Machine Learning (cs.LG)

Abstract: Explainable AI has become a popular tool for validating machine learning models. Mismatches between the explained model's decision strategy and the user's domain knowledge (e.g. Clever Hans effects) have also been recognized as a starting point for improving faulty models. However, it is less clear what to do when the user and the explanation agree. In this paper, we demonstrate that acceptance of explanations by the user is not a guarantee for a ML model to function well, in particular, some Clever Hans effects may remain undetected. Such hidden flaws of the model can nevertheless be mitigated, and we demonstrate this by contributing a new method, Explanation-Guided Exposure Minimization (EGEM), that premptively prunes variations in the ML model that have not been the subject of positive explanation feedback. Experiments on natural image data demonstrate that our approach leads to models that strongly reduce their reliance on hidden Clever Hans strategies, and consequently achieve higher accuracy on new data.

Abstract: In recent years, Dynamic Graph (DG) representations have been increasingly used for modeling dynamic systems due to their ability to integrate both topological and temporal information in a compact representation. Dynamic graphs allow to efficiently handle applications such as social network prediction, recommender systems, traffic forecasting or electroencephalography analysis, that can not be adressed using standard numeric representations. As a direct consequence of the emergence of dynamic graph representations, dynamic graph learning has emerged as a new machine learning problem, combining challenges from both sequential/temporal data processing and static graph learning. In this research area, Dynamic Graph Neural Network (DGNN) has became the state of the art approach and plethora of models have been proposed in the very recent years. This paper aims at providing a review of problems and models related to dynamic graph learning. The various dynamic graph supervised learning settings are analysed and discussed. We identify the similarities and differences between existing models with respect to the way time information is modeled. Finally, general guidelines for a DGNN designer when faced with a dynamic graph learning problem are provided.

Abstract: As data-driven intelligent systems advance, the need for reliable and transparent decision-making mechanisms has become increasingly important. Therefore, it is essential to integrate uncertainty quantification and model explainability approaches to foster trustworthy business and operational process analytics. This study explores how model uncertainty can be effectively communicated in global and local post-hoc explanation approaches, such as Partial Dependence Plots (PDP) and Individual Conditional Expectation (ICE) plots. In addition, this study examines appropriate visualization analytics approaches to facilitate such methodological integration. By combining these two research directions, decision-makers can not only justify the plausibility of explanation-driven actionable insights but also validate their reliability. Finally, the study includes expert interviews to assess the suitability of the proposed approach and designed interface for a real-world predictive process monitoring problem in the manufacturing domain.

Abstract: This study focuses on long-term forecasting (LTF) on continuous-time dynamic graph networks (CTDGNs), which is important for real-world modeling. Existing CTDGNs are effective for modeling temporal graph data due to their ability to capture complex temporal dependencies but perform poorly on LTF due to the substantial requirement for historical data, which is not practical in most cases. To relieve this problem, a most intuitive way is data augmentation. In this study, we propose \textbf{\underline{U}ncertainty \underline{M}asked \underline{M}ix\underline{U}p (UmmU)}: a plug-and-play module that conducts uncertainty estimation to introduce uncertainty into the embedding of intermediate layer of CTDGNs, and perform masked mixup to further enhance the uncertainty of the embedding to make it generalize to more situations. UmmU can be easily inserted into arbitrary CTDGNs without increasing the number of parameters. We conduct comprehensive experiments on three real-world dynamic graph datasets, the results demonstrate that UmmU can effectively improve the long-term forecasting performance for CTDGNs.

Abstract: We study the geometry of linear networks with one-dimensional convolutional layers. The function spaces of these networks can be identified with semi-algebraic families of polynomials admitting sparse factorizations. We analyze the impact of the network's architecture on the function space's dimension, boundary, and singular points. We also describe the critical points of the network's parameterization map. Furthermore, we study the optimization problem of training a network with the squared error loss. We prove that for architectures where all strides are larger than one and generic data, the non-zero critical points of that optimization problem are smooth interior points of the function space. This property is known to be false for dense linear networks and linear convolutional networks with stride one.

Abstract: Time series classification (TSC) is a challenging task due to the diversity of types of feature that may be relevant for different classification tasks, including trends, variance, frequency, magnitude, and various patterns. To address this challenge, several alternative classes of approach have been developed, including similarity-based, features and intervals, shapelets, dictionary, kernel, neural network, and hybrid approaches. While kernel, neural network, and hybrid approaches perform well overall, some specialized approaches are better suited for specific tasks. In this paper, we propose a new similarity-based classifier, Proximity Forest version 2.0 (PF 2.0), which outperforms previous state-of-the-art similarity-based classifiers across the UCR benchmark and outperforms state-of-the-art kernel, neural network, and hybrid methods on specific datasets in the benchmark that are best addressed by similarity-base methods. PF 2.0 incorporates three recent advances in time series similarity measures -- (1) computationally efficient early abandoning and pruning to speedup elastic similarity computations; (2) a new elastic similarity measure, Amerced Dynamic Time Warping (ADTW); and (3) cost function tuning. It rationalizes the set of similarity measures employed, reducing the eight base measures of the original PF to three and using the first derivative transform with all similarity measures, rather than a limited subset. We have implemented both PF 1.0 and PF 2.0 in a single C++ framework, making the PF framework more efficient.

Abstract: Federated learning (FL) is a popular way of edge computing that doesn't compromise users' privacy. Current FL paradigms assume that data only resides on the edge, while cloud servers only perform model averaging. However, in real-life situations such as recommender systems, the cloud server has the ability to store historical and interactive features. In this paper, our proposed Edge-Cloud Collaborative Knowledge Transfer Framework (ECCT) bridges the gap between the edge and cloud, enabling bi-directional knowledge transfer between both, sharing feature embeddings and prediction logits. ECCT consolidates various benefits, including enhancing personalization, enabling model heterogeneity, tolerating training asynchronization, and relieving communication burdens. Extensive experiments on public and industrial datasets demonstrate ECCT's effectiveness and potential for use in academia and industry.

Abstract: We study the design of sample-efficient algorithms for reinforcement learning in the presence of rich, high-dimensional observations, formalized via the Block MDP problem. Existing algorithms suffer from either 1) computational intractability, 2) strong statistical assumptions that are not necessarily satisfied in practice, or 3) suboptimal sample complexity. We address these issues by providing the first computationally efficient algorithm that attains rate-optimal sample complexity with respect to the desired accuracy level, with minimal statistical assumptions. Our algorithm, MusIK, combines systematic exploration with representation learning based on multi-step inverse kinematics, a learning objective in which the aim is to predict the learner's own action from the current observation and observations in the (potentially distant) future. MusIK is simple and flexible, and can efficiently take advantage of general-purpose function approximation. Our analysis leverages several new techniques tailored to non-optimistic exploration algorithms, which we anticipate will find broader use.

Abstract: Most real-world networks evolve over time. Existing literature proposes models for dynamic networks that are either unlabeled or assumed to have a single membership structure. On the other hand, a new family of Mixed Membership Stochastic Block Models (MMSBM) allows to model static labeled networks under the assumption of mixed-membership clustering. In this work, we propose to extend this later class of models to infer dynamic labeled networks under a mixed membership assumption. Our approach takes the form of a temporal prior on the model's parameters. It relies on the single assumption that dynamics are not abrupt. We show that our method significantly differs from existing approaches, and allows to model more complex systems --dynamic labeled networks. We demonstrate the robustness of our method with several experiments on both synthetic and real-world datasets. A key interest of our approach is that it needs very few training data to yield good results. The performance gain under challenging conditions broadens the variety of possible applications of automated learning tools --as in social sciences, which comprise many fields where small datasets are a major obstacle to the introduction of machine learning methods.

Abstract: Data augmentation forms the cornerstone of many modern machine learning training pipelines; yet, the mechanisms by which it works are not clearly understood. Much of the research on data augmentation (DA) has focused on improving existing techniques, examining its regularization effects in the context of neural network over-fitting, or investigating its impact on features. Here, we undertake a holistic examination of the effect of DA on three different classifiers, convolutional neural networks, support vector machines, and logistic regression models, which are commonly used in supervised classification of imbalanced data. We support our examination with testing on three image and five tabular datasets. Our research indicates that DA, when applied to imbalanced data, produces substantial changes in model weights, support vectors and feature selection; even though it may only yield relatively modest changes to global metrics, such as balanced accuracy or F1 measure. We hypothesize that DA works by facilitating variances in data, so that machine learning models can associate changes in the data with labels. By diversifying the range of feature amplitudes that a model must recognize to predict a label, DA improves a model's capacity to generalize when learning with imbalanced data.

Abstract: Diffusion Models (DMs) are powerful generative models that add Gaussian noise to the data and learn to remove it. We wanted to determine which noise distribution (Gaussian or non-Gaussian) led to better generated data in DMs. Since DMs do not work by design with non-Gaussian noise, we built a framework that allows reversing a diffusion process with non-Gaussian location-scale noise. We use that framework to show that the Gaussian distribution performs the best over a wide range of other distributions (Laplace, Uniform, t, Generalized-Gaussian).

Abstract: Localizing behaviors of neural networks to a subset of the network's components or a subset of interactions between components is a natural first step towards analyzing network mechanisms and possible failure modes. Existing work is often qualitative and ad-hoc, and there is no consensus on the appropriate way to evaluate localization claims. We introduce path patching, a technique for expressing and quantitatively testing a natural class of hypotheses expressing that behaviors are localized to a set of paths. We refine an explanation of induction heads, characterize a behavior of GPT-2, and open source a framework for efficiently running similar experiments.

Abstract: A new approach called NAF (the Neural Attention Forest) for solving regression and classification tasks under tabular training data is proposed. The main idea behind the proposed NAF model is to introduce the attention mechanism into the random forest by assigning attention weights calculated by neural networks of a specific form to data in leaves of decision trees and to the random forest itself in the framework of the Nadaraya-Watson kernel regression. In contrast to the available models like the attention-based random forest, the attention weights and the Nadaraya-Watson regression are represented in the form of neural networks whose weights can be regarded as trainable parameters. The first part of neural networks with shared weights is trained for all trees and computes attention weights of data in leaves. The second part aggregates outputs of the tree networks and aims to minimize the difference between the random forest prediction and the truth target value from a training set. The neural network is trained in an end-to-end manner. The combination of the random forest and neural networks implementing the attention mechanism forms a transformer for enhancing the forest predictions. Numerical experiments with real datasets illustrate the proposed method. The code implementing the approach is publicly available.

Abstract: This study presents a machine learning (ML) pipeline for clinical data classification in the context of a 30-day readmission problem, along with a fairness audit on subgroups based on sensitive attributes. A range of ML models are used for classification and the fairness audit is conducted on the model predictions. The fairness audit uncovers disparities in equal opportunity, predictive parity, false positive rate parity, and false negative rate parity criteria on the MIMIC III dataset based on attributes such as gender, ethnicity, language, and insurance group. The results identify disparities in the model's performance across different groups and highlights the need for better fairness and bias mitigation strategies. The study suggests the need for collaborative efforts among researchers, policymakers, and practitioners to address bias and fairness in artificial intelligence (AI) systems.

Abstract: Physics-informed neural networks (PINNs) have proven a suitable mathematical scaffold for solving inverse ordinary (ODE) and partial differential equations (PDE). Typical inverse PINNs are formulated as soft-constrained multi-objective optimization problems with several hyperparameters. In this work, we demonstrate that inverse PINNs can be framed in terms of maximum-likelihood estimators (MLE) to allow explicit error propagation from interpolation to the physical model space through Taylor expansion, without the need of hyperparameter tuning. We explore its application to high-dimensional coupled ODEs constrained by differential algebraic equations that are common in transient chemical and biological kinetics. Furthermore, we show that singular-value decomposition (SVD) of the ODE coupling matrices (reaction stoichiometry matrix) provides reduced uncorrelated subspaces in which PINNs solutions can be represented and over which residuals can be projected. Finally, SVD bases serve as preconditioners for the inversion of covariance matrices in this hyperparameter-free robust application of MLE to ``kinetics-informed neural networks''.

Abstract: Despite their potential in real-world applications, multi-agent reinforcement learning (MARL) algorithms often suffer from high sample complexity. To address this issue, we present a novel model-based MARL algorithm, BiLL (Bi-Level Latent Variable Model-based Learning), that learns a bi-level latent variable model from high-dimensional inputs. At the top level, the model learns latent representations of the global state, which encode global information relevant to behavior learning. At the bottom level, it learns latent representations for each agent, given the global latent representations from the top level. The model generates latent trajectories to use for policy learning. We evaluate our algorithm on complex multi-agent tasks in the challenging SMAC and Flatland environments. Our algorithm outperforms state-of-the-art model-free and model-based baselines in sample efficiency, including on two extremely challenging Super Hard SMAC maps.

Abstract: Heart disorder has just overtaken cancer as the world's biggest cause of mortality. Several cardiac failures, heart disease mortality, and diagnostic costs can all be reduced with early identification and treatment. Medical data is collected in large quantities by the healthcare industry, but it is not well mined. The discovery of previously unknown patterns and connections in this information can help with an improved decision when it comes to forecasting heart disorder risk. In the proposed study, we constructed an ML-based diagnostic system for heart illness forecasting, using a heart disorder dataset. We used data preprocessing techniques like outlier detection and removal, checking and removing missing entries, feature normalization, cross-validation, nine classification algorithms like RF, MLP, KNN, ETC, XGB, SVC, ADB, DT, and GBM, and eight classifier measuring performance metrics like ramification accuracy, precision, F1 score, specificity, ROC, sensitivity, log-loss, and Matthews' correlation coefficient, as well as eight classification performance evaluations. Our method can easily differentiate between people who have cardiac disease and those are normal. Receiver optimistic curves and also the region under the curves were determined by every classifier. Most of the classifiers, pretreatment strategies, validation methods, and performance assessment metrics for classification models have been discussed in this study. The performance of the proposed scheme has been confirmed, utilizing all of its capabilities. In this work, the impact of clinical decision support systems was evaluated using a stacked ensemble approach that included these nine algorithms