject options and empirically evaluate a specific implementation on a conformal prediction based reject option.

The Deep Ritz Method for Parametric $p$-Dirichlet Problems

We establish error estimates for the approximation of parametric $p$-Dirichlet problems deploying the Deep Ritz Method. Parametric dependencies include, e.g., varying geometries and exponents $p\in (1,\infty)$. Combining the derived error estimates with quantitative approximation theorems yields error decay rates and establishes that the Deep Ritz Method retains the favorable approximation capabilities of neural networks in the approximation of high dimensional functions which makes the method attractive for parametric problems. Finally, we present numerical examples to illustrate potential applications.

ICE-NODE: Integration of Clinical Embeddings with Neural Ordinary Differential Equations

Early diagnosis of disease can result in improved health outcomes, such as higher survival rates and lower treatment costs. With the massive amount of information in electronic health records (EHRs), there is great potential to use machine learning (ML) methods to model disease progression aimed at early prediction of disease onset and other outcomes. In this work, we employ recent innovations in neural ODEs to harness the full temporal information of EHRs. We propose ICE-NODE (Integration of Clinical Embeddings with Neural Ordinary Differential Equations), an architecture that temporally integrates embeddings of clinical codes and neural ODEs to learn and predict patient trajectories in EHRs. We apply our method to the publicly available MIMIC-III and MIMIC-IV datasets, reporting improved prediction results compared to state-of-the-art methods, specifically for clinical codes that are not frequently observed in EHRs. We also show that ICE-NODE is more competent at predicting certain medical conditions, like acute renal failure and pulmonary heart disease, and is also able to produce patient risk trajectories over time that can be exploited for further predictions.

Bayesian approaches for Quantifying Clinicians' Variability in Medical Image Quantification

Medical imaging, including MRI, CT, and Ultrasound, plays a vital role in clinical decisions. Accurate segmentation is essential to measure the structure of interest from the image. However, manual segmentation is highly operator-dependent, which leads to high inter and intra-variability of quantitative measurements. In this paper, we explore the feasibility that Bayesian predictive distribution parameterized by deep neural networks can capture the clinicians' inter-intra variability. By exploring and analyzing recently emerged approximate inference schemes, we evaluate whether approximate Bayesian deep learning with the posterior over segmentations can learn inter-intra rater variability both in segmentation and clinical measurements. The experiments are performed with two different imaging modalities: MRI and ultrasound. We empirically demonstrated that Bayesian predictive distribution parameterized by deep neural networks could approximate the clinicians' inter-intra variability. We show a new perspective in analyzing medical images quantitatively by providing clinical measurement uncertainty.

Meta-Learning a Real-Time Tabular AutoML Method For Small Data

We present TabPFN, an AutoML method that is competitive with the state of the art on small tabular datasets while being over 1,000$\times$ faster. Our method is very simple: it is fully entailed in the weights of a single neural network, and a single forward pass directly yields predictions for a new dataset. Our AutoML method is meta-learned using the Transformer-based Prior-Data Fitted Network (PFN) architecture and approximates Bayesian inference with a prior that is based on assumptions of simplicity and causal structures. The prior contains a large space of structural causal models and Bayesian neural networks with a bias for small architectures and thus low complexity. Furthermore, we extend the PFN approach to differentiably calibrate the prior's hyperparameters on real data. By doing so, we separate our abstract prior assumptions from their heuristic calibration on real data. Afterwards, the calibrated hyperparameters are fixed and TabPFN can be applied to any new tabular dataset at the push of a button. Finally, on 30 datasets from the OpenML-CC18 suite we show that our method outperforms boosted trees and performs on par with complex state-of-the-art AutoML systems with predictions produced in less than a second. We provide all our code and our final trained TabPFN in the supplementary materials.

PoF: Post-Training of Feature Extractor for Improving Generalization

It has been intensively investigated that the local shape, especially flatness, of the loss landscape near a minimum plays an important role for generalization of deep models. We developed a training algorithm called PoF: Post-Training of Feature Extractor that updates the feature extractor part of an already-trained deep model to search a flatter minimum. The characteristics are two-fold: 1) Feature extractor is trained under parameter perturbations in the higher-layer parameter space, based on observations that suggest flattening higher-layer parameter space, and 2) the perturbation range is determined in a data-driven manner aiming to reduce a part of test loss caused by the positive loss curvature. We provide a theoretical analysis that shows the proposed algorithm implicitly reduces the target Hessian components as well as the loss. Experimental results show that PoF improved model performance against baseline methods on both CIFAR-10 and CIFAR-100 datasets for only 10-epoch post-training, and on SVHN dataset for 50-epoch post-training. Source code is available at: \url{https://github.com/DensoITLab/PoF-v1

Efficient Representation Learning via Adaptive Context Pooling

Self-attention mechanisms model long-range context by using pairwise attention between all input tokens. In doing so, they assume a fixed attention granularity defined by the individual tokens (e.g., text characters or image pixels), which may not be optimal for modeling complex dependencies at higher levels. In this paper, we propose ContextPool to address this problem by adapting the attention granularity for each token. Inspired by the success of ConvNets that are combined with pooling to capture long-range dependencies, we learn to pool neighboring features for each token before computing attention in a given attention layer. The pooling weights and support size are adaptively determined, allowing the pooled features to encode meaningful context with varying scale. We show that ContextPool makes attention models more expressive, achieving strong performance often with fewer layers and thus significantly reduced cost. Experiments validate that our ContextPool module, when plugged into transformer models, matches or surpasses state-of-the-art performance using less compute on several language and image benchmarks, outperforms recent works with learned context sizes or sparse attention patterns, and is also applicable to ConvNets for efficient feature learning.

Randomized-to-Canonical Model Predictive Control for Real-world Visual Robotic Manipulation

Many works have recently explored Sim-to-real transferable visual model predictive control (MPC). However, such works are limited to one-shot transfer, where real-world data must be collected once to perform the sim-to-real transfer, which remains a significant human effort in transferring the models learned in simulations to new domains in the real world. To alleviate this problem, we first propose a novel model-learning framework called Kalman Randomized-to-Canonical Model (KRC-model). This framework is capable of extracting task-relevant intrinsic features and their dynamics from randomized images. We then propose Kalman Randomized-to-Canonical Model Predictive Control (KRC-MPC) as a zero-shot sim-to-real transferable visual MPC using KRC-model. The effectiveness of our method is evaluated through a valve rotation task by a robot hand in both simulation and the real world, and a block mating task in simulation. The experimental results show that KRC-MPC can be applied to various real domains and tasks in a zero-shot manner.

What Do Graph Convolutional Neural Networks Learn?

Graph neural networks (GNNs) have gained traction over the past few years for their superior performance in numerous machine learning tasks. Graph Convolutional Neural Networks (GCN) are a common variant of GNNs that are known to have high performance in semi-supervised node classification (SSNC), and work well under the assumption of homophily. Recent literature has highlighted that GCNs can achieve strong performance on heterophilous graphs under certain "special conditions". These arguments motivate us to understand why, and how, GCNs learn to perform SSNC. We find a positive correlation between similarity of latent node embeddings of nodes within a class and the performance of a GCN. Our investigation on underlying graph structures of a dataset finds that a GCN's SSNC performance is significantly influenced by the consistency and uniqueness in neighborhood structure of nodes within a class.

Ensemble feature selection with data-driven thresholding for Alzheimer's disease biomarker discovery

Healthcare datasets present many challenges to both machine learning and statistics as their data are typically heterogeneous, censored, high-dimensional and have missing information. Feature selection is often used to identify the important features but can produce unstable results when applied to high-dimensional data, selecting a different set of features on each iteration. The stability of feature selection can be improved with the use of feature selection ensembles, which aggregate the results of multiple base feature selectors. A threshold must be applied to the final aggregated feature set to separate the relevant features from the redundant ones. A fixed threshold, which is typically applied, offers no guarantee that the final set of selected features contains only relevant features. This work develops several data-driven thresholds to automatically identify the relevant features in an ensemble feature selector and evaluates their predictive accuracy and stability. To demonstrate the applicability of these methods to clinical data, they are applied to data from two real-world Alzheimer's disease (AD) studies. AD is a progressive neurodegenerative disease with no known cure, that begins at least 2-3 decades before overt symptoms appear, presenting an opportunity for researchers to identify early biomarkers that might identify patients at risk of developing AD. Features identified by applying these methods to both datasets reflect current findings in the AD literature.

Multimodal Frame-Scoring Transformer for Video Summarization

As the number of video content has mushroomed in recent years, automatic video summarization has come useful when we want to just peek at the content of the video. However, there are two underlying limitations in generic video summarization task. First, most previous approaches read in just visual features as input, leaving other modality features behind. Second, existing datasets for generic video summarization are relatively insufficient to train a caption generator and multimodal feature extractors. To address these two problems, this paper proposes the Multimodal Frame-Scoring Transformer (MFST) framework exploiting visual, text and audio features and scoring a video with respect to frames. Our MFST framework first extracts each modality features (visual-text-audio) using pretrained encoders. Then, MFST trains the multimodal frame-scoring transformer that uses video-text-audio representations as inputs and predicts frame-level scores. Our extensive experiments with previous models and ablation studies on TVSum and SumMe datasets demonstrate the effectiveness and superiority of our proposed method.

Deriving Surface Resistivity from Polarimetric SAR Data Using Dual-Input UNet

Traditional survey methods for finding surface resistivity are time-consuming and labor intensive. Very few studies have focused on finding the resistivity/conductivity using remote sensing data and deep learning techniques. In this line of work, we assessed the correlation between surface resistivity and Synthetic Aperture Radar (SAR) by applying various deep learning methods and tested our hypothesis in the Coso Geothermal Area, USA. For detecting the resistivity, L-band full polarimetric SAR data acquired by UAVSAR were used, and MT (Magnetotellurics) inverted resistivity data of the area were used as the ground truth. We conducted experiments to compare various deep learning architectures and suggest the use of Dual Input UNet (DI-UNet) architecture. DI-UNet uses a deep learning architecture to predict the resistivity using full polarimetric SAR data by promising a quick survey addition to the traditional method. Our proposed approach accomplished improved outcomes for the mapping of MT resistivity from SAR data.

Sedentary Behavior Estimation with Hip-worn Accelerometer Data: Segmentation, Classification and Thresholding

Cohort studies are increasingly using accelerometers for physical activity and sedentary behavior estimation. These devices tend to be less error-prone than self-report, can capture activity throughout the day, and are economical. However, previous methods for estimating sedentary behavior based on hip-worn data are often invalid or suboptimal under free-living situations and subject-to-subject variation. In this paper, we propose a local Markov switching model that takes this situation into account, and introduce a general procedure for posture classification and sedentary behavior analysis that fits the model naturally. Our method features changepoint detection methods in time series and also a two stage classification step that labels data into 3 classes(sitting, standing, stepping). Through a rigorous training-testing paradigm, we showed that our approach achieves > 80% accuracy. In addition, our method is robust and easy to interpret.

GSMFlow: Generation Shifts Mitigating Flow for Generalized Zero-Shot Learning

Generalized Zero-Shot Learning (GZSL) aims to recognize images from both the seen and unseen classes by transferring semantic knowledge from seen to unseen classes. It is a promising solution to take the advantage of generative models to hallucinate realistic unseen samples based on the knowledge learned from the seen classes. However, due to the generation shifts, the synthesized samples by most existing methods may drift from the real distribution of the unseen data. To address this issue, we propose a novel flow-based generative framework that consists of multiple conditional affine coupling layers for learning unseen data generation. Specifically, we discover and address three potential problems that trigger the generation shifts, i.e., semantic inconsistency, variance collapse, and structure disorder. First, to enhance the reflection of the semantic information in the generated samples, we explicitly embed the semantic information into the transformation in each conditional affine coupling layer. Second, to recover the intrinsic variance of the real unseen features, we introduce a boundary sample mining strategy with entropy maximization to discover more difficult visual variants of semantic prototypes and hereby adjust the decision boundary of the classifiers. Third, a relative positioning strategy is proposed to revise the attribute embeddings, guiding them to fully preserve the inter-class geometric structure and further avoid structure disorder in the semantic space. Extensive experimental results on four GZSL benchmark datasets demonstrate that GSMFlow achieves the state-of-the-art performance on GZSL.

Task-agnostic Defense against Adversarial Patch Attacks

Adversarial patch attacks mislead neural networks by injecting adversarial pixels within a designated local region. Patch attacks can be highly effective in a variety of tasks and physically realizable via attachment (e.g. a sticker) to the real-world objects. Despite the diversity in attack patterns, adversarial patches tend to be highly textured and different in appearance from natural images. We exploit this property and present PatchZero, a task-agnostic defense against white-box adversarial patches. Specifically, our defense detects the adversarial pixels and "zeros out" the patch region by repainting with mean pixel values. We formulate the patch detection problem as a semantic segmentation task such that our model can generalize to patches of any size and shape. We further design a two-stage adversarial training scheme to defend against the stronger adaptive attacks. We thoroughly evaluate PatchZero on the image classification (ImageNet, RESISC45), object detection (PASCAL VOC), and video classification (UCF101) datasets. Our method achieves SOTA robust accuracy without any degradation in the benign performance.

Features Based Adaptive Augmentation for Graph Contrastive Learning

Self-Supervised learning aims to eliminate the need for expensive annotation in graph representation learning, where graph contrastive learning (GCL) is trained with the self-supervision signals containing data-data pairs. These data-data pairs are generated with augmentation employing stochastic functions on the original graph. We argue that some features can be more critical than others depending on the downstream task, and applying stochastic function uniformly, will vandalize the influential features, leading to diminished accuracy. To fix this issue, we introduce a Feature Based Adaptive Augmentation (FebAA) approach, which identifies and preserves potentially influential features and corrupts the remaining ones. We implement FebAA as plug and play layer and use it with state-of-the-art Deep Graph Contrastive Learning (GRACE) and Bootstrapped Graph Latents (BGRL). We successfully improved the accuracy of GRACE and BGRL on eight graph representation learning's benchmark datasets.

Improved Global Guarantees for the Nonconvex Burer--Monteiro Factorization via Rank Overparameterization

We consider minimizing a twice-differentiable, $L$-smooth, and $\mu$-strongly convex objective $\phi$ over an $n\times n$ positive semidefinite matrix $M\succeq0$, under the assumption that the minimizer $M^{\star}$ has low rank $r^{\star}\ll n$. Following the Burer--Monteiro approach, we instead minimize the nonconvex objective $f(X)=\phi(XX^{T})$ over a factor matrix $X$ of size $n\times r$. This substantially reduces the number of variables from $O(n^{2})$ to as few as $O(n)$ and also enforces positive semidefiniteness for free, but at the cost of giving up the convexity of the original problem. In this paper, we prove that if the search rank $r\ge r^{\star}$ is overparameterized by a constant factor with respect to the true rank $r^{\star}$, namely as in $r>\frac{1}{4}(L/\mu-1)^{2}r^{\star}$, then despite nonconvexity, local optimization is guaranteed to globally converge from any initial point to the global optimum. This significantly improves upon a previous rank overparameterization threshold of $r\ge n$, which is known to be sharp if $\phi$ is allowed to be nonsmooth and/or non-strongly convex, but would increase the number of variables back up to $O(n^{2})$. Conversely, without rank overparameterization, we prove that such a global guarantee is possible if and only if $\phi$ is almost perfectly conditioned, with a condition number of $L/\mu<3$. Therefore, we conclude that a small amount of overparameterization can lead to large improvements in theoretical guarantees for the nonconvex Burer--Monteiro factorization.

A Unified Meta-Learning Framework for Dynamic Transfer Learning

Transfer learning refers to the transfer of knowledge or information from a relevant source task to a target task. However, most existing works assume both tasks are sampled from a stationary task distribution, thereby leading to the sub-optimal performance for dynamic tasks drawn from a non-stationary task distribution in real scenarios. To bridge this gap, in this paper, we study a more realistic and challenging transfer learning setting with dynamic tasks, i.e., source and target tasks are continuously evolving over time. We theoretically show that the expected error on the dynamic target task can be tightly bounded in terms of source knowledge and consecutive distribution discrepancy across tasks. This result motivates us to propose a generic meta-learning framework L2E for modeling the knowledge transferability on dynamic tasks. It is centered around a task-guided meta-learning problem with a group of meta-pairs of tasks, based on which we are able to learn the prior model initialization for fast adaptation on the newest target task. L2E enjoys the following properties: (1) effective knowledge transferability across dynamic tasks; (2) fast adaptation to the new target task; (3) mitigation of catastrophic forgetting on historical target tasks; and (4) flexibility in incorporating any existing static transfer learning algorithms. Extensive experiments on various image data sets demonstrate the effectiveness of the proposed L2E framework.

On A Mallows-type Model For (Ranked) Choices

In a preference learning setting, every participant chooses an ordered list of $k$ most preferred items among a displayed set of candidates. (The set can be different for every participant.) We identify a distance-based ranking model for the population's preferences and their (ranked) choice behavior. The ranking model resembles the Mallows model but uses a new distance function called Reverse Major Index (RMJ). We find that despite the need to sum over all permutations, the RMJ-based ranking distribution aggregates into (ranked) choice probabilities with simple closed-form expression. We develop effective methods to estimate the model parameters and showcase their generalization power using real data, especially when there is a limited variety of display sets.

CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning

Program synthesis or code generation aims to generate a program that satisfies a problem specification. Recent approaches using large-scale pretrained language models (LMs) have shown promising results, yet they have some critical limitations. In particular, they often follow a standard supervised fine-tuning procedure to train a code generation model only from the pairs of natural-language problem descriptions and ground-truth programs. Such paradigm largely ignores some important but potentially useful signals in the problem specification such as unit tests, which thus often results in poor performance when solving complex unseen coding tasks. To address the limitations, we propose "CodeRL", a new framework for program synthesis tasks through pretrained LMs and deep reinforcement learning (RL). Specifically, during training, we treat the code-generating LM as an actor network, and introduce a critic network that is trained to predict the functional correctness of generated programs and provide dense feedback signals to the actor. During inference, we introduce a new generation procedure with a critical sampling strategy that allows a model to automatically regenerate programs based on feedback from example unit tests and critic scores. For the model backbones, we extended the encoder-decoder architecture of CodeT5 with enhanced learning objectives, larger model sizes, and better pretraining data. Our method not only achieves new SOTA results on the challenging APPS benchmark, but also shows strong zero-shot transfer capability with new SOTA results on the simpler MBPP benchmark.

Approximating Discontinuous Nash Equilibrial Values of Two-Player General-Sum Differential Games

Finding Nash equilibrial policies for two-player differential games requires solving Hamilton-Jacobi-Isaacs PDEs. Recent studies achieved success in circumventing the curse of dimensionality in solving such PDEs with underlying applications to human-robot interactions (HRI), by adopting self-supervised (physics-informed) neural networks as universal value approximators. This paper extends from previous SOTA on zero-sum games with continuous values to general-sum games with discontinuous values, where the discontinuity is caused by that of the players' losses. We show that due to its lack of convergence proof and generalization analysis on discontinuous losses, the existing self-supervised learning technique fails to generalize and raises safety concerns in an autonomous driving application. Our solution is to first pre-train the value network on supervised Nash equilibria, and then refine it by minimizing a loss that combines the supervised data with the PDE and boundary conditions. Importantly, the demonstrated advantage of the proposed learning method against purely supervised and self-supervised approaches requires careful choice of the neural activation function: Among $\texttt{relu}$, $\texttt{sin}$, and $\texttt{tanh}$, we show that $\texttt{tanh}$ is the only choice that achieves optimal generalization and safety performance. Our conjecture is that $\texttt{tanh}$ (similar to $\texttt{sin}$) allows continuity of value and its gradient, which is sufficient for the convergence of learning, and at the same time is expressive enough (similar to $\texttt{relu}$) at approximating discontinuous value landscapes. Lastly, we apply our method to approximating control policies for an incomplete-information interaction and demonstrate its contribution to safe interactions.

A Generative Framework for Personalized Learning and Estimation: Theory, Algorithms, and Privacy

A distinguishing characteristic of federated learning is that the (local) client data could have statistical heterogeneity. This heterogeneity has motivated the design of personalized learning, where individual (personalized) models are trained, through collaboration. There have been various personalization methods proposed in literature, with seemingly very different forms and methods ranging from use of a single global model for local regularization and model interpolation, to use of multiple global models for personalized clustering, etc. In this work, we begin with a generative framework that could potentially unify several different algorithms as well as suggest new algorithms. We apply our generative framework to personalized estimation, and connect it to the classical empirical Bayes' methodology. We develop private personalized estimation under this framework. We then use our generative framework for learning, which unifies several known personalized FL algorithms and also suggests new ones; we propose and study a new algorithm AdaPeD based on a Knowledge Distillation, which numerically outperforms several known algorithms. We also develop privacy for personalized learning methods with guarantees for user-level privacy and composition. We numerically evaluate the performance as well as the privacy for both the estimation and learning problems, demonstrating the advantages of our proposed methods.

opPINN: Physics-Informed Neural Network with operator learning to approximate solutions to the Fokker-Planck-Landau equation

We propose a hybrid framework opPINN: physics-informed neural network (PINN) with operator learning for approximating the solution to the Fokker-Planck-Landau (FPL) equation. The opPINN framework is divided into two steps: Step 1 and Step 2. After the operator surrogate models are trained during Step 1, PINN can effectively approximate the solution to the FPL equation during Step 2 by using the pre-trained surrogate models. The operator surrogate models greatly reduce the computational cost and boost PINN by approximating the complex Landau collision integral in the FPL equation. The operator surrogate models can also be combined with the traditional numerical schemes. It provides a high efficiency in computational time when the number of velocity modes becomes larger. Using the opPINN framework, we provide the neural network solutions for the FPL equation under the various types of initial conditions, and interaction models in two and three dimensions. Furthermore, based on the theoretical properties of the FPL equation, we show that the approximated neural network solution converges to the a priori classical solution of the FPL equation as the pre-defined loss function is reduced.

TT-PINN: A Tensor-Compressed Neural PDE Solver for Edge Computing

Physics-informed neural networks (PINNs) have been increasingly employed due to their capability of modeling complex physics systems. To achieve better expressiveness, increasingly larger network sizes are required in many problems. This has caused challenges when we need to train PINNs on edge devices with limited memory, computing and energy resources. To enable training PINNs on edge devices, this paper proposes an end-to-end compressed PINN based on Tensor-Train decomposition. In solving a Helmholtz equation, our proposed model significantly outperforms the original PINNs with few parameters and achieves satisfactory prediction with up to 15$\times$ overall parameter reduction.

Federated Split GANs

Mobile devices and the immense amount and variety of data they generate are key enablers of machine learning (ML)-based applications. Traditional ML techniques have shifted toward new paradigms such as federated (FL) and split learning (SL) to improve the protection of user's data privacy. However, these paradigms often rely on server(s) located in the edge or cloud to train computationally-heavy parts of a ML model to avoid draining the limited resource on client devices, resulting in exposing device data to such third parties. This work proposes an alternative approach to train computationally-heavy ML models in user's devices themselves, where corresponding device data resides. Specifically, we focus on GANs (generative adversarial networks) and leverage their inherent privacy-preserving attribute. We train the discriminative part of a GAN with raw data on user's devices, whereas the generative model is trained remotely (e.g., server) for which there is no need to access sensor true data. Moreover, our approach ensures that the computational load of training the discriminative model is shared among user's devices-proportional to their computation capabilities-by means of SL. We implement our proposed collaborative training scheme of a computationally-heavy GAN model in real resource-constrained devices. The results show that our system preserves data privacy, keeps a short training time, and yields same accuracy of model training in unconstrained devices (e.g., cloud). Our code can be found on https://github.com/YukariSonz/FSL-GAN

Discrete Tree Flows via Tree-Structured Permutations

While normalizing flows for continuous data have been extensively researched, flows for discrete data have only recently been explored. These prior models, however, suffer from limitations that are distinct from those of continuous flows. Most notably, discrete flow-based models cannot be straightforwardly optimized with conventional deep learning methods because gradients of discrete functions are undefined or zero. Previous works approximate pseudo-gradients of the discrete functions but do not solve the problem on a fundamental level. In addition to that, backpropagation can be computationally burdensome compared to alternative discrete algorithms such as decision tree algorithms. Our approach seeks to reduce computational burden and remove the need for pseudo-gradients by developing a discrete flow based on decision trees -- building upon the success of efficient tree-based methods for classification and regression for discrete data. We first define a tree-structured permutation (TSP) that compactly encodes a permutation of discrete data where the inverse is easy to compute; thus, we can efficiently compute the density value and sample new data. We then propose a decision tree algorithm to build TSPs that learns the tree structure and permutations at each node via novel criteria. We empirically demonstrate the feasibility of our method on multiple datasets.

Anomaly-aware multiple instance learning for rare anemia disorder classification

Deep learning-based classification of rare anemia disorders is challenged by the lack of training data and instance-level annotations. Multiple Instance Learning (MIL) has shown to be an effective solution, yet it suffers from low accuracy and limited explainability. Although the inclusion of attention mechanisms has addressed these issues, their effectiveness highly depends on the amount and diversity of cells in the training samples. Consequently, the poor machine learning performance on rare anemia disorder classification from blood samples remains unresolved. In this paper, we propose an interpretable pooling method for MIL to address these limitations. By benefiting from instance-level information of negative bags (i.e., homogeneous benign cells from healthy individuals), our approach increases the contribution of anomalous instances. We show that our strategy outperforms standard MIL classification algorithms and provides a meaningful explanation behind its decisions. Moreover, it can denote anomalous instances of rare blood diseases that are not seen during the training phase.

Machine Learning in Access Control: A Taxonomy and Survey

An increasing body of work has recognized the importance of exploiting machine learning (ML) advancements to address the need for efficient automation in extracting access control attributes, policy mining, policy verification, access decisions, etc. In this work, we survey and summarize various ML approaches to solve different access control problems. We propose a novel taxonomy of the ML model's application in the access control domain. We highlight current limitations and open challenges such as lack of public real-world datasets, administration of ML-based access control systems, understanding a black-box ML model's decision, etc., and enumerate future research directions.

AutoSpeed: A Linked Autoencoder Approach for Pulse-Echo Speed-of-Sound Imaging for Medical Ultrasound

Quantitative ultrasound, e.g., speed-of-sound (SoS) in tissues, provides information about tissue properties that have diagnostic value. Recent studies showed the possibility of extracting SoS information from pulse-echo ultrasound raw data (a.k.a. RF data) using deep neural networks that are fully trained on simulated data. These methods take sensor domain data, i.e., RF data, as input and train a network in an end-to-end fashion to learn the implicit mapping between the RF data domain and SoS domain. However, such networks are prone to overfitting to simulated data which results in poor performance and instability when tested on measured data. We propose a novel method for SoS mapping employing learned representations from two linked autoencoders. We test our approach on simulated and measured data acquired from human breast mimicking phantoms. We show that SoS mapping is possible using linked autoencoders. The proposed method has a Mean Absolute Percentage Error (MAPE) of 2.39% on the simulated data. On the measured data, the predictions of the proposed method are close to the expected values with MAPE of 1.1%. Compared to an end-to-end trained network, the proposed method shows higher stability and reproducibility.

How Much More Data Do I Need? Estimating Requirements for Downstream Tasks

Given a small training data set and a learning algorithm, how much more data is necessary to reach a target validation or test performance? This question is of critical importance in applications such as autonomous driving or medical imaging where collecting data is expensive and time-consuming. Overestimating or underestimating data requirements incurs substantial costs that could be avoided with an adequate budget. Prior work on neural scaling laws suggest that the power-law function can fit the validation performance curve and extrapolate it to larger data set sizes. We find that this does not immediately translate to the more difficult downstream task of estimating the required data set size to meet a target performance. In this work, we consider a broad class of computer vision tasks and systematically investigate a family of functions that generalize the power-law function to allow for better estimation of data requirements. Finally, we show that incorporating a tuned correction factor and collecting over multiple rounds significantly improves the performance of the data estimators. Using our guidelines, practitioners can accurately estimate data requirements of machine learning systems to gain savings in both development time and data acquisition costs.

A Causal Approach for Business Optimization: Application on an Online Marketplace

A common sales strategy involves having account executives (AEs) actively reach out and contact potential customers. However, not all contact attempts have a positive effect: some attempts do not change customer decisions, while others might even interfere with the desired outcome. In this work we propose using causal inference to estimate the effect of contacting each potential customer and setting the contact policy accordingly. We demonstrate this approach on data from Worthy.com, an online jewelry marketplace. We examined the Worthy business process to identify relevant decisions and outcomes, and formalized assumptions on how they were made. Using causal tools, we selected a decision point where improving AE contact activity appeared to be promising. We then generated a personalized policy and recommended reaching out only to customers for whom it would be beneficial. Finally, we validated the results in an A\B test over a 3-month period, resulting in an increase in item delivery rate of the targeted population by 22% (p-value=0.026). This policy is now being used on an ongoing basis.

Do Not Take It for Granted: Comparing Open-Source Libraries for Software Development Effort Estimation

In the past two decades, several Machine Learning (ML) libraries have become freely available. Many studies have used such libraries to carry out empirical investigations on predictive Software Engineering (SE) tasks. However, the differences stemming from using one library over another have been overlooked, implicitly assuming that using any of these libraries would provide the user with the same or very similar results. This paper aims at raising awareness of the differences incurred when using different ML libraries for software development effort estimation (SEE), one of most widely studied SE prediction tasks. To this end, we investigate 4 deterministic machine learners as provided by 3 of the most popular ML open-source libraries written in different languages (namely, Scikit-Learn, Caret and Weka). We carry out a thorough empirical study comparing the performance of the machine learners on 5 SEE datasets in the two most common SEE scenarios (i.e., out-of-the-box-ml and tuned-ml) as well as an in-depth analysis of the documentation and code of their APIs. The results of our study reveal that the predictions provided by the 3 libraries differ in 95% of the cases on average across a total of 105 cases studied. These differences are significantly large in most cases and yield misestimations of up to approx. 3,000 hours per project. Moreover, our API analysis reveals that these libraries provide the user with different levels of control on the parameters one can manipulate, and a lack of clarity and consistency, overall, which might mislead users. Our findings highlight that the ML library is an important design choice for SEE studies, which can lead to a difference in performance. However, such a difference is under-documented. We conclude by highlighting open-challenges with suggestions for the developers of libraries as well as for the researchers and practitioners using them.

An adaptive music generation architecture for games based on the deep learning Transformer mode

This paper presents an architecture for generating music for video games based on the Transformer deep learning model. The system generates music in various layers, following the standard layering strategy currently used by composers designing video game music. The music is adaptive to the psychological context of the player, according to the arousal-valence model. Our motivation is to customize music according to the player's tastes, who can select his preferred style of music through a set of training examples of music. We discuss current limitations and prospects for the future, such as collaborative and interactive control of the musical components.

FACT: High-Dimensional Random Forests Inference

Random forests is one of the most widely used machine learning methods over the past decade thanks to its outstanding empirical performance. Yet, because of its black-box nature, the results by random forests can be hard to interpret in many big data applications. Quantifying the usefulness of individual features in random forests learning can greatly enhance its interpretability. Existing studies have shown that some popularly used feature importance measures for random forests suffer from the bias issue. In addition, there lack comprehensive size and power analyses for most of these existing methods. In this paper, we approach the problem via hypothesis testing, and suggest a framework of the self-normalized feature-residual correlation test (FACT) for evaluating the significance of a given feature in the random forests model with bias-resistance property, where our null hypothesis concerns whether the feature is conditionally independent of the response given all other features. Such an endeavor on random forests inference is empowered by some recent developments on high-dimensional random forests consistency. The vanilla version of our FACT test can suffer from the bias issue in the presence of feature dependency. We exploit the techniques of imbalancing and conditioning for bias correction. We further incorporate the ensemble idea into the FACT statistic through feature transformations for the enhanced power. Under a fairly general high-dimensional nonparametric model setting with dependent features, we formally establish that FACT can provide theoretically justified random forests feature p-values and enjoy appealing power through nonasymptotic analyses. The theoretical results and finite-sample advantages of the newly suggested method are illustrated with several simulation examples and an economic forecasting application in relation to COVID-19.

Breaking Feedback Loops in Recommender Systems with Causal Inference

Recommender systems play a key role in shaping modern web ecosystems. These systems alternate between (1) making recommendations (2) collecting user responses to these recommendations, and (3) retraining the recommendation algorithm based on this feedback. During this process the recommender system influences the user behavioral data that is subsequently used to update it, thus creating a feedback loop. Recent work has shown that feedback loops may compromise recommendation quality and homogenize user behavior, raising ethical and performance concerns when deploying recommender systems. To address these issues, we propose the Causal Adjustment for Feedback Loops (CAFL), an algorithm that provably breaks feedback loops using causal inference and can be applied to any recommendation algorithm that optimizes a training loss. Our main observation is that a recommender system does not suffer from feedback loops if it reasons about causal quantities, namely the intervention distributions of recommendations on user ratings. Moreover, we can calculate this intervention distribution from observational data by adjusting for the recommender system's predictions of user preferences. Using simulated environments, we demonstrate that CAFL improves recommendation quality when compared to prior correction methods.

Beyond mAP: Re-evaluating and Improving Performance in Instance Segmentation with Semantic Sorting and Contrastive Flow

Top-down instance segmentation methods improve mAP by hedging bets on low-confidence predictions to match a ground truth. Moreover, the query-key paradigm of top-down methods leads to the instance merging problem. An excessive number of duplicate predictions leads to the (over)counting error, and the independence of category and localization branches leads to the naming error. The de-facto mAP metric doesn't capture these errors, as we show that a trivial dithering scheme can simultaneously increase mAP with hedging errors. To this end, we propose two graph-based metrics that quantifies the amount of hedging both inter-and intra-class. We conjecture the source of the hedging problem is due to feature merging and propose a) Contrastive Flow Field to encode contextual differences between instances as a supervisory signal, and b) Semantic Sorting and NMS step to suppress duplicates and incorrectly categorized prediction. Ablations show that our method encodes contextual information better than baselines, and experiments on COCO our method simultaneously reduces merging and hedging errors compared to state-of-the-art instance segmentation methods.

Doubly-Asynchronous Value Iteration: Making Value Iteration Asynchronous in Actions

Value iteration (VI) is a foundational dynamic programming method, important for learning and planning in optimal control and reinforcement learning. VI proceeds in batches, where the update to the value of each state must be completed before the next batch of updates can begin. Completing a single batch is prohibitively expensive if the state space is large, rendering VI impractical for many applications. Asynchronous VI helps to address the large state space problem by updating one state at a time, in-place and in an arbitrary order. However, Asynchronous VI still requires a maximization over the entire action space, making it impractical for domains with large action space. To address this issue, we propose doubly-asynchronous value iteration (DAVI), a new algorithm that generalizes the idea of asynchrony from states to states and actions. More concretely, DAVI maximizes over a sampled subset of actions that can be of any user-defined size. This simple approach of using sampling to reduce computation maintains similarly appealing theoretical properties to VI without the need to wait for a full sweep through the entire action space in each update. In this paper, we show DAVI converges to the optimal value function with probability one, converges at a near-geometric rate with probability 1-delta, and returns a near-optimal policy in computation time that nearly matches a previously established bound for VI. We also empirically demonstrate DAVI's effectiveness in several experiments.

Recommendation Systems with Distribution-Free Reliability Guarantees

When building recommendation systems, we seek to output a helpful set of items to the user. Under the hood, a ranking model predicts which of two candidate items is better, and we must distill these pairwise comparisons into the user-facing output. However, a learned ranking model is never perfect, so taking its predictions at face value gives no guarantee that the user-facing output is reliable. Building from a pre-trained ranking model, we show how to return a set of items that is rigorously guaranteed to contain mostly good items. Our procedure endows any ranking model with rigorous finite-sample control of the false discovery rate (FDR), regardless of the (unknown) data distribution. Moreover, our calibration algorithm enables the easy and principled integration of multiple objectives in recommender systems. As an example, we show how to optimize for recommendation diversity subject to a user-specified level of FDR control, circumventing the need to specify ad hoc weights of a diversity loss against an accuracy loss. Throughout, we focus on the problem of learning to rank a set of possible recommendations, evaluating our methods on the Yahoo! Learning to Rank and MSMarco datasets.

Invariant and Transportable Representations for Anti-Causal Domain Shifts

Real-world classification problems must contend with domain shift, the (potential) mismatch between the domain where a model is deployed and the domain(s) where the training data was gathered. Methods to handle such problems must specify what structure is common between the domains and what varies. A natural assumption is that causal (structural) relationships are invariant in all domains. Then, it is tempting to learn a predictor for label $Y$ that depends only on its causal parents. However, many real-world problems are "anti-causal" in the sense that $Y$ is a cause of the covariates $X$ -- in this case, $Y$ has no causal parents and the naive causal invariance is useless. In this paper, we study representation learning under a particular notion of domain shift that both respects causal invariance and that naturally handles the "anti-causal" structure. We show how to leverage the shared causal structure of the domains to learn a representation that both admits an invariant predictor and that also allows fast adaptation in new domains. The key is to translate causal assumptions into learning principles that disentangle "invariant" and "non-stable" features. Experiments on both synthetic and real-world data demonstrate the effectiveness of the proposed learning algorithm. Code is available at https://github.com/ybjiaang/ACTIR.

Deep Learning for Short-term Instant Energy Consumption Forecasting in the Manufacturing Sector

Electricity is a volatile power source that requires great planning and resource management for both short and long term. More specifically, in the short-term, accurate instant energy consumption forecasting contributes greatly to improve the efficiency of buildings, opening new avenues for the adoption of renewable energy. In that regard, data-driven approaches, namely the ones based on machine learning, are begin to be preferred over more traditional ones since they provide not only more simplified ways of deployment but also state of the art results. In that sense, this work applies and compares the performance of several deep learning algorithms, LSTM, CNN, mixed CNN-LSTM and TCN, in a real testbed within the manufacturing sector. The experimental results suggest that the TCN is the most reliable method for predicting instant energy consumption in the short-term.

E2Efold-3D: End-to-End Deep Learning Method for accurate de novo RNA 3D Structure Prediction

RNA structure determination and prediction can promote RNA-targeted drug development and engineerable synthetic elements design. But due to the intrinsic structural flexibility of RNAs, all the three mainstream structure determination methods (X-ray crystallography, NMR, and Cryo-EM) encounter challenges when resolving the RNA structures, which leads to the scarcity of the resolved RNA structures. Computational prediction approaches emerge as complementary to the experimental techniques. However, none of the \textit{de novo} approaches is based on deep learning since too few structures are available. Instead, most of them apply the time-consuming sampling-based strategies, and their performance seems to hit the plateau. In this work, we develop the first end-to-end deep learning approach, E2Efold-3D, to accurately perform the \textit{de novo} RNA structure prediction. Several novel components are proposed to overcome the data scarcity, such as a fully-differentiable end-to-end pipeline, secondary structure-assisted self-distillation, and parameter-efficient backbone formulation. Such designs are validated on the independent, non-overlapping RNA puzzle testing dataset and reach an average sub-4 \AA{} root-mean-square deviation, demonstrating its superior performance compared to state-of-the-art approaches. Interestingly, it also achieves promising results when predicting RNA complex structures, a feat that none of the previous systems could accomplish. When E2Efold-3D is coupled with the experimental techniques, the RNA structure prediction field can be greatly advanced.

Interpretable Fusion Analytics Framework for fMRI Connectivity: Self-Attention Mechanism and Latent Space Item-Response Model

There have been several attempts to use deep learning based on brain fMRI signals to classify cognitive impairment diseases. However, deep learning is a hidden black box model that makes it difficult to interpret the process of classification. To address this issue, we propose a novel analytical framework that interprets the classification result from deep learning processes. We first derive the region of interest (ROI) functional connectivity network (FCN) by embedding functions based on their similar signal patterns. Then, using the self-attention equipped deep learning model, we classify diseases based on their FCN. Finally, in order to interpret the classification results, we employ a latent space item-response interaction network model to identify the significant functions that exhibit distinct connectivity patterns when compared to other diseases. The application of this proposed framework to the four types of cognitive impairment shows that our approach is valid for determining the significant ROI functions.

Dynamic Spatial Sparsification for Efficient Vision Transformers and Convolutional Neural Networks

In this paper, we present a new approach for model acceleration by exploiting spatial sparsity in visual data. We observe that the final prediction in vision Transformers is only based on a subset of the most informative tokens, which is sufficient for accurate image recognition. Based on this observation, we propose a dynamic token sparsification framework to prune redundant tokens progressively and dynamically based on the input to accelerate vision Transformers. Specifically, we devise a lightweight prediction module to estimate the importance score of each token given the current features. The module is added to different layers to prune redundant tokens hierarchically. While the framework is inspired by our observation of the sparse attention in vision Transformers, we find the idea of adaptive and asymmetric computation can be a general solution for accelerating various architectures. We extend our method to hierarchical models including CNNs and hierarchical vision Transformers as well as more complex dense prediction tasks that require structured feature maps by formulating a more generic dynamic spatial sparsification framework with progressive sparsification and asymmetric computation for different spatial locations. By applying lightweight fast paths to less informative features and using more expressive slow paths to more important locations, we can maintain the structure of feature maps while significantly reducing the overall computations. Extensive experiments demonstrate the effectiveness of our framework on various modern architectures and different visual recognition tasks. Our results clearly demonstrate that dynamic spatial sparsification offers a new and more effective dimension for model acceleration. Code is available at https://github.com/raoyongming/DynamicViT

Spatiotemporal Feature Learning Based on Two-Step LSTM and Transformer for CT Scans

Computed tomography (CT) imaging could be very practical for diagnosing various diseases. However, the nature of the CT images is even more diverse since the resolution and number of the slices of a CT scan are determined by the machine and its settings. Conventional deep learning models are hard to tickle such diverse data since the essential requirement of the deep neural network is the consistent shape of the input data. In this paper, we propose a novel, effective, two-step-wise approach to tickle this issue for COVID-19 symptom classification thoroughly. First, the semantic feature embedding of each slice for a CT scan is extracted by conventional backbone networks. Then, we proposed a long short-term memory (LSTM) and Transformer-based sub-network to deal with temporal feature learning, leading to spatiotemporal feature representation learning. In this fashion, the proposed two-step LSTM model could prevent overfitting, as well as increase performance. Comprehensive experiments reveal that the proposed two-step method not only shows excellent performance but also could be compensated for each other. More specifically, the two-step LSTM model has a lower false-negative rate, while the 2-step Swin model has a lower false-positive rate. In summary, it is suggested that the model ensemble could be adopted for more stable and promising performance in real-world applications.

ViRel: Unsupervised Visual Relations Discovery with Graph-level Analogy

Visual relations form the basis of understanding our compositional world, as relationships between visual objects capture key information in a scene. It is then advantageous to learn relations automatically from the data, as learning with predefined labels cannot capture all possible relations. However, current relation learning methods typically require supervision, and are not designed to generalize to scenes with more complicated relational structures than those seen during training. Here, we introduce ViRel, a method for unsupervised discovery and learning of Visual Relations with graph-level analogy. In a setting where scenes within a task share the same underlying relational subgraph structure, our learning method of contrasting isomorphic and non-isomorphic graphs discovers the relations across tasks in an unsupervised manner. Once the relations are learned, ViRel can then retrieve the shared relational graph structure for each task by parsing the predicted relational structure. Using a dataset based on grid-world and the Abstract Reasoning Corpus, we show that our method achieves above 95% accuracy in relation classification, discovers the relation graph structure for most tasks, and further generalizes to unseen tasks with more complicated relational structures.

De-Biasing Generative Models using Counterfactual Methods

Variational autoencoders (VAEs) and other generative methods have garnered growing interest not just for their generative properties but also for the ability to dis-entangle a low-dimensional latent variable space. However, few existing generative models take causality into account. We propose a new decoder based framework named the Causal Counterfactual Generative Model (CCGM), which includes a partially trainable causal layer in which a part of a causal model can be learned without significantly impacting reconstruction fidelity. By learning the causal relationships between image semantic labels or tabular variables, we can analyze biases, intervene on the generative model, and simulate new scenarios. Furthermore, by modifying the causal structure, we can generate samples outside the domain of the original training data and use such counterfactual models to de-bias datasets. Thus, datasets with known biases can still be used to train the causal generative model and learn the causal relationships, but we can produce de-biased datasets on the generative side. Our proposed method combines a causal latent space VAE model with specific modification to emphasize causal fidelity, enabling finer control over the causal layer and the ability to learn a robust intervention framework. We explore how better disentanglement of causal learning and encoding/decoding generates higher causal intervention quality. We also compare our model against similar research to demonstrate the need for explicit generative de-biasing beyond interventions. Our initial experiments show that our model can generate images and tabular data with high fidelity to the causal framework and accommodate explicit de-biasing to ignore undesired relationships in the causal data compared to the baseline.

Goal-Conditioned Generators of Deep Policies

Goal-conditioned Reinforcement Learning (RL) aims at learning optimal policies, given goals encoded in special command inputs. Here we study goal-conditioned neural nets (NNs) that learn to generate deep NN policies in form of context-specific weight matrices, similar to Fast Weight Programmers and other methods from the 1990s. Using context commands of the form "generate a policy that achieves a desired expected return," our NN generators combine powerful exploration of parameter space with generalization across commands to iteratively find better and better policies. A form of weight-sharing HyperNetworks and policy embeddings scales our method to generate deep NNs. Experiments show how a single learned policy generator can produce policies that achieve any return seen during training. Finally, we evaluate our algorithm on a set of continuous control tasks where it exhibits competitive performance. Our code is public.

General Policy Evaluation and Improvement by Learning to Identify Few But Crucial States

Learning to evaluate and improve policies is a core problem of Reinforcement Learning (RL). Traditional RL algorithms learn a value function defined for a single policy. A recently explored competitive alternative is to learn a single value function for many policies. Here we combine the actor-critic architecture of Parameter-Based Value Functions and the policy embedding of Policy Evaluation Networks to learn a single value function for evaluating (and thus helping to improve) any policy represented by a deep neural network (NN). The method yields competitive experimental results. In continuous control problems with infinitely many states, our value function minimizes its prediction error by simultaneously learning a small set of `probing states' and a mapping from actions produced in probing states to the policy's return. The method extracts crucial abstract knowledge about the environment in form of very few states sufficient to fully specify the behavior of many policies. A policy improves solely by changing actions in probing states, following the gradient of the value function's predictions. Surprisingly, it is possible to clone the behavior of a near-optimal policy in Swimmer-v3 and Hopper-v3 environments only by knowing how to act in 3 and 5 such learned states, respectively. Remarkably, our value function trained to evaluate NN policies is also invariant to changes of the policy architecture: we show that it allows for zero-shot learning of linear policies competitive with the best policy seen during training. Our code is public.

Fidelity of Ensemble Aggregation for Saliency Map Explanations using Bayesian Optimization Techniques

In recent years, an abundance of feature attribution methods for explaining neural networks have been developed. Especially in the field of computer vision, many methods for generating saliency maps providing pixel attributions exist. However, their explanations often contradict each other and it is not clear which explanation to trust. A natural solution to this problem is the aggregation of multiple explanations. We present and compare different pixel-based aggregation schemes with the goal of generating a new explanation, whose fidelity to the model's decision is higher than each individual explanation. Using methods from the field of Bayesian Optimization, we incorporate the variance between the individual explanations into the aggregation process. Additionally, we analyze the effect of multiple normalization techniques on ensemble aggregation.

Progressive Latent Replay for efficient Generative Rehearsal

We introduce a new method for internal replay that modulates the frequency of rehearsal based on the depth of the network. While replay strategies mitigate the effects of catastrophic forgetting in neural networks, recent works on generative replay show that performing the rehearsal only on the deeper layers of the network improves the performance in continual learning. However, the generative approach introduces additional computational overhead, limiting its applications. Motivated by the observation that earlier layers of neural networks forget less abruptly, we propose to update network layers with varying frequency using intermediate-level features during replay. This reduces the computational burden by omitting computations for both deeper layers of the generator and earlier layers of the main model. We name our method Progressive Latent Replay and show that it outperforms Internal Replay while using significantly fewer resources.

Selectively increasing the diversity of GAN-generated samples

Generative Adversarial Networks (GANs) are powerful models able to synthesize data samples closely resembling the distribution of real data, yet the diversity of those generated samples is limited due to the so-called mode collapse phenomenon observed in GANs. Especially prone to mode collapse are conditional GANs, which tend to ignore the input noise vector and focus on the conditional information. Recent methods proposed to mitigate this limitation increase the diversity of generated samples, yet they reduce the performance of the models when similarity of samples is required. To address this shortcoming, we propose a novel method to selectively increase the diversity of GAN-generated samples. By adding a simple, yet effective regularization to the training loss function we encourage the generator to discover new data modes for inputs related to diverse outputs while generating consistent samples for the remaining ones. More precisely, we maximise the ratio of distances between generated images and input latent vectors scaling the effect according to the diversity of samples for a given conditional input. We show the superiority of our method in a synthetic benchmark as well as a real-life scenario of simulating data from the Zero Degree Calorimeter of ALICE experiment in LHC, CERN.

High-Dimensional Private Empirical Risk Minimization by Greedy Coordinate Descent

In this paper, we study differentially private empirical risk minimization (DP-ERM). It has been shown that the (worst-case) utility of DP-ERM reduces as the dimension increases. This is a major obstacle to privately learning large machine learning models. In high dimension, it is common for some model's parameters to carry more information than others. To exploit this, we propose a differentially private greedy coordinate descent (DP-GCD) algorithm. At each iteration, DP-GCD privately performs a coordinate-wise gradient step along the gradients' (approximately) greatest entry. We show theoretically that DP-GCD can improve utility by exploiting structural properties of the problem's solution (such as sparsity or quasi-sparsity), with very fast progress in early iterations. We then illustrate this numerically, both on synthetic and real datasets. Finally, we describe promising directions for future work.

Learning from Multiple Unlabeled Datasets with Partial Risk Regularization

Recent years have witnessed a great success of supervised deep learning, where predictive models were trained from a large amount of fully labeled data. However, in practice, labeling such big data can be very costly and may not even be possible for privacy reasons. Therefore, in this paper, we aim to learn an accurate classifier without any class labels. More specifically, we consider the case where multiple sets of unlabeled data and only their class priors, i.e., the proportions of each class, are available. Under this problem setup, we first derive an unbiased estimator of the classification risk that can be estimated from the given unlabeled sets and theoretically analyze the generalization error of the learned classifier. We then find that the classifier obtained as such tends to cause overfitting as its empirical risks go negative during training. To prevent overfitting, we further propose a partial risk regularization that maintains the partial risks with respect to unlabeled datasets and classes to certain levels. Experiments demonstrate that our method effectively mitigates overfitting and outperforms state-of-the-art methods for learning from multiple unlabeled sets.

Counterbalancing Teacher: Regularizing Batch Normalized Models for Robustness

Batch normalization (BN) is a ubiquitous technique for training deep neural networks that accelerates their convergence to reach higher accuracy. However, we demonstrate that BN comes with a fundamental drawback: it incentivizes the model to rely on low-variance features that are highly specific to the training (in-domain) data, hurting generalization performance on out-of-domain examples. In this work, we investigate this phenomenon by first showing that removing BN layers across a wide range of architectures leads to lower out-of-domain and corruption errors at the cost of higher in-domain errors. We then propose Counterbalancing Teacher (CT), a method which leverages a frozen copy of the same model without BN as a teacher to enforce the student network's learning of robust representations by substantially adapting its weights through a consistency loss function. This regularization signal helps CT perform well in unforeseen data shifts, even without information from the target domain as in prior works. We theoretically show in an overparameterized linear regression setting why normalization leads to a model's reliance on such in-domain features, and empirically demonstrate the efficacy of CT by outperforming several baselines on robustness benchmarks such as CIFAR-10-C, CIFAR-100-C, and VLCS.

Wild Networks: Exposure of 5G Network Infrastructures to Adversarial Examples

Fifth Generation (5G) networks must support billions of heterogeneous devices while guaranteeing optimal Quality of Service (QoS). Such requirements are impossible to meet with human effort alone, and Machine Learning (ML) represents a core asset in 5G. ML, however, is known to be vulnerable to adversarial examples; moreover, as our paper will show, the 5G context is exposed to a yet another type of adversarial ML attacks that cannot be formalized with existing threat models. Proactive assessment of such risks is also challenging due to the lack of ML-powered 5G equipment available for adversarial ML research. To tackle these problems, we propose a novel adversarial ML threat model that is particularly suited to 5G scenarios, and is agnostic to the precise function solved by ML. In contrast to existing ML threat models, our attacks do not require any compromise of the target 5G system while still being viable due to the QoS guarantees and the open nature of 5G networks. Furthermore, we propose an original framework for realistic ML security assessments based on public data. We proactively evaluate our threat model on 6 applications of ML envisioned in 5G. Our attacks affect both the training and the inference stages, can degrade the performance of state-of-the-art ML systems, and have a lower entry barrier than previous attacks.

Variational Neural Networks

Bayesian Neural Networks (BNNs) provide a tool to estimate the uncertainty of a neural network by considering a distribution over weights and sampling different models for each input. In this paper, we propose a method for uncertainty estimation in neural networks called Variational Neural Network that, instead of considering a distribution over weights, generates parameters for the output distribution of a layer by transforming its inputs with learnable sub-layers. In uncertainty quality estimation experiments, we show that VNNs achieve better uncertainty quality than Monte Carlo Dropout or Bayes By Backpropagation methods.

Learning state machines via efficient hashing of future traces

State machines are popular models to model and visualize discrete systems such as software systems, and to represent regular grammars. Most algorithms that passively learn state machines from data assume all the data to be available from the beginning and they load this data into memory. This makes it hard to apply them to continuously streaming data and results in large memory requirements when dealing with large datasets. In this paper we propose a method to learn state machines from data streams using the count-min-sketch data structure to reduce memory requirements. We apply state merging using the well-known red-blue-framework to reduce the search space. We implemented our approach in an established framework for learning state machines, and evaluated it on a well know dataset to provide experimental data, showing the effectiveness of our approach with respect to quality of the results and run-time.

Mix and Match: An Empirical Study on Training Corpus Composition for Polyglot Text-To-Speech (TTS)

Training multilingual Neural Text-To-Speech (NTTS) models using only monolingual corpora has emerged as a popular way for building voice cloning based Polyglot NTTS systems. In order to train these models, it is essential to understand how the composition of the training corpora affects the quality of multilingual speech synthesis. In this context, it is common to hear questions such as "Would including more Spanish data help my Italian synthesis, given the closeness of both languages?". Unfortunately, we found existing literature on the topic lacking in completeness in this regard. In the present work, we conduct an extensive ablation study aimed at understanding how various factors of the training corpora, such as language family affiliation, gender composition, and the number of speakers, contribute to the quality of Polyglot synthesis. Our findings include the observation that female speaker data are preferred in most scenarios, and that it is not always beneficial to have more speakers from the target language variant in the training corpus. The findings herein are informative for the process of data procurement and corpora building.

Slice-by-slice deep learning aided oropharyngeal cancer segmentation with adaptive thresholding for spatial uncertainty on FDG PET and CT images

Tumor segmentation is a fundamental step for radiotherapy treatment planning. To define an accurate segmentation of the primary tumor (GTVp) of oropharyngeal cancer patients (OPC), simultaneous assessment of different image modalities is needed, and each image volume is explored slice-by-slice from different orientations. Moreover, the manual fixed boundary of segmentation neglects the spatial uncertainty known to occur in tumor delineation. This study proposes a novel automatic deep learning (DL) model to assist radiation oncologists in a slice-by-slice adaptive GTVp segmentation on registered FDG PET/CT images. We included 138 OPC patients treated with (chemo)radiation in our institute. Our DL framework exploits both inter and intra-slice context. Sequences of 3 consecutive 2D slices of concatenated FDG PET/CT images and GTVp contours were used as input. A 3-fold cross validation was performed three times, training on sequences extracted from the Axial (A), Sagittal (S), and Coronal (C) plane of 113 patients. Since consecutive sequences in a volume contain overlapping slices, each slice resulted in three outcome predictions that were averaged. In the A, S, and C planes, the output shows areas with different probabilities of predicting the tumor. The performance of the models was assessed on 25 patients at different probability thresholds using the mean Dice Score Coefficient (DSC). Predictions were the closest to the ground truth at a probability threshold of 0.9 (DSC of 0.70 in the A, 0.77 in the S, and 0.80 in the C plane). The promising results of the proposed DL model show that the probability maps on registered FDG PET/CT images could guide radiation oncologists in a slice-by-slice adaptive GTVp segmentation.

Deep Contrastive One-Class Time Series Anomaly Detection

The accumulation of time series data and the absence of labels make time-series Anomaly Detection (AD) a self-supervised deep learning task. Single-assumption-based methods may only touch on a certain aspect of the whole normality, not sufficient to detect various anomalies. Among them, contrastive learning methods adopted for AD always choose negative pairs that are both normal to push away, which is objecting to AD tasks' purpose. Existing multi-assumption-based methods are usually two-staged, firstly applying a pre-training process whose target may differ from AD, so the performance is limited by the pre-trained representations. This paper proposes a deep Contrastive One-Class Anomaly detection method of time series (COCA), which combines the normality assumptions of contrastive learning and one-class classification. The key idea is to treat the representation and reconstructed representation as the positive pair of negative-samples-free contrastive learning, and we name it sequence contrast. Then we apply a contrastive one-class loss function composed of invariance and variance terms, the former optimizing loss of the two assumptions simultaneously, and the latter preventing hypersphere collapse. Extensive experiments conducted on four real-world time-series datasets show the superior performance of the proposed method achieves state-of-the-art. The code is publicly available at https://github.com/ruiking04/COCA.

Generalisable Methods for Early Prediction in Interactive Simulations for Education

Interactive simulations allow students to discover the underlying principles of a scientific phenomenon through their own exploration. Unfortunately, students often struggle to learn effectively in these environments. Classifying students' interaction data in the simulations based on their expected performance has the potential to enable adaptive guidance and consequently improve students' learning. Previous research in this field has mainly focused on a-posteriori analyses or investigations limited to one specific predictive model and simulation. In this paper, we investigate the quality and generalisability of models for an early prediction of conceptual understanding based on clickstream data of students across interactive simulations. We first measure the students' conceptual understanding through their in-task performance. Then, we suggest a novel type of features that, starting from clickstream data, encodes both the state of the simulation and the action performed by the student. We finally propose to feed these features into GRU-based models, with and without attention, for prediction. Experiments on two different simulations and with two different populations show that our proposed models outperform shallow learning baselines and better generalise to different learning environments and populations. The inclusion of attention into the model increases interpretability in terms of effective inquiry. The source code is available on Github (https://github.com/epfl-ml4ed/beerslaw-lab.git).

GlowVC: Mel-spectrogram space disentangling model for language-independent text-free voice conversion

In this paper, we propose GlowVC: a multilingual multi-speaker flow-based model for language-independent text-free voice conversion. We build on Glow-TTS, which provides an architecture that enables use of linguistic features during training without the necessity of using them for VC inference. We consider two versions of our model: GlowVC-conditional and GlowVC-explicit. GlowVC-conditional models the distribution of mel-spectrograms with speaker-conditioned flow and disentangles the mel-spectrogram space into content- and pitch-relevant dimensions, while GlowVC-explicit models the explicit distribution with unconditioned flow and disentangles said space into content-, pitch- and speaker-relevant dimensions. We evaluate our models in terms of intelligibility, speaker similarity and naturalness for intra- and cross-lingual conversion in seen and unseen languages. GlowVC models greatly outperform AutoVC baseline in terms of intelligibility, while achieving just as high speaker similarity in intra-lingual VC, and slightly worse in the cross-lingual setting. Moreover, we demonstrate that GlowVC-explicit surpasses both GlowVC-conditional and AutoVC in terms of naturalness.

Solving the Traveling Salesperson Problem with Precedence Constraints by Deep Reinforcement Learning

This work presents solutions to the Traveling Salesperson Problem with precedence constraints (TSPPC) using Deep Reinforcement Learning (DRL) by adapting recent approaches that work well for regular TSPs. Common to these approaches is the use of graph models based on multi-head attention (MHA) layers. One idea for solving the pickup and delivery problem (PDP) is using heterogeneous attentions to embed the different possible roles each node can take. In this work, we generalize this concept of heterogeneous attentions to the TSPPC. Furthermore, we adapt recent ideas to sparsify attentions for better scalability. Overall, we contribute to the research community through the application and evaluation of recent DRL methods in solving the TSPPC.

Comparing Feature Importance and Rule Extraction for Interpretability on Text Data

Complex machine learning algorithms are used more and more often in critical tasks involving text data, leading to the development of interpretability methods. Among local methods, two families have emerged: those computing importance scores for each feature and those extracting simple logical rules. In this paper we show that using different methods can lead to unexpectedly different explanations, even when applied to simple models for which we would expect qualitative coincidence. To quantify this effect, we propose a new approach to compare explanations produced by different methods.

The Neural-Prediction based Acceleration Algorithm of Column Generation for Graph-Based Set Covering Problems

Set covering problem is an important class of combinatorial optimization problems, which has been widely applied and studied in many fields. In this paper, we propose an improved column generation algorithm with neural prediction (CG-P) for solving graph-based set covering problems. We leverage a graph neural network based neural prediction model to predict the probability to be included in the final solution for each edge. Our CG-P algorithm constructs a reduced graph that only contains the edges with higher predicted probability, and this graph reduction process significantly speeds up the solution process. We evaluate the CG-P algorithm on railway crew scheduling problems and it outperforms the baseline column generation algorithm. We provide two solution modes for our CG-P algorithm. In the optimal mode, we can obtain a solution with an optimality guarantee while reducing the time cost to 63.12%. In the fast mode, we can obtain a sub-optimal solution with a 7.62% optimality gap in only 2.91% computation time.

Using contextual sentence analysis models to recognize ESG concepts

This paper summarizes the joint participation of the Trading Central Labs and the L3i laboratory of the University of La Rochelle on both sub-tasks of the Shared Task FinSim-4 evaluation campaign. The first sub-task aims to enrich the 'Fortia ESG taxonomy' with new lexicon entries while the second one aims to classify sentences to either 'sustainable' or 'unsustainable' with respect to ESG (Environment, Social and Governance) related factors. For the first sub-task, we proposed a model based on pre-trained Sentence-BERT models to project sentences and concepts in a common space in order to better represent ESG concepts. The official task results show that our system yields a significant performance improvement compared to the baseline and outperforms all other submissions on the first sub-task. For the second sub-task, we combine the RoBERTa model with a feed-forward multi-layer perceptron in order to extract the context of sentences and classify them. Our model achieved high accuracy scores (over 92%) and was ranked among the top 5 systems.

Hessian-Free Second-Order Adversarial Examples for Adversarial Learning

Recent studies show deep neural networks (DNNs) are extremely vulnerable to the elaborately designed adversarial examples. Adversarial learning with those adversarial examples has been proved as one of the most effective methods to defend against such an attack. At present, most existing adversarial examples generation methods are based on first-order gradients, which can hardly further improve models' robustness, especially when facing second-order adversarial attacks. Compared with first-order gradients, second-order gradients provide a more accurate approximation of the loss landscape with respect to natural examples. Inspired by this, our work crafts second-order adversarial examples and uses them to train DNNs. Nevertheless, second-order optimization involves time-consuming calculation for Hessian-inverse. We propose an approximation method through transforming the problem into an optimization in the Krylov subspace, which remarkably reduce the computational complexity to speed up the training procedure. Extensive experiments conducted on the MINIST and CIFAR-10 datasets show that our adversarial learning with second-order adversarial examples outperforms other fisrt-order methods, which can improve the model robustness against a wide range of attacks.

Autonomous Drug Design with Multi-armed Bandits

Recent developments in artificial intelligence and automation could potentially enable a new drug design paradigm: autonomous drug design. Under this paradigm, generative models provide suggestions on thousands of molecules with specific properties. However, since only a limited number of molecules can be synthesized and tested, an obvious challenge is how to efficiently select these. We formulate this task as a contextual stochastic multi-armed bandit problem with multiple plays and volatile arms. Then, to solve it, we extend previous work on multi-armed bandits to reflect this setting, and compare our solution with random sampling, greedy selection and decaying-epsilon-greedy selection. To investigate how the different selection strategies affect the cumulative reward and the diversity of the selections, we simulate the drug design process. According to the simulation results, our approach has the potential for better exploring and exploiting the chemical space for autonomous drug design.

Task-oriented Self-supervised Learning for Anomaly Detection in Electroencephalography

Accurate automated analysis of electroencephalography (EEG) would largely help clinicians effectively monitor and diagnose patients with various brain diseases. Compared to supervised learning with labelled disease EEG data which can train a model to analyze specific diseases but would fail to monitor previously unseen statuses, anomaly detection based on only normal EEGs can detect any potential anomaly in new EEGs. Different from existing anomaly detection strategies which do not consider any property of unavailable abnormal data during model development, a task-oriented self-supervised learning approach is proposed here which makes use of available normal EEGs and expert knowledge about abnormal EEGs to train a more effective feature extractor for the subsequent development of anomaly detector. In addition, a specific two branch convolutional neural network with larger kernels is designed as the feature extractor such that it can more easily extract both larger scale and small-scale features which often appear in unavailable abnormal EEGs. The effectively designed and trained feature extractor has shown to be able to extract better feature representations from EEGs for development of anomaly detector based on normal data and future anomaly detection for new EEGs, as demonstrated on three EEG datasets. The code is available at https://github.com/ironing/EEG-AD.

GAN-based generation of realistic 3D data: A systematic review and taxonomy

Data has become the most valuable resource in today's world. With the massive proliferation of data-driven algorithms, such as deep learning-based approaches, the availability of data is of great interest. In this context, high-quality training, validation and testing datasets are particularly needed. Volumetric data is a very important resource in medicine, as it ranges from disease diagnoses to therapy monitoring. When the dataset is sufficient, models can be trained to help doctors with these tasks. Unfortunately, there are scenarios and applications where large amounts of data is unavailable. For example, in the medical field, rare diseases and privacy issues can lead to restricted data availability. In non-medical fields, the high cost of obtaining a sufficient amount of high-quality data can also be a concern. A solution to these problems can be the generation of synthetic data to perform data augmentation in combination with other more traditional methods of data augmentation. Therefore, most of the publications on 3D Generative Adversarial Networks (GANs) are within the medical domain. The existence of mechanisms to generate realistic synthetic data is a good asset to overcome this challenge, especially in healthcare, as the data must be of good quality and close to reality, i.e. realistic, and without privacy issues. In this review, we provide a summary of works that generate realistic 3D synthetic data using GANs. We therefore outline GAN-based methods in these areas with common architectures, advantages and disadvantages. We present a novel taxonomy, evaluations, challenges and research opportunities to provide a holistic overview of the current state of GANs in medicine and other fields.

Task Discrepancy Maximization for Fine-grained Few-Shot Classification

Recognizing discriminative details such as eyes and beaks is important for distinguishing fine-grained classes since they have similar overall appearances. In this regard, we introduce Task Discrepancy Maximization (TDM), a simple module for fine-grained few-shot classification. Our objective is to localize the class-wise discriminative regions by highlighting channels encoding distinct information of the class. Specifically, TDM learns task-specific channel weights based on two novel components: Support Attention Module (SAM) and Query Attention Module (QAM). SAM produces a support weight to represent channel-wise discriminative power for each class. Still, since the SAM is basically only based on the labeled support sets, it can be vulnerable to bias toward such support set. Therefore, we propose QAM which complements SAM by yielding a query weight that grants more weight to object-relevant channels for a given query image. By combining these two weights, a class-wise task-specific channel weight is defined. The weights are then applied to produce task-adaptive feature maps more focusing on the discriminative details. Our experiments validate the effectiveness of TDM and its complementary benefits with prior methods in fine-grained few-shot classification.

Probabilistic forecasting for geosteering in fluvial successions using a generative adversarial network

Quantitative workflows utilizing real-time data to constrain ahead-of-bit uncertainty have the potential to improve geosteering significantly. Fast updates based on real-time data are essential when drilling in complex reservoirs with high uncertainties in pre-drill models. However, practical assimilation of real-time data requires effective geological modeling and mathematically robust parameterization. We propose a generative adversarial deep neural network (GAN), trained to reproduce geologically consistent 2D sections of fluvial successions. Offline training produces a fast GAN-based approximation of complex geology parameterized as a 60-dimensional model vector with standard Gaussian distribution of each component. Probabilistic forecasts are generated using an ensemble of equiprobable model vector realizations. A forward-modeling sequence, including a GAN, converts the initial (prior) ensemble of realizations into EM log predictions. An ensemble smoother minimizes statistical misfits between predictions and real-time data, yielding an update of model vectors and reduced uncertainty around the well. Updates can be then translated to probabilistic predictions of facies and resistivities. The present paper demonstrates a workflow for geosteering in an outcrop-based, synthetic fluvial succession. In our example, the method reduces uncertainty and correctly predicts most major geological features up to 500 meters ahead of drill-bit.

Multi-scale alignment and Spatial ROI Module for COVID-19 Diagnosis

Coronavirus Disease 2019 (COVID-19) has spread globally and become a health crisis faced by humanity since first reported. Radiology imaging technologies such as computer tomography (CT) and chest X-ray imaging (CXR) are effective tools for diagnosing COVID-19. However, in CT and CXR images, the infected area occupies only a small part of the image. Some common deep learning methods that integrate large-scale receptive fields may cause the loss of image detail, resulting in the omission of the region of interest (ROI) in COVID-19 images and are therefore not suitable for further processing. To this end, we propose a deep spatial pyramid pooling (D-SPP) module to integrate contextual information over different resolutions, aiming to extract information under different scales of COVID-19 images effectively. Besides, we propose a COVID-19 infection detection (CID) module to draw attention to the lesion area and remove interference from irrelevant information. Extensive experiments on four CT and CXR datasets have shown that our method produces higher accuracy of detecting COVID-19 lesions in CT and CXR images. It can be used as a computer-aided diagnosis tool to help doctors effectively diagnose and screen for COVID-19.

Safe Reinforcement Learning via Confidence-Based Filters

Ensuring safety is a crucial challenge when deploying reinforcement learning (RL) to real-world systems. We develop confidence-based safety filters, a control-theoretic approach for certifying state safety constraints for nominal policies learned via standard RL techniques, based on probabilistic dynamics models. Our approach is based on a reformulation of state constraints in terms of cost functions, reducing safety verification to a standard RL task. By exploiting the concept of hallucinating inputs, we extend this formulation to determine a "backup" policy that is safe for the unknown system with high probability. Finally, the nominal policy is minimally adjusted at every time step during a roll-out towards the backup policy, such that safe recovery can be guaranteed afterwards. We provide formal safety guarantees, and empirically demonstrate the effectiveness of our approach.

Spectral Power Profile Optimization of Field-Deployed WDM Network by Remote Link Modeling

A digital twin model of a multi-node WDM network is obtained from a single access point. The model is used to predict and optimize the transmit power profile for each link in the network and up to 2.2~dB of margin improvements are obtained w.r.t. unoptimized transmission.

The least-control principle for learning at equilibrium

Equilibrium systems are a powerful way to express neural computations. As special cases, they include models of great current interest in both neuroscience and machine learning, such as equilibrium recurrent neural networks, deep equilibrium models, or meta-learning. Here, we present a new principle for learning such systems with a temporally- and spatially-local rule. Our principle casts learning as a least-control problem, where we first introduce an optimal controller to lead the system towards a solution state, and then define learning as reducing the amount of control needed to reach such a state. We show that incorporating learning signals within a dynamics as an optimal control enables transmitting credit assignment information in previously unknown ways, avoids storing intermediate states in memory, and does not rely on infinitesimal learning signals. In practice, our principle leads to strong performance matching that of leading gradient-based learning methods when applied to an array of problems involving recurrent neural networks and meta-learning. Our results shed light on how the brain might learn and offer new ways of approaching a broad class of machine learning problems.

Assessing the Performance of Automated Prediction and Ranking of Patient Age from Chest X-rays Against Clinicians

Understanding the internal physiological changes accompanying the aging process is an important aspect of medical image interpretation, with the expected changes acting as a baseline when reporting abnormal findings. Deep learning has recently been demonstrated to allow the accurate estimation of patient age from chest X-rays, and shows potential as a health indicator and mortality predictor. In this paper we present a novel comparative study of the relative performance of radiologists versus state-of-the-art deep learning models on two tasks: (a) patient age estimation from a single chest X-ray, and (b) ranking of two time-separated images of the same patient by age. We train our models with a heterogeneous database of 1.8M chest X-rays with ground truth patient ages and investigate the limitations on model accuracy imposed by limited training data and image resolution, and demonstrate generalisation performance on public data. To explore the large performance gap between the models and humans on these age-prediction tasks compared with other radiological reporting tasks seen in the literature, we incorporate our age prediction model into a conditional Generative Adversarial Network (cGAN) allowing visualisation of the semantic features identified by the prediction model as significant to age prediction, comparing the identified features with those relied on by clinicians.

NodeTrans: A Graph Transfer Learning Approach for Traffic Prediction

Recently, deep learning methods have made great progress in traffic prediction, but their performance depends on a large amount of historical data. In reality, we may face the data scarcity issue. In this case, deep learning models fail to obtain satisfactory performance. Transfer learning is a promising approach to solve the data scarcity issue. However, existing transfer learning approaches in traffic prediction are mainly based on regular grid data, which is not suitable for the inherent graph data in the traffic network. Moreover, existing graph-based models can only capture shared traffic patterns in the road network, and how to learn node-specific patterns is also a challenge. In this paper, we propose a novel transfer learning approach to solve the traffic prediction with few data, which can transfer the knowledge learned from a data-rich source domain to a data-scarce target domain. First, a spatial-temporal graph neural network is proposed, which can capture the node-specific spatial-temporal traffic patterns of different road networks. Then, to improve the robustness of transfer, we design a pattern-based transfer strategy, where we leverage a clustering-based mechanism to distill common spatial-temporal patterns in the source domain, and use these knowledge to further improve the prediction performance of the target domain. Experiments on real-world datasets verify the effectiveness of our approach.

A New Index for Clustering Evaluation Based on Density Estimation

A new index for internal evaluation of clustering is introduced. The index is defined as a mixture of two sub-indices. The first sub-index $ I_a $ is called the Ambiguous Index; the second sub-index $ I_s $ is called the Similarity Index. Calculation of the two sub-indices is based on density estimation to each cluster of a partition of the data. An experiment is conducted to test the performance of the new index, and compared with three popular internal clustering evaluation indices -- Calinski-Harabasz index, Silhouette coefficient, and Davies-Bouldin index, on a set of 145 datasets. The result shows the new index improves the three popular indices by 59\%, 34\%, and 74\%, correspondingly.

DiffML: End-to-end Differentiable ML Pipelines

In this paper, we present our vision of differentiable ML pipelines called DiffML to automate the construction of ML pipelines in an end-to-end fashion. The idea is that DiffML allows to jointly train not just the ML model itself but also the entire pipeline including data preprocessing steps, e.g., data cleaning, feature selection, etc. Our core idea is to formulate all pipeline steps in a differentiable way such that the entire pipeline can be trained using backpropagation. However, this is a non-trivial problem and opens up many new research questions. To show the feasibility of this direction, we demonstrate initial ideas and a general principle of how typical preprocessing steps such as data cleaning, feature selection and dataset selection can be formulated as differentiable programs and jointly learned with the ML model. Moreover, we discuss a research roadmap and core challenges that have to be systematically tackled to enable fully differentiable ML pipelines.

CPrune: Compiler-Informed Model Pruning for Efficient Target-Aware DNN Execution

Mobile devices run deep learning models for various purposes, such as image classification and speech recognition. Due to the resource constraints of mobile devices, researchers have focused on either making a lightweight deep neural network (DNN) model using model pruning or generating an efficient code using compiler optimization. Surprisingly, we found that the straightforward integration between model compression and compiler auto-tuning often does not produce the most efficient model for a target device. We propose CPrune, a compiler-informed model pruning for efficient target-aware DNN execution to support an application with a required target accuracy. CPrune makes a lightweight DNN model through informed pruning based on the structural information of subgraphs built during the compiler tuning process. Our experimental results show that CPrune increases the DNN execution speed up to 2.73x compared to the state-of-the-art TVM auto-tune while satisfying the accuracy requirement.

Learning Optimal Transport Between two Empirical Distributions with Normalizing Flows

Optimal transport (OT) provides effective tools for comparing and mapping probability measures. We propose to leverage the flexibility of neural networks to learn an approximate optimal transport map. More precisely, we present a new and original method to address the problem of transporting a finite set of samples associated with a first underlying unknown distribution towards another finite set of samples drawn from another unknown distribution. We show that a particular instance of invertible neural networks, namely the normalizing flows, can be used to approximate the solution of this OT problem between a pair of empirical distributions. To this aim, we propose to relax the Monge formulation of OT by replacing the equality constraint on the push-forward measure by the minimization of the corresponding Wasserstein distance. The push-forward operator to be retrieved is then restricted to be a normalizing flow which is trained by optimizing the resulting cost function. This approach allows the transport map to be discretized as a composition of functions. Each of these functions is associated to one sub-flow of the network, whose output provides intermediate steps of the transport between the original and target measures. This discretization yields also a set of intermediate barycenters between the two measures of interest. Experiments conducted on toy examples as well as a challenging task of unsupervised translation demonstrate the interest of the proposed method. Finally, some experiments show that the proposed approach leads to a good approximation of the true OT.

Parametric and Multivariate Uncertainty Calibration for Regression and Object Detection

Reliable spatial uncertainty evaluation of object detection models is of special interest and has been subject of recent work. In this work, we review the existing definitions for uncertainty calibration of probabilistic regression tasks. We inspect the calibration properties of common detection networks and extend state-of-the-art recalibration methods. Our methods use a Gaussian process (GP) recalibration scheme that yields parametric distributions as output (e.g. Gaussian or Cauchy). The usage of GP recalibration allows for a local (conditional) uncertainty calibration by capturing dependencies between neighboring samples. The use of parametric distributions such as as Gaussian allows for a simplified adaption of calibration in subsequent processes, e.g., for Kalman filtering in the scope of object tracking. In addition, we use the GP recalibration scheme to perform covariance estimation which allows for post-hoc introduction of local correlations between the output quantities, e.g., position, width, or height in object detection. To measure the joint calibration of multivariate and possibly correlated data, we introduce the quantile calibration error which is based on the Mahalanobis distance between the predicted distribution and the ground truth to determine whether the ground truth is within a predicted quantile. Our experiments show that common detection models overestimate the spatial uncertainty in comparison to the observed error. We show that the simple Isotonic Regression recalibration method is sufficient to achieve a good uncertainty quantification in terms of calibrated quantiles. In contrast, if normal distributions are required for subsequent processes, our GP-Normal recalibration method yields the best results. Finally, we show that our covariance estimation method is able to achieve best calibration results for joint multivariate calibration.

Approximate Vanishing Ideal Computations at Scale

The approximate vanishing ideal of a set of points $X = \{\mathbf{x}_1, \ldots, \mathbf{x}_m\}\subseteq [0,1]^n$ is the set of polynomials that approximately evaluate to $0$ over all points $\mathbf{x} \in X$ and admits an efficient representation by a finite set of polynomials called generators. Algorithms that construct this set of generators are extensively studied but ultimately find little practical application because their computational complexities are thought to be superlinear in the number of samples $m$. In this paper, we focus on scaling up the Oracle Approximate Vanishing Ideal algorithm (OAVI), one of the most powerful of these methods. We prove that the computational complexity of OAVI is not superlinear but linear in the number of samples $m$ and polynomial in the number of features $n$, making OAVI an attractive preprocessing technique for large-scale machine learning. To further accelerate OAVI's training time, we propose two changes: First, as the name suggests, OAVI makes repeated oracle calls to convex solvers throughout its execution. By replacing the Pairwise Conditional Gradients algorithm, one of the standard solvers used in OAVI, with the faster Blended Pairwise Conditional Gradients algorithm, we illustrate how OAVI directly benefits from advancements in the study of convex solvers. Second, we propose Inverse Hessian Boosting (IHB): IHB exploits the fact that OAVI repeatedly solves quadratic convex optimization problems that differ only by very little and whose solutions can be written in closed form using inverse Hessian information. By efficiently updating the inverse of the Hessian matrix, the convex optimization problems can be solved almost instantly, accelerating OAVI's training time by up to multiple orders of magnitude. We complement our theoretical analysis with extensive numerical experiments on data sets whose sample numbers are in the millions.

Look beyond labels: Incorporating functional summary information in Bayesian neural networks

Bayesian deep learning offers a principled approach to train neural networks that accounts for both aleatoric and epistemic uncertainty. In variational inference, priors are often specified over the weight parameters, but they do not capture the true prior knowledge in large and complex neural network architectures. We present a simple approach to incorporate summary information about the predicted probability (such as sigmoid or softmax score) outputs in Bayesian neural networks (BNNs). The available summary information is incorporated as augmented data and modeled with a Dirichlet process, and we derive the corresponding \emph{Summary Evidence Lower BOund}. We show how the method can inform the model about task difficulty or class imbalance. Extensive empirical experiments show that, with negligible computational overhead, the proposed method yields a BNN with a better calibration of uncertainty.

Masked Self-Supervision for Remaining Useful Lifetime Prediction in Machine Tools

Prediction of Remaining Useful Lifetime(RUL) in the modern manufacturing and automation workplace for machines and tools is essential in Industry 4.0. This is clearly evident as continuous tool wear, or worse, sudden machine breakdown will lead to various manufacturing failures which would clearly cause economic loss. With the availability of deep learning approaches, the great potential and prospect of utilizing these for RUL prediction have resulted in several models which are designed driven by operation data of manufacturing machines. Current efforts in these which are based on fully-supervised models heavily rely on the data labeled with their RULs. However, the required RUL prediction data (i.e. the annotated and labeled data from faulty and/or degraded machines) can only be obtained after the machine breakdown occurs. The scarcity of broken machines in the modern manufacturing and automation workplace in real-world situations increases the difficulty of getting sufficient annotated and labeled data. In contrast, the data from healthy machines is much easier to be collected. Noting this challenge and the potential for improved effectiveness and applicability, we thus propose (and also fully develop) a method based on the idea of masked autoencoders which will utilize unlabeled data to do self-supervision. In thus the work here, a noteworthy masked self-supervised learning approach is developed and utilized. This is designed to seek to build a deep learning model for RUL prediction by utilizing unlabeled data. The experiments to verify the effectiveness of this development are implemented on the C-MAPSS datasets (which are collected from the data from the NASA turbofan engine). The results rather clearly show that our development and approach here perform better, in both accuracy and effectiveness, for RUL prediction when compared with approaches utilizing a fully-supervised model.

WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents

Existing benchmarks for grounding language in interactive environments either lack real-world linguistic elements, or prove difficult to scale up due to substantial human involvement in the collection of data or feedback signals. To bridge this gap, we develop WebShop -- a simulated e-commerce website environment with $1.18$ million real-world products and $12,087$ crowd-sourced text instructions. Given a text instruction specifying a product requirement, an agent needs to navigate multiple types of webpages and issue diverse actions to find, customize, and purchase an item. WebShop provides several challenges for language grounding including understanding compositional instructions, query (re-)formulation, comprehending and acting on noisy text in webpages, and performing strategic exploration. We collect over $1,600$ human demonstrations for the task, and train and evaluate a diverse range of agents using reinforcement learning, imitation learning, and pre-trained image and language models. Our best model achieves a task success rate of $29\%$, which outperforms rule-based heuristics ($9.6\%$) but is far lower than human expert performance ($59\%$). We also analyze agent and human trajectories and ablate various model components to provide insights for developing future agents with stronger language understanding and decision making abilities. Finally, we show that agents trained on WebShop exhibit non-trivial sim-to-real transfer when evaluated on amazon.com, indicating the potential value of WebShop in developing practical web-based agents that can operate in the wild.

Pareto Optimization for Active Learning under Out-of-Distribution Data Scenarios

Pool-based Active Learning (AL) has achieved great success in minimizing labeling cost by sequentially selecting informative unlabeled samples from a large unlabeled data pool and querying their labels from oracle/annotators. However, existing AL sampling strategies might not work well in out-of-distribution (OOD) data scenarios, where the unlabeled data pool contains some data samples that do not belong to the classes of the target task. Achieving good AL performance under OOD data scenarios is a challenging task due to the natural conflict between AL sampling strategies and OOD sample detection. AL selects data that are hard to be classified by the current basic classifier (e.g., samples whose predicted class probabilities have high entropy), while OOD samples tend to have more uniform predicted class probabilities (i.e., high entropy) than in-distribution (ID) data. In this paper, we propose a sampling scheme, Monte-Carlo Pareto Optimization for Active Learning (POAL), which selects optimal subsets of unlabeled samples with fixed batch size from the unlabeled data pool. We cast the AL sampling task as a multi-objective optimization problem, and thus we utilize Pareto optimization based on two conflicting objectives: (1) the normal AL data sampling scheme (e.g., maximum entropy), and (2) the confidence of not being an OOD sample. Experimental results show its effectiveness on both classical Machine Learning (ML) and Deep Learning (DL) tasks.

Learning node embeddings via summary graphs: a brief theoretical analysis

Graph representation learning plays an important role in many graph mining applications, but learning embeddings of large-scale graphs remains a problem. Recent works try to improve scalability via graph summarization -- i.e., they learn embeddings on a smaller summary graph, and then restore the node embeddings of the original graph. However, all existing works depend on heuristic designs and lack theoretical analysis. Different from existing works, we contribute an in-depth theoretical analysis of three specific embedding learning methods based on introduced kernel matrix, and reveal that learning embeddings via graph summarization is actually learning embeddings on a approximate graph constructed by the configuration model. We also give analysis about approximation error. To the best of our knowledge, this is the first work to give theoretical analysis of this approach. Furthermore, our analysis framework gives interpretation of some existing methods and provides great insights for future work on this problem.

ETF Portfolio Construction via Neural Network trained on Financial Statement Data

Recently, the application of advanced machine learning methods for asset management has become one of the most intriguing topics. Unfortunately, the application of these methods, such as deep neural networks, is difficult due to the data shortage problem. To address this issue, we propose a novel approach using neural networks to construct a portfolio of exchange traded funds (ETFs) based on the financial statement data of their components. Although a number of ETFs and ETF-managed portfolios have emerged in the past few decades, the ability to apply neural networks to manage ETF portfolios is limited since the number and historical existence of ETFs are relatively smaller and shorter, respectively, than those of individual stocks. Therefore, we use the data of individual stocks to train our neural networks to predict the future performance of individual stocks and use these predictions and the portfolio deposit file (PDF) to construct a portfolio of ETFs. Multiple experiments have been performed, and we have found that our proposed method outperforms the baselines. We believe that our approach can be more beneficial when managing recently listed ETFs, such as thematic ETFs, of which there is relatively limited historical data for training advanced machine learning methods.