Machine Learning (cs.LG)

Abstract: Genes are fundamental for analyzing biological systems and many recent works proposed to utilize gene expression for various biological tasks by deep learning models. Despite their promising performance, it is hard for deep neural networks to provide biological insights for humans due to their black-box nature. Recently, some works integrated biological knowledge with neural networks to improve the transparency and performance of their models. However, these methods can only incorporate partial biological knowledge, leading to suboptimal performance. In this paper, we propose the Biological Factor Regulatory Neural Network (BFReg-NN), a generic framework to model relations among biological factors in cell systems. BFReg-NN starts from gene expression data and is capable of merging most existing biological knowledge into the model, including the regulatory relations among genes or proteins (e.g., gene regulatory networks (GRN), protein-protein interaction networks (PPI)) and the hierarchical relations among genes, proteins and pathways (e.g., several genes/proteins are contained in a pathway). Moreover, BFReg-NN also has the ability to provide new biologically meaningful insights because of its white-box characteristics. Experimental results on different gene expression-based tasks verify the superiority of BFReg-NN compared with baselines. Our case studies also show that the key insights found by BFReg-NN are consistent with the biological literature.

Abstract: Graph Neural Networks (GNNs) have been widely applied on a variety of real-world applications, such as social recommendation. However, existing GNN-based models on social recommendation suffer from serious problems of generalization and oversmoothness, because of the underexplored negative sampling method and the direct implanting of the off-the-shelf GNN models. In this paper, we propose a succinct multi-network GNN-based neural model (NeMo) for social recommendation. Compared with the existing methods, the proposed model explores a generative negative sampling strategy, and leverages both the positive and negative user-item interactions for users' interest propagation. The experiments show that NeMo outperforms the state-of-the-art baselines on various real-world benchmark datasets (e.g., by up to 38.8% in terms of NDCG@15).

Abstract: How to behave efficiently and flexibly is a central problem for understanding biological agents and creating intelligent embodied AI. It has been well known that behavior can be classified as two types: reward-maximizing habitual behavior, which is fast while inflexible; and goal-directed behavior, which is flexible while slow. Conventionally, habitual and goal-directed behaviors are considered handled by two distinct systems in the brain. Here, we propose to bridge the gap between the two behaviors, drawing on the principles of variational Bayesian theory. We incorporate both behaviors in one framework by introducing a Bayesian latent variable called "intention". The habitual behavior is generated by using prior distribution of intention, which is goal-less; and the goal-directed behavior is generated by the posterior distribution of intention, which is conditioned on the goal. Building on this idea, we present a novel Bayesian framework for modeling behaviors. Our proposed framework enables skill sharing between the two kinds of behaviors, and by leveraging the idea of predictive coding, it enables an agent to seamlessly generalize from habitual to goal-directed behavior without requiring additional training. The proposed framework suggests a fresh perspective for cognitive science and embodied AI, highlighting the potential for greater integration between habitual and goal-directed behaviors.

Abstract: Machine Learning models are increasingly used for decision making, in particular in high-stakes applications such as credit scoring, medicine or recidivism prediction. However, there are growing concerns about these models with respect to their lack of interpretability and the undesirable biases they can generate or reproduce. While the concepts of interpretability and fairness have been extensively studied by the scientific community in recent years, few works have tackled the general multi-class classification problem under fairness constraints, and none of them proposes to generate fair and interpretable models for multi-class classification. In this paper, we use Mixed-Integer Linear Programming (MILP) techniques to produce inherently interpretable scoring systems under sparsity and fairness constraints, for the general multi-class classification setup. Our work generalizes the SLIM (Supersparse Linear Integer Models) framework that was proposed by Rudin and Ustun to learn optimal scoring systems for binary classification. The use of MILP techniques allows for an easy integration of diverse operational constraints (such as, but not restricted to, fairness or sparsity), but also for the building of certifiably optimal models (or sub-optimal models with bounded optimality gap).

Abstract: Graph representation learning aims to effectively encode high-dimensional sparse graph-structured data into low-dimensional dense vectors, which is a fundamental task that has been widely studied in a range of fields, including machine learning and data mining. Classic graph embedding methods follow the basic idea that the embedding vectors of interconnected nodes in the graph can still maintain a relatively close distance, thereby preserving the structural information between the nodes in the graph. However, this is sub-optimal due to: (i) traditional methods have limited model capacity which limits the learning performance; (ii) existing techniques typically rely on unsupervised learning strategies and fail to couple with the latest learning paradigms; (iii) representation learning and downstream tasks are dependent on each other which should be jointly enhanced. With the remarkable success of deep learning, deep graph representation learning has shown great potential and advantages over shallow (traditional) methods, there exist a large number of deep graph representation learning techniques have been proposed in the past decade, especially graph neural networks. In this survey, we conduct a comprehensive survey on current deep graph representation learning algorithms by proposing a new taxonomy of existing state-of-the-art literature. Specifically, we systematically summarize the essential components of graph representation learning and categorize existing approaches by the ways of graph neural network architectures and the most recent advanced learning paradigms. Moreover, this survey also provides the practical and promising applications of deep graph representation learning. Last but not least, we state new perspectives and suggest challenging directions which deserve further investigations in the future.

Abstract: Learning unbiased node representations for imbalanced samples in the graph has become a more remarkable and important topic. For the graph, a significant challenge is that the topological properties of the nodes (e.g., locations, roles) are unbalanced (topology-imbalance), other than the number of training labeled nodes (quantity-imbalance). Existing studies on topology-imbalance focus on the location or the local neighborhood structure of nodes, ignoring the global underlying hierarchical properties of the graph, i.e., hierarchy. In the real-world scenario, the hierarchical structure of graph data reveals important topological properties of graphs and is relevant to a wide range of applications. We find that training labeled nodes with different hierarchical properties have a significant impact on the node classification tasks and confirm it in our experiments. It is well known that hyperbolic geometry has a unique advantage in representing the hierarchical structure of graphs. Therefore, we attempt to explore the hierarchy-imbalance issue for node classification of graph neural networks with a novelty perspective of hyperbolic geometry, including its characteristics and causes. Then, we propose a novel hyperbolic geometric hierarchy-imbalance learning framework, named HyperIMBA, to alleviate the hierarchy-imbalance issue caused by uneven hierarchy-levels and cross-hierarchy connectivity patterns of labeled nodes.Extensive experimental results demonstrate the superior effectiveness of HyperIMBA for hierarchy-imbalance node classification tasks.

Abstract: The most relevant problems in discounted reinforcement learning involve estimating the mean of a function under the stationary distribution of a Markov reward process, such as the expected return in policy evaluation, or the policy gradient in policy optimization. In practice, these estimates are produced through a finite-horizon episodic sampling, which neglects the mixing properties of the Markov process. It is mostly unclear how this mismatch between the practical and the ideal setting affects the estimation, and the literature lacks a formal study on the pitfalls of episodic sampling, and how to do it optimally. In this paper, we present a minimax lower bound on the discounted mean estimation problem that explicitly connects the estimation error with the mixing properties of the Markov process and the discount factor. Then, we provide a statistical analysis on a set of notable estimators and the corresponding sampling procedures, which includes the finite-horizon estimators often used in practice. Crucially, we show that estimating the mean by directly sampling from the discounted kernel of the Markov process brings compelling statistical properties w.r.t. the alternative estimators, as it matches the lower bound without requiring a careful tuning of the episode horizon.

Abstract: Multivariate time series classification (MTSC) is an important data mining task, which can be effectively solved by popular deep learning technology. Unfortunately, the existing deep learning-based methods neglect the hidden dependencies in different dimensions and also rarely consider the unique dynamic features of time series, which lack sufficient feature extraction capability to obtain satisfactory classification accuracy. To address this problem, we propose a novel temporal dynamic graph neural network (TodyNet) that can extract hidden spatio-temporal dependencies without undefined graph structure. It enables information flow among isolated but implicit interdependent variables and captures the associations between different time slots by dynamic graph mechanism, which further improves the classification performance of the model. Meanwhile, the hierarchical representations of graphs cannot be learned due to the limitation of GNNs. Thus, we also design a temporal graph pooling layer to obtain a global graph-level representation for graph learning with learnable temporal parameters. The dynamic graph, graph information propagation, and temporal convolution are jointly learned in an end-to-end framework. The experiments on 26 UEA benchmark datasets illustrate that the proposed TodyNet outperforms existing deep learning-based methods in the MTSC tasks.

Abstract: Accurate state-of-health (SOH) estimation is critical to guarantee the safety, efficiency and reliability of battery-powered applications. Most SOH estimation methods focus on the 0-100\% full state-of-charge (SOC) range that has similar distributions. However, the batteries in real-world applications usually work in the partial SOC range under shallow-cycle conditions and follow different degradation profiles with no labeled data available, thus making SOH estimation challenging. To estimate shallow-cycle battery SOH, a novel unsupervised deep transfer learning method is proposed to bridge different domains using self-attention distillation module and multi-kernel maximum mean discrepancy technique. The proposed method automatically extracts domain-variant features from charge curves to transfer knowledge from the large-scale labeled full cycles to the unlabeled shallow cycles. The CALCE and SNL battery datasets are employed to verify the effectiveness of the proposed method to estimate the battery SOH for different SOC ranges, temperatures, and discharge rates. The proposed method achieves a root-mean-square error within 2\% and outperforms other transfer learning methods for different SOC ranges. When applied to batteries with different operating conditions and from different manufacturers, the proposed method still exhibits superior SOH estimation performance. The proposed method is the first attempt at accurately estimating battery SOH under shallow-cycle conditions without needing a full-cycle characteristic test.

Abstract: We focus on learning composable policies to control a variety of physical agents with possibly different structures. Among state-of-the-art methods, prominent approaches exploit graph-based representations and weight-sharing modular policies based on the message-passing framework. However, as shown by recent literature, message passing can create bottlenecks in information propagation and hinder global coordination. This drawback can become even more problematic in tasks where high-level planning is crucial. In fact, in similar scenarios, each modular policy - e.g., controlling a joint of a robot - would request to coordinate not only for basic locomotion but also achieve high-level goals, such as navigating a maze. A classical solution to avoid similar pitfalls is to resort to hierarchical decision-making. In this work, we adopt the Feudal Reinforcement Learning paradigm to develop agents where control actions are the outcome of a hierarchical (pyramidal) message-passing process. In the proposed Feudal Graph Reinforcement Learning (FGRL) framework, high-level decisions at the top level of the hierarchy are propagated through a layered graph representing a hierarchy of policies. Lower layers mimic the morphology of the physical system and upper layers can capture more abstract sub-modules. The purpose of this preliminary work is to formalize the framework and provide proof-of-concept experiments on benchmark environments (MuJoCo locomotion tasks). Empirical evaluation shows promising results on both standard benchmarks and zero-shot transfer learning settings.

Abstract: These lecture notes provide an overview of Neural Network architectures from a mathematical point of view. Especially, Machine Learning with Neural Networks is seen as an optimization problem. Covered are an introduction to Neural Networks and the following architectures: Feedforward Neural Network, Convolutional Neural Network, ResNet, and Recurrent Neural Network.

Abstract: Federated learning (FL) provides a variety of privacy advantages by allowing clients to collaboratively train a model without sharing their private data. However, recent studies have shown that private information can still be leaked through shared gradients. To further minimize the risk of privacy leakage, existing defenses usually require clients to locally modify their gradients (e.g., differential privacy) prior to sharing with the server. While these approaches are effective in certain cases, they regard the entire data as a single entity to protect, which usually comes at a large cost in model utility. In this paper, we seek to reconcile utility and privacy in FL by proposing a user-configurable privacy defense, RecUP-FL, that can better focus on the user-specified sensitive attributes while obtaining significant improvements in utility over traditional defenses. Moreover, we observe that existing inference attacks often rely on a machine learning model to extract the private information (e.g., attributes). We thus formulate such a privacy defense as an adversarial learning problem, where RecUP-FL generates slight perturbations that can be added to the gradients before sharing to fool adversary models. To improve the transferability to un-queryable black-box adversary models, inspired by the idea of meta-learning, RecUP-FL forms a model zoo containing a set of substitute models and iteratively alternates between simulations of the white-box and the black-box adversarial attack scenarios to generate perturbations. Extensive experiments on four datasets under various adversarial settings (both attribute inference attack and data reconstruction attack) show that RecUP-FL can meet user-specified privacy constraints over the sensitive attributes while significantly improving the model utility compared with state-of-the-art privacy defenses.

Abstract: Spider webs are incredible biological structures, comprising thin but strong silk filament and arranged into complex hierarchical architectures with striking mechanical properties (e.g., lightweight but high strength, achieving diverse mechanical responses). While simple 2D orb webs can easily be mimicked, the modeling and synthesis of 3D-based web structures remain challenging, partly due to the rich set of design features. Here we provide a detailed analysis of the heterogenous graph structures of spider webs, and use deep learning as a way to model and then synthesize artificial, bio-inspired 3D web structures. The generative AI models are conditioned based on key geometric parameters (including average edge length, number of nodes, average node degree, and others). To identify graph construction principles, we use inductive representation sampling of large experimentally determined spider web graphs, to yield a dataset that is used to train three conditional generative models: 1) An analog diffusion model inspired by nonequilibrium thermodynamics, with sparse neighbor representation, 2) a discrete diffusion model with full neighbor representation, and 3) an autoregressive transformer architecture with full neighbor representation. All three models are scalable, produce complex, de novo bio-inspired spider web mimics, and successfully construct graphs that meet the design objectives. We further propose algorithm that assembles web samples produced by the generative models into larger-scale structures based on a series of geometric design targets, including helical and parametric shapes, mimicking, and extending natural design principles towards integration with diverging engineering objectives. Several webs are manufactured using 3D printing and tested to assess mechanical properties.

Abstract: Classifying incomplete multi-view data is inevitable since arbitrary view missing widely exists in real-world applications. Although great progress has been achieved, existing incomplete multi-view methods are still difficult to obtain a trustworthy prediction due to the relatively high uncertainty nature of missing views. First, the missing view is of high uncertainty, and thus it is not reasonable to provide a single deterministic imputation. Second, the quality of the imputed data itself is of high uncertainty. To explore and exploit the uncertainty, we propose an Uncertainty-induced Incomplete Multi-View Data Classification (UIMC) model to classify the incomplete multi-view data under a stable and reliable framework. We construct a distribution and sample multiple times to characterize the uncertainty of missing views, and adaptively utilize them according to the sampling quality. Accordingly, the proposed method realizes more perceivable imputation and controllable fusion. Specifically, we model each missing data with a distribution conditioning on the available views and thus introducing uncertainty. Then an evidence-based fusion strategy is employed to guarantee the trustworthy integration of the imputed views. Extensive experiments are conducted on multiple benchmark data sets and our method establishes a state-of-the-art performance in terms of both performance and trustworthiness.

Abstract: Learning skills by imitation is a promising concept for the intuitive teaching of robots. A common way to learn such skills is to learn a parametric model by maximizing the likelihood given the demonstrations. Yet, human demonstrations are often multi-modal, i.e., the same task is solved in multiple ways which is a major challenge for most imitation learning methods that are based on such a maximum likelihood (ML) objective. The ML objective forces the model to cover all data, it prevents specialization in the context space and can cause mode-averaging in the behavior space, leading to suboptimal or potentially catastrophic behavior. Here, we alleviate those issues by introducing a curriculum using a weight for each data point, allowing the model to specialize on data it can represent while incentivizing it to cover as much data as possible by an entropy bonus. We extend our algorithm to a Mixture of (linear) Experts (MoE) such that the single components can specialize on local context regions, while the MoE covers all data points. We evaluate our approach in complex simulated and real robot control tasks and show it learns from versatile human demonstrations and significantly outperforms current SOTA methods. A reference implementation can be found at https://github.com/intuitive-robots/ml-cur

Abstract: This article presents a novel hybrid approach using statistics and machine learning to forecast the national demand of electricity. As investment and operation of future energy systems require long-term electricity demand forecasts with hourly resolution, our mathematical model fills a gap in energy forecasting. The proposed methodology was constructed using hourly data from Ukraine's electricity consumption ranging from 2013 to 2020. To this end, we analysed the underlying structure of the hourly, daily and yearly time series of electricity consumption. The long-term yearly trend is evaluated using macroeconomic regression analysis. The mid-term model integrates temperature and calendar regressors to describe the underlying structure, and combines ARIMA and LSTM ``black-box'' pattern-based approaches to describe the error term. The short-term model captures the hourly seasonality through calendar regressors and multiple ARMA models for the residual. Results show that the best forecasting model is composed by combining multiple regression models and a LSTM hybrid model for residual prediction. Our hybrid model is very effective at forecasting long-term electricity consumption on an hourly resolution. In two years of out-of-sample forecasts with 17520 timesteps, it is shown to be within 96.83 \% accuracy.

Abstract: Anomaly detection on attributed graphs is a crucial topic for its practical application. Existing methods suffer from semantic mixture and imbalance issue because they mainly focus on anomaly discrimination, ignoring representation learning. It conflicts with the assortativity assumption that anomalous nodes commonly connect with normal nodes directly. Additionally, there are far fewer anomalous nodes than normal nodes, indicating a long-tailed data distribution. To address these challenges, a unique algorithm,Decoupled Self-supervised Learning forAnomalyDetection (DSLAD), is proposed in this paper. DSLAD is a self-supervised method with anomaly discrimination and representation learning decoupled for anomaly detection. DSLAD employs bilinear pooling and masked autoencoder as the anomaly discriminators. By decoupling anomaly discrimination and representation learning, a balanced feature space is constructed, in which nodes are more semantically discriminative, as well as imbalance issue can be resolved. Experiments conducted on various six benchmark datasets reveal the effectiveness of DSLAD.

Abstract: The architecture of a deep neural network is defined explicitly in terms of the number of layers, the width of each layer and the general network topology. Existing optimisation frameworks neglect this information in favour of implicit architectural information (e.g. second-order methods) or architecture-agnostic distance functions (e.g. mirror descent). Meanwhile, the most popular optimiser in practice, Adam, is based on heuristics. This paper builds a new framework for deriving optimisation algorithms that explicitly leverage neural architecture. The theory extends mirror descent to non-convex composite objective functions: the idea is to transform a Bregman divergence to account for the non-linear structure of neural architecture. Working through the details for deep fully-connected networks yields automatic gradient descent: a first-order optimiser without any hyperparameters. Automatic gradient descent trains both fully-connected and convolutional networks out-of-the-box and at ImageNet scale. A PyTorch implementation is available at https://github.com/jxbz/agd and also in Appendix B. Overall, the paper supplies a rigorous theoretical foundation for a next-generation of architecture-dependent optimisers that work automatically and without hyperparameters.

Abstract: Numerous research studies in the field of federated learning (FL) have attempted to use personalization to address the heterogeneity among clients, one of FL's most crucial and challenging problems. However, existing works predominantly focus on tailoring models. Yet, due to the heterogeneity of clients, they may each require different choices of hyperparameters, which have not been studied so far. We pinpoint two challenges of personalized federated hyperparameter optimization (pFedHPO): handling the exponentially increased search space and characterizing each client without compromising its data privacy. To overcome them, we propose learning a \textsc{H}yper\textsc{P}arameter \textsc{N}etwork (HPN) fed with client encoding to decide personalized hyperparameters. The client encoding is calculated with a random projection-based procedure to protect each client's privacy. Besides, we design a novel mechanism to debias the low-fidelity function evaluation samples for learning HPN. We conduct extensive experiments on FL tasks from various domains, demonstrating the superiority of HPN.

Abstract: Tiny machine learning (TinyML) is a rapidly growing field aiming to democratize machine learning (ML) for resource-constrained microcontrollers (MCUs). Given the pervasiveness of these tiny devices, it is inherent to ask whether TinyML applications can benefit from aggregating their knowledge. Federated learning (FL) enables decentralized agents to jointly learn a global model without sharing sensitive local data. However, a common global model may not work for all devices due to the complexity of the actual deployment environment and the heterogeneity of the data available on each device. In addition, the deployment of TinyML hardware has significant computational and communication constraints, which traditional ML fails to address. Considering these challenges, we propose TinyReptile, a simple but efficient algorithm inspired by meta-learning and online learning, to collaboratively learn a solid initialization for a neural network (NN) across tiny devices that can be quickly adapted to a new device with respect to its data. We demonstrate TinyReptile on Raspberry Pi 4 and Cortex-M4 MCU with only 256-KB RAM. The evaluations on various TinyML use cases confirm a resource reduction and training time saving by at least two factors compared with baseline algorithms with comparable performance.

Abstract: Multivariate time series data comprises various channels of variables. The multivariate forecasting models need to capture the relationship between the channels to accurately predict future values. However, recently, there has been an emergence of methods that employ the Channel Independent (CI) strategy. These methods view multivariate time series data as separate univariate time series and disregard the correlation between channels. Surprisingly, our empirical results have shown that models trained with the CI strategy outperform those trained with the Channel Dependent (CD) strategy, usually by a significant margin. Nevertheless, the reasons behind this phenomenon have not yet been thoroughly explored in the literature. This paper provides comprehensive empirical and theoretical analyses of the characteristics of multivariate time series datasets and the CI/CD strategy. Our results conclude that the CD approach has higher capacity but often lacks robustness to accurately predict distributionally drifted time series. In contrast, the CI approach trades capacity for robust prediction. Practical measures inspired by these analyses are proposed to address the capacity and robustness dilemma, including a modified CD method called Predict Residuals with Regularization (PRReg) that can surpass the CI strategy. We hope our findings can raise awareness among researchers about the characteristics of multivariate time series and inspire the construction of better forecasting models.

Abstract: Rule-based surrogate models are an effective and interpretable way to approximate a Deep Neural Network's (DNN) decision boundaries, allowing humans to easily understand deep learning models. Current state-of-the-art decompositional methods, which are those that consider the DNN's latent space to extract more exact rule sets, manage to derive rule sets at high accuracy. However, they a) do not guarantee that the surrogate model has learned from the same variables as the DNN (alignment), b) only allow to optimise for a single objective, such as accuracy, which can result in excessively large rule sets (complexity), and c) use decision tree algorithms as intermediate models, which can result in different explanations for the same DNN (stability). This paper introduces the CGX (Column Generation eXplainer) to address these limitations - a decompositional method using dual linear programming to extract rules from the hidden representations of the DNN. This approach allows to optimise for any number of objectives and empowers users to tweak the explanation model to their needs. We evaluate our results on a wide variety of tasks and show that CGX meets all three criteria, by having exact reproducibility of the explanation model that guarantees stability and reduces the rule set size by >80% (complexity) at equivalent or improved accuracy and fidelity across tasks (alignment).

Abstract: Classic online prediction algorithms, such as Hedge, are inherently unfair by design, as they try to play the most rewarding arm as many times as possible while ignoring the sub-optimal arms to achieve sublinear regret. In this paper, we consider a fair online prediction problem in the adversarial setting with hard lower bounds on the rate of accrual of rewards for all arms. By combining elementary queueing theory with online learning, we propose a new online prediction policy, called BanditQ, that achieves the target rate constraints while achieving a regret of $O(T^{3/4})$ in the full-information setting. The design and analysis of BanditQ involve a novel use of the potential function method and are of independent interest.

Abstract: We study estimation of piecewise smooth signals over a graph. We propose a $\ell_{2,0}$-norm penalized Graph Trend Filtering (GTF) model to estimate piecewise smooth graph signals that exhibits inhomogeneous levels of smoothness across the nodes. We prove that the proposed GTF model is simultaneously a k-means clustering on the signal over the nodes and a minimum graph cut on the edges of the graph, where the clustering and the cut share the same assignment matrix. We propose two methods to solve the proposed GTF model: a spectral decomposition method and a method based on simulated annealing. In the experiment on synthetic and real-world datasets, we show that the proposed GTF model has a better performances compared with existing approaches on the tasks of denoising, support recovery and semi-supervised classification. We also show that the proposed GTF model can be solved more efficiently than existing models for the dataset with a large edge set.

Abstract: Nowadays artificial neural network models achieve remarkable results in many disciplines. Functions mapping the representation provided by the model to the probability distribution are the inseparable aspect of deep learning solutions. Although softmax is a commonly accepted probability mapping function in the machine learning community, it cannot return sparse outputs and always spreads the positive probability to all positions. In this paper, we propose r-softmax, a modification of the softmax, outputting sparse probability distribution with controllable sparsity rate. In contrast to the existing sparse probability mapping functions, we provide an intuitive mechanism for controlling the output sparsity level. We show on several multi-label datasets that r-softmax outperforms other sparse alternatives to softmax and is highly competitive with the original softmax. We also apply r-softmax to the self-attention module of a pre-trained transformer language model and demonstrate that it leads to improved performance when fine-tuning the model on different natural language processing tasks.

Abstract: Despite the vast body of literature on Active Learning (AL), there is no comprehensive and open benchmark allowing for efficient and simple comparison of proposed samplers. Additionally, the variability in experimental settings across the literature makes it difficult to choose a sampling strategy, which is critical due to the one-off nature of AL experiments. To address those limitations, we introduce OpenAL, a flexible and open-source framework to easily run and compare sampling AL strategies on a collection of realistic tasks. The proposed benchmark is augmented with interpretability metrics and statistical analysis methods to understand when and why some samplers outperform others. Last but not least, practitioners can easily extend the benchmark by submitting their own AL samplers.

Abstract: In this paper, we present a transformer architecture for predicting student performance on standardized tests. Specifically, we leverage students historical data, including their past test scores, study habits, and other relevant information, to create a personalized model for each student. We then use these models to predict their future performance on a given test. Applying this model to the RIIID dataset, we demonstrate that using multiple granularities for temporal features as the decoder input significantly improve model performance. Our results also show the effectiveness of our approach, with substantial improvements over the LightGBM method. Our work contributes to the growing field of AI in education, providing a scalable and accurate tool for predicting student outcomes.

Abstract: In Federated Learning, a global model is learned by aggregating model updates computed at a set of independent client nodes, to reduce communication costs multiple gradient steps are performed at each node prior to aggregation. A key challenge in this setting is data heterogeneity across clients resulting in differing local objectives which can lead clients to overly minimize their own local objective, diverging from the global solution. We demonstrate that individual client models experience a catastrophic forgetting with respect to data from other clients and propose an efficient approach that modifies the cross-entropy objective on a per-client basis by re-weighting the softmax logits prior to computing the loss. This approach shields classes outside a client's label set from abrupt representation change and we empirically demonstrate it can alleviate client forgetting and provide consistent improvements to standard federated learning algorithms. Our method is particularly beneficial under the most challenging federated learning settings where data heterogeneity is high and client participation in each round is low.

Abstract: Parameter regularization or allocation methods are effective in overcoming catastrophic forgetting in lifelong learning. However, they solve all tasks in a sequence uniformly and ignore the differences in the learning difficulty of different tasks. So parameter regularization methods face significant forgetting when learning a new task very different from learned tasks, and parameter allocation methods face unnecessary parameter overhead when learning simple tasks. In this paper, we propose the Parameter Allocation & Regularization (PAR), which adaptively select an appropriate strategy for each task from parameter allocation and regularization based on its learning difficulty. A task is easy for a model that has learned tasks related to it and vice versa. We propose a divergence estimation method based on the Nearest-Prototype distance to measure the task relatedness using only features of the new task. Moreover, we propose a time-efficient relatedness-aware sampling-based architecture search strategy to reduce the parameter overhead for allocation. Experimental results on multiple benchmarks demonstrate that, compared with SOTAs, our method is scalable and significantly reduces the model's redundancy while improving the model's performance. Further qualitative analysis indicates that PAR obtains reasonable task-relatedness.

Abstract: Neural Ordinary Differential Equations (NODEs), a framework of continuous-depth neural networks, have been widely applied, showing exceptional efficacy in coping with representative datasets. Recently, an augmented framework has been developed to overcome some limitations that emerged in the application of the original framework. In this paper, we propose a new class of continuous-depth neural networks with delay, named Neural Delay Differential Equations (NDDEs). To compute the corresponding gradients, we use the adjoint sensitivity method to obtain the delayed dynamics of the adjoint. Differential equations with delays are typically seen as dynamical systems of infinite dimension that possess more fruitful dynamics. Compared to NODEs, NDDEs have a stronger capacity of nonlinear representations. We use several illustrative examples to demonstrate this outstanding capacity. Firstly, we successfully model the delayed dynamics where the trajectories in the lower-dimensional phase space could be mutually intersected and even chaotic in a model-free or model-based manner. Traditional NODEs, without any argumentation, are not directly applicable for such modeling. Secondly, we achieve lower loss and higher accuracy not only for the data produced synthetically by complex models but also for the CIFAR10, a well-known image dataset. Our results on the NDDEs demonstrate that appropriately articulating the elements of dynamical systems into the network design is truly beneficial in promoting network performance.

Abstract: Denoising diffusion models are a recent class of generative models which achieve state-of-the-art results in many domains such as unconditional image generation and text-to-speech tasks. They consist of a noising process destroying the data and a backward stage defined as the time-reversal of the noising diffusion. Building on their success, diffusion models have recently been extended to the Riemannian manifold setting. Yet, these Riemannian diffusion models require geodesics to be defined for all times. While this setting encompasses many important applications, it does not include manifolds defined via a set of inequality constraints, which are ubiquitous in many scientific domains such as robotics and protein design. In this work, we introduce two methods to bridge this gap. First, we design a noising process based on the logarithmic barrier metric induced by the inequality constraints. Second, we introduce a noising process based on the reflected Brownian motion. As existing diffusion model techniques cannot be applied in this setting, we derive new tools to define such models in our framework. We empirically demonstrate the applicability of our methods to a number of synthetic and real-world tasks, including the constrained conformational modelling of protein backbones and robotic arms.

Abstract: There is a growing interest in using reinforcement learning (RL) to personalize sequences of treatments in digital health to support users in adopting healthier behaviors. Such sequential decision-making problems involve decisions about when to treat and how to treat based on the user's context (e.g., prior activity level, location, etc.). Online RL is a promising data-driven approach for this problem as it learns based on each user's historical responses and uses that knowledge to personalize these decisions. However, to decide whether the RL algorithm should be included in an ``optimized'' intervention for real-world deployment, we must assess the data evidence indicating that the RL algorithm is actually personalizing the treatments to its users. Due to the stochasticity in the RL algorithm, one may get a false impression that it is learning in certain states and using this learning to provide specific treatments. We use a working definition of personalization and introduce a resampling-based methodology for investigating whether the personalization exhibited by the RL algorithm is an artifact of the RL algorithm stochasticity. We illustrate our methodology with a case study by analyzing the data from a physical activity clinical trial called HeartSteps, which included the use of an online RL algorithm. We demonstrate how our approach enhances data-driven truth-in-advertising of algorithm personalization both across all users as well as within specific users in the study.

Abstract: No free lunch theorems for supervised learning state that no learner can solve all problems or that all learners achieve exactly the same accuracy on average over a uniform distribution on learning problems. Accordingly, these theorems are often referenced in support of the notion that individual problems require specially tailored inductive biases. While virtually all uniformly sampled datasets have high complexity, real-world problems disproportionately generate low-complexity data, and we argue that neural network models share this same preference, formalized using Kolmogorov complexity. Notably, we show that architectures designed for a particular domain, such as computer vision, can compress datasets on a variety of seemingly unrelated domains. Our experiments show that pre-trained and even randomly initialized language models prefer to generate low-complexity sequences. Whereas no free lunch theorems seemingly indicate that individual problems require specialized learners, we explain how tasks that often require human intervention such as picking an appropriately sized model when labeled data is scarce or plentiful can be automated into a single learning algorithm. These observations justify the trend in deep learning of unifying seemingly disparate problems with an increasingly small set of machine learning models.

Abstract: Self-Supervised Learning (SSL) models rely on a pretext task to learn representations. Because this pretext task differs from the downstream tasks used to evaluate the performance of these models, there is an inherent misalignment or pretraining bias. A commonly used trick in SSL, shown to make deep networks more robust to such bias, is the addition of a small projector (usually a 2 or 3 layer multi-layer perceptron) on top of a backbone network during training. In contrast to previous work that studied the impact of the projector architecture, we here focus on a simpler, yet overlooked lever to control the information in the backbone representation. We show that merely changing its dimensionality -- by changing only the size of the backbone's very last block -- is a remarkably effective technique to mitigate the pretraining bias. It significantly improves downstream transfer performance for both Self-Supervised and Supervised pretrained models.