Machine Learning (cs.LG)

Abstract: Federated learning is an approach to collaboratively training machine learning models for multiple parties that prohibit data sharing. One of the challenges in federated learning is non-IID data between clients, as a single model can not fit the data distribution for all clients. Meta-learning, such as Per-FedAvg, is introduced to cope with the challenge. Meta-learning learns shared initial parameters for all clients. Each client employs gradient descent to adapt the initialization to local data distributions quickly to realize model personalization. However, due to non-convex loss function and randomness of sampling update, meta-learning approaches have unstable goals in local adaptation for the same client. This fluctuation in different adaptation directions hinders the convergence in meta-learning. To overcome this challenge, we use the historical local adapted model to restrict the direction of the inner loop and propose an elastic-constrained method. As a result, the current round inner loop keeps historical goals and adapts to better solutions. Experiments show our method boosts meta-learning convergence and improves personalization without additional calculation and communication. Our method achieved SOTA on all metrics in three public datasets.

Abstract: The vulnerability of deep neural network models to adversarial example attacks is a practical challenge in many artificial intelligence applications. A recent line of work shows that the use of randomization in adversarial training is the key to find optimal strategies against adversarial example attacks. However, in a fully randomized setting where both the defender and the attacker can use randomized strategies, there are no efficient algorithm for finding such an optimal strategy. To fill the gap, we propose the first algorithm of its kind, called FRAT, which models the problem with a new infinite-dimensional continuous-time flow on probability distribution spaces. FRAT maintains a lightweight mixture of models for the defender, with flexibility to efficiently update mixing weights and model parameters at each iteration. Furthermore, FRAT utilizes lightweight sampling subroutines to construct a random strategy for the attacker. We prove that the continuous-time limit of FRAT converges to a mixed Nash equilibria in a zero-sum game formed by a defender and an attacker. Experimental results also demonstrate the efficiency of FRAT on CIFAR-10 and CIFAR-100 datasets.

Abstract: We propose a novel value approximation method, namely Eigensubspace Regularized Critic (ERC) for deep reinforcement learning (RL). ERC is motivated by an analysis of the dynamics of Q-value approximation error in the Temporal-Difference (TD) method, which follows a path defined by the 1-eigensubspace of the transition kernel associated with the Markov Decision Process (MDP). It reveals a fundamental property of TD learning that has remained unused in previous deep RL approaches. In ERC, we propose a regularizer that guides the approximation error tending towards the 1-eigensubspace, resulting in a more efficient and stable path of value approximation. Moreover, we theoretically prove the convergence of the ERC method. Besides, theoretical analysis and experiments demonstrate that ERC effectively reduces the variance of value functions. Among 26 tasks in the DMControl benchmark, ERC outperforms state-of-the-art methods for 20. Besides, it shows significant advantages in Q-value approximation and variance reduction. Our code is available at https://sites.google.com/view/erc-ecml23/.

Abstract: Molecular property is usually observed with a limited number of samples, and researchers have considered property prediction as a few-shot problem. One important fact that has been ignored by prior works is that each molecule can be recorded with several different properties simultaneously. To effectively utilize many-to-many correlations of molecules and properties, we propose a Graph Sampling-based Meta-learning (GS-Meta) framework for few-shot molecular property prediction. First, we construct a Molecule-Property relation Graph (MPG): molecule and properties are nodes, while property labels decide edges. Then, to utilize the topological information of MPG, we reformulate an episode in meta-learning as a subgraph of the MPG, containing a target property node, molecule nodes, and auxiliary property nodes. Third, as episodes in the form of subgraphs are no longer independent of each other, we propose to schedule the subgraph sampling process with a contrastive loss function, which considers the consistency and discrimination of subgraphs. Extensive experiments on 5 commonly-used benchmarks show GS-Meta consistently outperforms state-of-the-art methods by 5.71%-6.93% in ROC-AUC and verify the effectiveness of each proposed module. Our code is available at https://github.com/HICAI-ZJU/GS-Meta.

Abstract: Neural networks can be significantly compressed by pruning, leading to sparse models requiring considerably less storage and floating-point operations while maintaining predictive performance. Model soups (Wortsman et al., 2022) improve generalization and out-of-distribution performance by averaging the parameters of multiple models into a single one without increased inference time. However, identifying models in the same loss basin to leverage both sparsity and parameter averaging is challenging, as averaging arbitrary sparse models reduces the overall sparsity due to differing sparse connectivities. In this work, we address these challenges by demonstrating that exploring a single retraining phase of Iterative Magnitude Pruning (IMP) with varying hyperparameter configurations, such as batch ordering or weight decay, produces models that are suitable for averaging and share the same sparse connectivity by design. Averaging these models significantly enhances generalization performance compared to their individual components. Building on this idea, we introduce Sparse Model Soups (SMS), a novel method for merging sparse models by initiating each prune-retrain cycle with the averaged model of the previous phase. SMS maintains sparsity, exploits sparse network benefits being modular and fully parallelizable, and substantially improves IMP's performance. Additionally, we demonstrate that SMS can be adapted to enhance the performance of state-of-the-art pruning during training approaches.

Abstract: To make reinforcement learning more sample efficient, we need better credit assignment methods that measure an action's influence on future rewards. Building upon Hindsight Credit Assignment (HCA), we introduce Counterfactual Contribution Analysis (COCOA), a new family of model-based credit assignment algorithms. Our algorithms achieve precise credit assignment by measuring the contribution of actions upon obtaining subsequent rewards, by quantifying a counterfactual query: "Would the agent still have reached this reward if it had taken another action?". We show that measuring contributions w.r.t. rewarding states, as is done in HCA, results in spurious estimates of contributions, causing HCA to degrade towards the high-variance REINFORCE estimator in many relevant environments. Instead, we measure contributions w.r.t. rewards or learned representations of the rewarding objects, resulting in gradient estimates with lower variance. We run experiments on a suite of problems specifically designed to evaluate long-term credit assignment capabilities. By using dynamic programming, we measure ground-truth policy gradients and show that the improved performance of our new model-based credit assignment methods is due to lower bias and variance compared to HCA and common baselines. Our results demonstrate how modeling action contributions towards rewarding outcomes can be leveraged for credit assignment, opening a new path towards sample-efficient reinforcement learning.

Abstract: Neural networks are very effective when trained on large datasets for a large number of iterations. However, when they are trained on non-stationary streams of data and in an online fashion, their performance is reduced (1) by the online setup, which limits the availability of data, (2) due to catastrophic forgetting because of the non-stationary nature of the data. Furthermore, several recent works (Caccia et al., 2022; Lange et al., 2023) arXiv:2205.1345(2) showed that replay methods used in continual learning suffer from the stability gap, encountered when evaluating the model continually (rather than only on task boundaries). In this article, we study the effect of model ensembling as a way to improve performance and stability in online continual learning. We notice that naively ensembling models coming from a variety of training tasks increases the performance in online continual learning considerably. Starting from this observation, and drawing inspirations from semi-supervised learning ensembling methods, we use a lightweight temporal ensemble that computes the exponential moving average of the weights (EMA) at test time, and show that it can drastically increase the performance and stability when used in combination with several methods from the literature.

Abstract: Inpatient length of stay (LoS) is an important managerial metric which if known in advance can be used to efficiently plan admissions, allocate resources and improve care. Using historical patient data and machine learning techniques, LoS prediction models can be developed. Ethically, these models can not be used for patient discharge in lieu of unit heads but are of utmost necessity for hospital management systems in charge of effective hospital planning. Therefore, the design of the prediction system should be adapted to work in a true hospital setting. In this study, we predict early hospital LoS at the granular level of admission units by applying domain adaptation to leverage information learned from a potential source domain. Time-varying data from 110,079 and 60,492 patient stays to 8 and 9 intensive care units were respectively extracted from eICU-CRD and MIMIC-IV. These were fed into a Long-Short Term Memory and a Fully connected network to train a source domain model, the weights of which were transferred either partially or fully to initiate training in target domains. Shapley Additive exPlanations (SHAP) algorithms were used to study the effect of weight transfer on model explanability. Compared to the benchmark, the proposed weight transfer model showed statistically significant gains in prediction accuracy (between 1% and 5%) as well as computation time (up to 2hrs) for some target domains. The proposed method thus provides an adapted clinical decision support system for hospital management that can ease processes of data access via ethical committee, computation infrastructures and time.

Abstract: Over recent years, denoising diffusion generative models have come to be considered as state-of-the-art methods for synthetic data generation, especially in the case of generating images. These approaches have also proved successful in other applications such as tabular and graph data generation. However, due to computational complexity, to this date, the application of these techniques to graph data has been restricted to small graphs, such as those used in molecular modeling. In this paper, we propose SaGess, a discrete denoising diffusion approach, which is able to generate large real-world networks by augmenting a diffusion model (DiGress) with a generalized divide-and-conquer framework. The algorithm is capable of generating larger graphs by sampling a covering of subgraphs of the initial graph in order to train DiGress. SaGess then constructs a synthetic graph using the subgraphs that have been generated by DiGress. We evaluate the quality of the synthetic data sets against several competitor methods by comparing graph statistics between the original and synthetic samples, as well as evaluating the utility of the synthetic data set produced by using it to train a task-driven model, namely link prediction. In our experiments, SaGess, outperforms most of the one-shot state-of-the-art graph generating methods by a significant factor, both on the graph metrics and on the link prediction task.

Abstract: We introduce a probability distribution, combined with an efficient sampling algorithm, for weights and biases of fully-connected neural networks. In a supervised learning context, no iterative optimization or gradient computations of internal network parameters are needed to obtain a trained network. The sampling is based on the idea of random feature models. However, instead of a data-agnostic distribution, e.g., a normal distribution, we use both the input and the output training data of the supervised learning problem to sample both shallow and deep networks. We prove that the sampled networks we construct are universal approximators. We also show that our sampling scheme is invariant to rigid body transformations and scaling of the input data. This implies many popular pre-processing techniques are no longer required. For Barron functions, we show that the $L^2$-approximation error of sampled shallow networks decreases with the square root of the number of neurons. In numerical experiments, we demonstrate that sampled networks achieve comparable accuracy as iteratively trained ones, but can be constructed orders of magnitude faster. Our test cases involve a classification benchmark from OpenML, sampling of neural operators to represent maps in function spaces, and transfer learning using well-known architectures.

Abstract: The development of very large-scale integration (VLSI) technology has posed new challenges for electronic design automation (EDA) techniques in chip floorplanning. During this process, macro placement is an important subproblem, which tries to determine the positions of all macros with the aim of minimizing half-perimeter wirelength (HPWL) and avoiding overlapping. Previous methods include packing-based, analytical and reinforcement learning methods. In this paper, we propose a new black-box optimization (BBO) framework (called WireMask-BBO) for macro placement, by using a wire-mask-guided greedy procedure for objective evaluation. Equipped with different BBO algorithms, WireMask-BBO empirically achieves significant improvements over previous methods, i.e., achieves significantly shorter HPWL by using much less time. Furthermore, it can fine-tune existing placements by treating them as initial solutions, which can bring up to 50% improvement in HPWL. WireMask-BBO has the potential to significantly improve the quality and efficiency of chip floorplanning, which makes it appealing to researchers and practitioners in EDA and will also promote the application of BBO.

Abstract: We examine the assumption that the hidden-state vectors of recurrent neural networks (RNNs) tend to form clusters of semantically similar vectors, which we dub the clustering hypothesis. While this hypothesis has been assumed in the analysis of RNNs in recent years, its validity has not been studied thoroughly on modern neural network architectures. We examine the clustering hypothesis in the context of RNNs that were trained to recognize regular languages. This enables us to draw on perfect ground-truth automata in our evaluation, against which we can compare the RNN's accuracy and the distribution of the hidden-state vectors. We start with examining the (piecewise linear) separability of an RNN's hidden-state vectors into semantically different classes. We continue the analysis by computing clusters over the hidden-state vector space with multiple state-of-the-art unsupervised clustering approaches. We formally analyze the accuracy of computed clustering functions and the validity of the clustering hypothesis by determining whether clusters group semantically similar vectors to the same state in the ground-truth model. Our evaluation supports the validity of the clustering hypothesis in the majority of examined cases. We observed that the hidden-state vectors of well-trained RNNs are separable, and that the unsupervised clustering techniques succeed in finding clusters of similar state vectors.

Abstract: Deep neural networks (DNNs) have become ubiquitous in machine learning, but their energy consumption remains a notable issue. Lowering the supply voltage is an effective strategy for reducing energy consumption. However, aggressively scaling down the supply voltage can lead to accuracy degradation due to random bit flips in static random access memory (SRAM) where model parameters are stored. To address this challenge, we introduce NeuralFuse, a novel add-on module that addresses the accuracy-energy tradeoff in low-voltage regimes by learning input transformations to generate error-resistant data representations. NeuralFuse protects DNN accuracy in both nominal and low-voltage scenarios. Moreover, NeuralFuse is easy to implement and can be readily applied to DNNs with limited access, such as non-configurable hardware or remote access to cloud-based APIs. Experimental results demonstrate that, at a 1% bit error rate, NeuralFuse can reduce SRAM memory access energy by up to 24% while improving accuracy by up to 57%. To the best of our knowledge, this is the first model-agnostic approach (i.e., no model retraining) to address low-voltage-induced bit errors. The source code is available at https://github.com/IBM/NeuralFuse.

Abstract: Despite the success of two-stage few-shot classification methods, in the episodic meta-training stage, the model suffers severe overfitting. We hypothesize that it is caused by over-discrimination, i.e., the model learns to over-rely on the superficial features that fit for base class discrimination while suppressing the novel class generalization. To penalize over-discrimination, we introduce knowledge distillation techniques to keep novel generalization knowledge from the teacher model during training. Specifically, we select the teacher model as the one with the best validation accuracy during meta-training and restrict the symmetric Kullback-Leibler (SKL) divergence between the output distribution of the linear classifier of the teacher model and that of the student model. This simple approach outperforms the standard meta-training process. We further propose the Nearest Neighbor Symmetric Kullback-Leibler (NNSKL) divergence for meta-training to push the limits of knowledge distillation techniques. NNSKL takes few-shot tasks as input and penalizes the output of the nearest neighbor classifier, which possesses an impact on the relationships between query embedding and support centers. By combining SKL and NNSKL in meta-training, the model achieves even better performance and surpasses state-of-the-art results on several benchmarks.

Abstract: Purpose: The development of machine learning models for surgical workflow and instrument recognition from temporal data represents a challenging task due to the complex nature of surgical workflows. In particular, the imbalanced distribution of data is one of the major challenges in the domain of surgical workflow recognition. In order to obtain meaningful results, careful partitioning of data into training, validation, and test sets, as well as the selection of suitable evaluation metrics are crucial. Methods: In this work, we present an openly available web-based application that enables interactive exploration of dataset partitions. The proposed visual framework facilitates the assessment of dataset splits for surgical workflow recognition, especially with regard to identifying sub-optimal dataset splits. Currently, it supports visualization of surgical phase and instrument annotations. Results: In order to validate the dedicated interactive visualizations, we use a dataset split of the Cholec80 dataset. This dataset split was specifically selected to reflect a case of strong data imbalance. Using our software, we were able to identify phases, phase transitions, and combinations of surgical instruments that were not represented in one of the sets. Conclusion: In order to obtain meaningful results in highly unbalanced class distributions, special care should be taken with respect to the selection of an appropriate split. Interactive data visualization represents a promising approach for the assessment of machine learning datasets. The source code is available at https://github.com/Cardio-AI/endovis-ml

Abstract: Feature transformation aims to reconstruct an effective representation space by mathematically refining the existing features. It serves as a pivotal approach to combat the curse of dimensionality, enhance model generalization, mitigate data sparsity, and extend the applicability of classical models. Existing research predominantly focuses on domain knowledge-based feature engineering or learning latent representations. However, these methods, while insightful, lack full automation and fail to yield a traceable and optimal representation space. An indispensable question arises: Can we concurrently address these limitations when reconstructing a feature space for a machine-learning task? Our initial work took a pioneering step towards this challenge by introducing a novel self-optimizing framework. This framework leverages the power of three cascading reinforced agents to automatically select candidate features and operations for generating improved feature transformation combinations. Despite the impressive strides made, there was room for enhancing its effectiveness and generalization capability. In this extended journal version, we advance our initial work from two distinct yet interconnected perspectives: 1) We propose a refinement of the original framework, which integrates a graph-based state representation method to capture the feature interactions more effectively and develop different Q-learning strategies to alleviate Q-value overestimation further. 2) We utilize a new optimization technique (actor-critic) to train the entire self-optimizing framework in order to accelerate the model convergence and improve the feature transformation performance. Finally, to validate the improved effectiveness and generalization capability of our framework, we perform extensive experiments and conduct comprehensive analyses.

Abstract: Optimizing a machine learning pipeline for a task at hand requires careful configuration of various hyperparameters, typically supported by an AutoML system that optimizes the hyperparameters for the given training dataset. Yet, depending on the AutoML system's own second-order meta-configuration, the performance of the AutoML process can vary significantly. Current AutoML systems cannot automatically adapt their own configuration to a specific use case. Further, they cannot compile user-defined application constraints on the effectiveness and efficiency of the pipeline and its generation. In this paper, we propose Caml, which uses meta-learning to automatically adapt its own AutoML parameters, such as the search strategy, the validation strategy, and the search space, for a task at hand. The dynamic AutoML strategy of Caml takes user-defined constraints into account and obtains constraint-satisfying pipelines with high predictive performance.

Abstract: We introduce ordered transfer hyperparameter optimisation (OTHPO), a version of transfer learning for hyperparameter optimisation (HPO) where the tasks follow a sequential order. Unlike for state-of-the-art transfer HPO, the assumption is that each task is most correlated to those immediately before it. This matches many deployed settings, where hyperparameters are retuned as more data is collected; for instance tuning a sequence of movie recommendation systems as more movies and ratings are added. We propose a formal definition, outline the differences to related problems and propose a basic OTHPO method that outperforms state-of-the-art transfer HPO. We empirically show the importance of taking order into account using ten benchmarks. The benchmarks are in the setting of gradually accumulating data, and span XGBoost, random forest, approximate k-nearest neighbor, elastic net, support vector machines and a separate real-world motivated optimisation problem. We open source the benchmarks to foster future research on ordered transfer HPO.

Abstract: Experimental results have shown that curriculum learning, i.e., presenting simpler examples before more complex ones, can improve the efficiency of learning. Some recent theoretical results also showed that changing the sampling distribution can help neural networks learn parities, with formal results only for large learning rates and one-step arguments. Here we show a separation result in the number of training steps with standard (bounded) learning rates on a common sample distribution: if the data distribution is a mixture of sparse and dense inputs, there exists a regime in which a 2-layer ReLU neural network trained by a curriculum noisy-GD (or SGD) algorithm that uses sparse examples first, can learn parities of sufficiently large degree, while any fully connected neural network of possibly larger width or depth trained by noisy-GD on the unordered samples cannot learn without additional steps. We also provide experimental results supporting the qualitative separation beyond the specific regime of the theoretical results.

Abstract: Invariance to spatial transformations such as translations and rotations is a desirable property and a basic design principle for classification neural networks. However, the commonly used convolutional neural networks (CNNs) are actually very sensitive to even small translations. There exist vast works to achieve exact or approximate transformation invariance by designing transformation-invariant models or assessing the transformations. These works usually make changes to the standard CNNs and harm the performance on standard datasets. In this paper, rather than modifying the classifier, we propose a pre-classifier restorer to recover translated (or even rotated) inputs to the original ones which will be fed into any classifier for the same dataset. The restorer is based on a theoretical result which gives a sufficient and necessary condition for an affine operator to be translational equivariant on a tensor space.

Abstract: High-order Graph Neural Networks (HO-GNNs) have been developed to infer consistent latent spaces in the heterophilic regime, where the label distribution is not correlated with the graph structure. However, most of the existing HO-GNNs are hop-based, i.e., they rely on the powers of the transition matrix. As a result, these architectures are not fully reactive to the classification loss and the achieved structural filters have static supports. In other words, neither the filters' supports nor their coefficients can be learned with these networks. They are confined, instead, to learn combinations of filters. To address the above concerns, we propose Diffusion-jump GNNs a method relying on asymptotic diffusion distances that operates on jumps. A diffusion-pump generates pairwise distances whose projections determine both the support and coefficients of each structural filter. These filters are called jumps because they explore a wide range of scales in order to find bonds between scattered nodes with the same label. Actually, the full process is controlled by the classification loss. Both the jumps and the diffusion distances react to classification errors (i.e. they are learnable). Homophiliation, i.e., the process of learning piecewise smooth latent spaces in the heterophilic regime, is formulated as a Dirichlet problem: the known labels determine the border nodes and the diffusion-pump ensures a minimal deviation of the semi-supervised grouping from a canonical unsupervised grouping. This triggers the update of both the diffusion distances and, consequently, the jumps in order to minimize the classification error. The Dirichlet formulation has several advantages. It leads to the definition of structural heterophily, a novel measure beyond edge heterophily. It also allows us to investigate links with (learnable) diffusion distances, absorbing random walks and stochastic diffusion.

Abstract: Regularization is a set of techniques that are used to improve the generalization ability of deep neural networks. In this paper, we introduce weight compander (WC), a novel effective method to improve generalization by reparameterizing each weight in deep neural networks using a nonlinear function. It is a general, intuitive, cheap and easy to implement method, which can be combined with various other regularization techniques. Large weights in deep neural networks are a sign of a more complex network that is overfitted to the training data. Moreover, regularized networks tend to have a greater range of weights around zero with fewer weights centered at zero. We introduce a weight reparameterization function which is applied to each weight and implicitly reduces overfitting by restricting the magnitude of the weights while forcing them away from zero at the same time. This leads to a more democratic decision-making in the network. Firstly, individual weights cannot have too much influence in the prediction process due to the restriction of their magnitude. Secondly, more weights are used in the prediction process, since they are forced away from zero during the training. This promotes the extraction of more features from the input data and increases the level of weight redundancy, which makes the network less sensitive to statistical differences between training and test data. We extend our method to learn the hyperparameters of the introduced weight reparameterization function. This avoids hyperparameter search and gives the network the opportunity to align the weight reparameterization with the training progress. We show experimentally that using weight compander in addition to standard regularization methods improves the performance of neural networks.

Abstract: Compositionality is a critical aspect of scalable system design. Reinforcement learning (RL) has recently shown substantial success in task learning, but has only recently begun to truly leverage composition. In this paper, we focus on Boolean composition of learned tasks as opposed to functional or sequential composition. Existing Boolean composition for RL focuses on reaching a satisfying absorbing state in environments with discrete action spaces, but does not support composable safety (i.e., avoidance) constraints. We advance the state of the art in Boolean composition of learned tasks with three contributions: i) introduce two distinct notions of safety in this framework; ii) show how to enforce either safety semantics, prove correctness (under some assumptions), and analyze the trade-offs between the two safety notions; and iii) extend Boolean composition from discrete action spaces to continuous action spaces. We demonstrate these techniques using modified versions of value iteration in a grid world, Deep Q-Network (DQN) in a grid world with image observations, and Twin Delayed DDPG (TD3) in a continuous-observation and continuous-action Bullet physics environment. We believe that these contributions advance the theory of safe reinforcement learning by allowing zero-shot composition of policies satisfying safety properties.

Abstract: Evolutionary differential equation discovery proved to be a tool to obtain equations with less a priori assumptions than conventional approaches, such as sparse symbolic regression over the complete possible terms library. The equation discovery field contains two independent directions. The first one is purely mathematical and concerns differentiation, the object of optimization and its relation to the functional spaces and others. The second one is dedicated purely to the optimizational problem statement. Both topics are worth investigating to improve the algorithm's ability to handle experimental data a more artificial intelligence way, without significant pre-processing and a priori knowledge of their nature. In the paper, we consider the prevalence of either single-objective optimization, which considers only the discrepancy between selected terms in the equation, or multi-objective optimization, which additionally takes into account the complexity of the obtained equation. The proposed comparison approach is shown on classical model examples -- Burgers equation, wave equation, and Korteweg - de Vries equation.

Abstract: Many applications, e.g., in shared mobility, require coordinating a large number of agents. Mean-field reinforcement learning addresses the resulting scalability challenge by optimizing the policy of a representative agent. In this paper, we address an important generalization where there exist global constraints on the distribution of agents (e.g., requiring capacity constraints or minimum coverage requirements to be met). We propose Safe-$\text{M}^3$-UCRL, the first model-based algorithm that attains safe policies even in the case of unknown transition dynamics. As a key ingredient, it uses epistemic uncertainty in the transition model within a log-barrier approach to ensure pessimistic constraints satisfaction with high probability. We showcase Safe-$\text{M}^3$-UCRL on the vehicle repositioning problem faced by many shared mobility operators and evaluate its performance through simulations built on Shenzhen taxi trajectory data. Our algorithm effectively meets the demand in critical areas while ensuring service accessibility in regions with low demand.

Abstract: In recent years, channel state information (CSI) at sub-6 GHz has been widely exploited for Wi-Fi sensing, particularly for activity and gesture recognition. In this work, we instead explore mmWave (60 GHz) Wi-Fi signals for gesture recognition/pose estimation. Our focus is on the mmWave Wi-Fi signals so that they can be used not only for high data rate communication but also for improved sensing e.g., for extended reality (XR) applications. For this reason, we extract spatial beam signal-to-noise ratios (SNRs) from the periodic beam training employed by IEEE 802.11ad devices. We consider a set of 10 gestures/poses motivated by XR applications. We conduct experiments in two environments and with three people.As a comparison, we also collect CSI from IEEE 802.11ac devices. To extract features from the CSI and the beam SNR, we leverage a deep neural network (DNN). The DNN classifier achieves promising results on the beam SNR task with state-of-the-art 96.7% accuracy in a single environment, even with a limited dataset. We also investigate the robustness of the beam SNR against CSI across different environments. Our experiments reveal that features from the CSI generalize without additional re-training, while those from beam SNRs do not. Therefore, re-training is required in the latter case.

Abstract: Temporal Point Processes (TPPs) serve as the standard mathematical framework for modeling asynchronous event sequences in continuous time. However, classical TPP models are often constrained by strong assumptions, limiting their ability to capture complex real-world event dynamics. To overcome this limitation, researchers have proposed Neural TPPs, which leverage neural network parametrizations to offer more flexible and efficient modeling. While recent studies demonstrate the effectiveness of Neural TPPs, they often lack a unified setup, relying on different baselines, datasets, and experimental configurations. This makes it challenging to identify the key factors driving improvements in predictive accuracy, hindering research progress. To bridge this gap, we present a comprehensive large-scale experimental study that systematically evaluates the predictive accuracy of state-of-the-art neural TPP models. Our study encompasses multiple real-world and synthetic event sequence datasets, following a carefully designed unified setup. We thoroughly investigate the influence of major architectural components such as event encoding, history encoder, and decoder parametrization on both time and mark prediction tasks. Additionally, we delve into the less explored area of probabilistic calibration for neural TPP models. By analyzing our results, we draw insightful conclusions regarding the significance of history size and the impact of architectural components on predictive accuracy. Furthermore, we shed light on the miscalibration of mark distributions in neural TPP models. Our study aims to provide valuable insights into the performance and characteristics of neural TPP models, contributing to a better understanding of their strengths and limitations.

Abstract: Large Language Models (LLMs) have been successfully used in many natural-language tasks and applications including text generation and AI chatbots. They also are a promising new technology for concept-oriented deep learning (CODL). However, the prerequisite is that LLMs understand concepts and ensure conceptual consistency. We discuss these in this paper, as well as major uses of LLMs for CODL including concept extraction from text, concept graph extraction from text, and concept learning. Human knowledge consists of both symbolic (conceptual) knowledge and embodied (sensory) knowledge. Text-only LLMs, however, can represent only symbolic (conceptual) knowledge. Multimodal LLMs, on the other hand, are capable of representing the full range (conceptual and sensory) of human knowledge. We discuss conceptual understanding in visual-language LLMs, the most important multimodal LLMs, and major uses of them for CODL including concept extraction from image, concept graph extraction from image, and concept learning. While uses of LLMs for CODL are valuable standalone, they are particularly valuable as part of LLM applications such as AI chatbots.

Abstract: Graph neural networks (GNNs) have been widely applied in multi-variate time-series forecasting (MTSF) tasks because of their capability in capturing the correlations among different time-series. These graph-based learning approaches improve the forecasting performance by discovering and understanding the underlying graph structures, which represent the data correlation. When the explicit prior graph structures are not available, most existing works cannot guarantee the sparsity of the generated graphs that make the overall model computational expensive and less interpretable. In this work, we propose a decoupled training method, which includes a graph generating module and a GNNs forecasting module. First, we use Graphical Lasso (or GraphLASSO) to directly exploit the sparsity pattern from data to build graph structures in both static and time-varying cases. Second, we fit these graph structures and the input data into a Graph Convolutional Recurrent Network (GCRN) to train a forecasting model. The experimental results on three real-world datasets show that our novel approach has competitive performance against existing state-of-the-art forecasting algorithms while providing sparse, meaningful and explainable graph structures and reducing training time by approximately 40%. Our PyTorch implementation is publicly available at https://github.com/HySonLab/GraphLASSO

Abstract: We introduce RL4CO, an extensive reinforcement learning (RL) for combinatorial optimization (CO) benchmark. RL4CO employs state-of-the-art software libraries as well as best practices in implementation, such as modularity and configuration management, to be efficient and easily modifiable by researchers for adaptations of neural network architecture, environments, and algorithms. Contrary to the existing focus on specific tasks like the traveling salesman problem (TSP) for performance assessment, we underline the importance of scalability and generalization capabilities for diverse optimization tasks. We also systematically benchmark sample efficiency, zero-shot generalization, and adaptability to changes in data distributions of various models. Our experiments show that some recent state-of-the-art methods fall behind their predecessors when evaluated using these new metrics, suggesting the necessity for a more balanced view of the performance of neural CO solvers. We hope RL4CO will encourage the exploration of novel solutions to complex real-world tasks, allowing to compare with existing methods through a standardized interface that decouples the science from the software engineering. We make our library publicly available at https://github.com/kaist-silab/rl4co.

Abstract: Recent work has observed an intriguing ''Neural Collapse'' phenomenon in well-trained neural networks, where the last-layer representations of training samples with the same label collapse into each other. This appears to suggest that the last-layer representations are completely determined by the labels, and do not depend on the intrinsic structure of input distribution. We provide evidence that this is not a complete description, and that the apparent collapse hides important fine-grained structure in the representations. Specifically, even when representations apparently collapse, the small amount of remaining variation can still faithfully and accurately captures the intrinsic structure of input distribution. As an example, if we train on CIFAR-10 using only 5 coarse-grained labels (by combining two classes into one super-class) until convergence, we can reconstruct the original 10-class labels from the learned representations via unsupervised clustering. The reconstructed labels achieve $93\%$ accuracy on the CIFAR-10 test set, nearly matching the normal CIFAR-10 accuracy for the same architecture. We also provide an initial theoretical result showing the fine-grained representation structure in a simplified synthetic setting. Our results show concretely how the structure of input data can play a significant role in determining the fine-grained structure of neural representations, going beyond what Neural Collapse predicts.

Abstract: There has been an explosion in interest in machine learning (ML) in recent years due to its applications to science and engineering. However, as ML techniques have advanced, tools for explaining and visualizing novel ML algorithms have lagged behind. Animation has been shown to be a powerful tool for making engaging visualizations of systems that dynamically change over time, which makes it well suited to the task of communicating ML algorithms. However, the current approach to animating ML algorithms is to handcraft applications that highlight specific algorithms or use complex generalized animation software. We developed ManimML, an open-source Python library for easily generating animations of ML algorithms directly from code. We sought to leverage ML practitioners' preexisting knowledge of programming rather than requiring them to learn complex animation software. ManimML has a familiar syntax for specifying neural networks that mimics popular deep learning frameworks like Pytorch. A user can take a preexisting neural network architecture and easily write a specification for an animation in ManimML, which will then automatically compose animations for different components of the system into a final animation of the entire neural network. ManimML is open source and available at https://github.com/helblazer811/ManimML.

Abstract: Using machine learning models to generate synthetic data has become common in many fields. Technology to generate synthetic transactions that can be used to detect fraud is also growing fast. Generally, this synthetic data contains only information about the transaction, such as the time, place, and amount of money. It does not usually contain the individual user's characteristics (age and gender are occasionally included). Using relatively complex synthetic demographic data may improve the complexity of transaction data features, thus improving the fraud detection performance. Benefiting from developments of machine learning, some deep learning models have potential to perform better than other well-established synthetic data generation methods, such as microsimulation. In this study, we built a deep-learning Generative Adversarial Network (GAN), called DGGAN, which will be used for demographic data generation. Our model generates samples during model training, which we found important to overcame class imbalance issues. This study can help improve the cognition of synthetic data and further explore the application of synthetic data generation in card fraud detection.