Authors: Lei Yang
Abstract: Intelligent transportation systems (ITS) rely heavily on complete and high-quality spatiotemporal traffic data to achieve optimal performance. Nevertheless, in real-word traffic data collection processes, issues such as communication failures and sensor malfunctions often lead to incomplete or corrupted datasets, thereby posing significant challenges to the advancement of ITS. Among various methods for imputing missing spatiotemporal traffic data, the latent factorization of tensors (LFT) model has emerged as a widely adopted and effective solution. However, conventional LFT models typically employ the standard L2-norm in their learning objective, which makes them vulnerable to the influence of outliers. To overcome this limitation, this paper proposes a threshold distance weighted (TDW) loss-incorporated Latent Factorization of Tensors (TDWLFT) model. The proposed loss function effectively reduces the model's sensitivity to outliers by assigning differentiated weights to individual samples. Extensive experiments conducted on two traffic speed datasets sourced from diverse urban environments confirm that the proposed TDWLFT model consistently outperforms state-of-the-art approaches in terms of both in both prediction accuracy and computational efficiency.
Authors: Piotr Makarevich
Abstract: In everyday reasoning, when we think about a particular object, we associate it with a unique set of expected properties such as weight, size, or more abstract attributes like density or horsepower. These expectations are shaped by our prior knowledge and the conceptual categories we have formed through experience. This paper investigates how such knowledge-based structured thinking can be reproduced in deep learning models using features based embeddings. Specially, it introduces an specific approach to build feature-grounded embedding, aiming to align shareable representations of operable dictionary with interpretable domain-specific conceptual features.
Authors: Sarah Seifi, Tobias Sukianto, Cecilia Carbonelli, Lorenzo Servadei, Robert Wille
Abstract: Rule-based models offer interpretability but struggle with complex data, while deep neural networks excel in performance yet lack transparency. This work investigates a neuro-symbolic rule learning neural network named RL-Net that learns interpretable rule lists through neural optimization, applied for the first time to radar-based hand gesture recognition (HGR). We benchmark RL-Net against a fully transparent rule-based system (MIRA) and an explainable black-box model (XentricAI), evaluating accuracy, interpretability, and user adaptability via transfer learning. Our results show that RL-Net achieves a favorable trade-off, maintaining strong performance (93.03% F1) while significantly reducing rule complexity. We identify optimization challenges specific to rule pruning and hierarchy bias and propose stability-enhancing modifications. Compared to MIRA and XentricAI, RL-Net emerges as a practical middle ground between transparency and performance. This study highlights the real-world feasibility of neuro-symbolic models for interpretable HGR and offers insights for extending explainable AI to edge-deployable sensing systems.
Authors: Jing Wang, Amar Sra, Jeremy C. Weiss
Abstract: The long-term effects of Postacute Sequelae of SARS-CoV-2, known as PASC, pose a significant challenge to healthcare systems worldwide. Accurate identification of progression events, such as hospitalization and reinfection, is essential for effective patient management and resource allocation. However, traditional models trained on structured data struggle to capture the nuanced progression of PASC. In this study, we introduce the first publicly available cohort of 18 PASC patients, with text time series features based on Large Language Model Llama-3.1-70B-Instruct and clinical risk annotated by clinical expert. We propose an Active Attention Network to predict the clinical risk and identify progression events related to the risk. By integrating human expertise with active learning, we aim to enhance clinical risk prediction accuracy and enable progression events identification with fewer number of annotation. The ultimate goal is to improves patient care and decision-making for SARS-CoV-2 patient.
Authors: Saad Alqithami
Abstract: Cyber-Physical Systems play a critical role in the infrastructure of various sectors, including manufacturing, energy distribution, and autonomous transportation systems. However, their increasing connectivity renders them highly vulnerable to sophisticated cyber threats, such as adaptive and zero-day attacks, against which traditional security methods like rule-based intrusion detection and single-agent reinforcement learning prove insufficient. To overcome these challenges, this paper introduces a novel Hierarchical Adversarially-Resilient Multi-Agent Reinforcement Learning (HAMARL) framework. HAMARL employs a hierarchical structure consisting of local agents dedicated to subsystem security and a global coordinator that oversees and optimizes comprehensive, system-wide defense strategies. Furthermore, the framework incorporates an adversarial training loop designed to simulate and anticipate evolving cyber threats, enabling proactive defense adaptation. Extensive experimental evaluations conducted on a simulated industrial IoT testbed indicate that HAMARL substantially outperforms traditional multi-agent reinforcement learning approaches, significantly improving attack detection accuracy, reducing response times, and ensuring operational continuity. The results underscore the effectiveness of combining hierarchical multi-agent coordination with adversarially-aware training to enhance the resilience and security of next-generation CPS.
Authors: Aakash Tripathi, Asim Waqas, Matthew B. Schabath, Yasin Yilmaz, Ghulam Rasool
Abstract: Accurate cancer survival prediction requires integration of diverse data modalities that reflect the complex interplay between imaging, clinical parameters, and textual reports. However, existing multimodal approaches suffer from simplistic fusion strategies, massive computational requirements, and lack of interpretability-critical barriers to clinical adoption. We present EAGLE (Efficient Alignment of Generalized Latent Embeddings), a novel deep learning framework that addresses these limitations through attention-based multimodal fusion with comprehensive attribution analysis. EAGLE introduces four key innovations: (1) dynamic cross-modal attention mechanisms that learn hierarchical relationships between modalities, (2) massive dimensionality reduction (99.96%) while maintaining predictive performance, (3) three complementary attribution methods providing patient-level interpretability, and (4) a unified pipeline enabling seamless adaptation across cancer types. We evaluated EAGLE on 911 patients across three distinct malignancies: glioblastoma (GBM, n=160), intraductal papillary mucinous neoplasms (IPMN, n=171), and non-small cell lung cancer (NSCLC, n=580). Patient-level analysis showed high-risk individuals relied more heavily on adverse imaging features, while low-risk patients demonstrated balanced modality contributions. Risk stratification identified clinically meaningful groups with 4-fold (GBM) to 5-fold (NSCLC) differences in median survival, directly informing treatment intensity decisions. By combining state-of-the-art performance with clinical interpretability, EAGLE bridges the gap between advanced AI capabilities and practical healthcare deployment, offering a scalable solution for multimodal survival prediction that enhances both prognostic accuracy and physician trust in automated predictions.
Authors: Fabio Merizzi, Harilaos Loukos
Abstract: Global Climate Models (GCMs) are critical for simulating large-scale climate dynamics, but their coarse spatial resolution limits their applicability in regional studies. Regional Climate Models (RCMs) refine this through dynamic downscaling, albeit at considerable computational cost and with limited flexibility. While deep learning has emerged as an efficient data-driven alternative, most existing studies have focused on single-variable models that downscale one variable at a time. This approach can lead to limited contextual awareness, redundant computation, and lack of cross-variable interaction. Our study addresses these limitations by proposing a multi-task, multi-variable Vision Transformer (ViT) architecture with a shared encoder and variable-specific decoders (1EMD). The proposed architecture jointly predicts three key climate variables: surface temperature (tas), wind speed (sfcWind), and 500 hPa geopotential height (zg500), directly from GCM-resolution inputs, emulating RCM-scale downscaling over Europe. We show that our multi-variable approach achieves positive cross-variable knowledge transfer and consistently outperforms single-variable baselines trained under identical conditions, while also improving computational efficiency. These results demonstrate the effectiveness of multi-variable modeling for high-resolution climate downscaling.
Authors: Matvei Anoshin, Olga Tsurkan, Vadim Lopatkin, Leonid Fedichkin
Abstract: The stabilization of time series processes is a crucial problem that is ubiquitous in various industrial fields. The application of machine learning to its solution can have a decisive impact, improving both the quality of the resulting stabilization with less computational resources required. In this work, we present a simple pipeline consisting of two neural networks: the oracle predictor and the optimizer, proposing a substitution of the point-wise values optimization to the problem of the neural network training, which successfully improves stability in terms of the temperature control by about 3 times compared to ordinary solvers.
Authors: Jakub Pele\v{s}ka, Gustav \v{S}\'ir
Abstract: Relational Deep Learning (RDL) is an emerging paradigm that leverages Graph Neural Network principles to learn directly from relational databases by representing them as heterogeneous graphs. However, existing RDL models typically rely on task-specific supervised learning, requiring training separate models for each predictive task, which may hamper scalability and reuse. In this work, we propose a novel task-agnostic contrastive pretraining approach for RDL that enables database-wide representation learning. For that aim, we introduce three levels of contrastive objectives$-$row-level, link-level, and context-level$-$designed to capture the structural and semantic heterogeneity inherent to relational data. We implement the respective pretraining approach through a modular RDL architecture and an efficient sampling strategy tailored to the heterogeneous database setting. Our preliminary results on standard RDL benchmarks demonstrate that fine-tuning the pretrained models measurably outperforms training from scratch, validating the promise of the proposed methodology in learning transferable representations for relational data.
Authors: Jacob Adamczyk
Abstract: Exploration remains a fundamental challenge in reinforcement learning (RL), particularly in environments with sparse or adversarial reward structures. In this work, we study how the architecture of deep neural policies implicitly shapes exploration before training. We theoretically and empirically demonstrate strategies for generating ballistic or diffusive trajectories from untrained policies in a toy model. Using the theory of infinite-width networks and a continuous-time limit, we show that untrained policies return correlated actions and result in non-trivial state-visitation distributions. We discuss the distributions of the corresponding trajectories for a standard architecture, revealing insights into inductive biases for tackling exploration. Our results establish a theoretical and experimental framework for using policy initialization as a design tool to understand exploration behavior in early training.
Authors: Xufei Lv, Haoyuan Sun, Xuefeng Bai, Min Zhang, Houde Liu, Kehai Chen
Abstract: Alignment of large language models (LLMs) with human values has recently garnered significant attention, with prominent examples including the canonical yet costly Reinforcement Learning from Human Feedback (RLHF) and the simple Direct Preference Optimization (DPO). In this work, we demonstrate that both RLHF and DPO can be interpreted from the perspective of mutual information (MI) maximization, uncovering a profound connection to contrastive learning. Within this framework, both RLHF and DPO can be viewed as methods that perform contrastive learning based on the positive and negative samples derived from the base model, leveraging the Donsker-Varadhan (DV) lower bound on MI (equivalently, the MINE estimator). This paradigm further explains why RLHF may not intrinsically incentivize reasoning capacities in LLMs beyond what is already present in the base model. Building on this perspective, we replace the DV/MINE bound with the Jensen-Shannon MI estimator and propose Mutual Information Optimization (MIO). Comprehensive theoretical analysis and extensive empirical evaluations demonstrate that MIO mitigates the late-stage decline in chosen-likelihood observed in DPO, achieving competitive or superior performance across various challenging reasoning and mathematical benchmarks. We will release the model and code upon acceptance.
Authors: Joshua C. Zhao, Saurabh Bagchi
Abstract: Transfer learning is often used to decrease the computational cost of model training, as fine-tuning a model allows a downstream task to leverage the features learned from the pre-training dataset and quickly adapt them to a new task. This is particularly useful for achieving adversarial robustness, as adversarially training models from scratch is very computationally expensive. However, high robustness in transfer learning still requires adversarial training during the fine-tuning phase, which requires up to an order of magnitude more time than standard fine-tuning. In this work, we revisit the use of the fast gradient sign method (FGSM) in robust transfer learning to improve the computational cost of adversarial fine-tuning. We surprisingly find that FGSM is much more stable in adversarial fine-tuning than when training from scratch. In particular, FGSM fine-tuning does not suffer from any issues with catastrophic overfitting at standard perturbation budgets of $\varepsilon=4$ or $\varepsilon=8$. This stability is further enhanced with parameter-efficient fine-tuning methods, where FGSM remains stable even up to $\varepsilon=32$ for linear probing. We demonstrate how this stability translates into performance across multiple datasets. Compared to fine-tuning with the more commonly used method of projected gradient descent (PGD), on average, FGSM only loses 0.39% and 1.39% test robustness for $\varepsilon=4$ and $\varepsilon=8$ while using $4\times$ less training time. Surprisingly, FGSM may not only be a significantly more efficient alternative to PGD in adversarially robust transfer learning but also a well-performing one.
Authors: Paul Saves, Edward Hall\'e-Hannan, Jasper Bussemaker, Youssef Diouane, Nathalie Bartoli
Abstract: Simulation-based problems involving mixed-variable inputs frequently feature domains that are hierarchical, conditional, heterogeneous, or tree-structured. These characteristics pose challenges for data representation, modeling, and optimization. This paper reviews extensive literature on these structured input spaces and proposes a unified framework that generalizes existing approaches. In this framework, input variables may be continuous, integer, or categorical. A variable is described as meta if its value governs the presence of other decreed variables, enabling the modeling of conditional and hierarchical structures. We further introduce the concept of partially-decreed variables, whose activation depends on contextual conditions. To capture these inter-variable hierarchical relationships, we introduce design space graphs, combining principles from feature modeling and graph theory. This allows the definition of general hierarchical domains suitable for describing complex system architectures. The framework supports the use of surrogate models over such domains and integrates hierarchical kernels and distances for efficient modeling and optimization. The proposed methods are implemented in the open-source Surrogate Modeling Toolbox (SMT 2.0), and their capabilities are demonstrated through applications in Bayesian optimization for complex system design, including a case study in green aircraft architecture.
Authors: Dmitry B. Rokhlin
Abstract: We study the problem of online regression with the unconstrained quadratic loss against a time-varying sequence of functions from a Reproducing Kernel Hilbert Space (RKHS). Recently, Jacobsen and Cutkosky (2024) introduced a discounted Vovk-Azoury-Warmuth (DVAW) forecaster that achieves optimal dynamic regret in the finite-dimensional case. In this work, we lift their approach to the non-parametric domain by synthesizing the DVAW framework with a random feature approximation. We propose a fully adaptive, hierarchical algorithm, which we call H-VAW-D (Hierarchical Vovk-Azoury-Warmuth with Discounting), that learns both the discount factor and the number of random features. We prove that this algorithm, which has a per-iteration computational complexity of $O(T\ln T)$, achieves an expected dynamic regret of $O(T^{2/3}P_T^{1/3} + \sqrt{T}\ln T)$, where $P_T$ is the functional path length of a comparator sequence.
Authors: Aadim Nepal, Safal Shrestha, Anubhav Shrestha, Minwu Kim, Keith Ross
Abstract: Large language models can exhibit improved mathematical reasoning capabilities following post-training with instruction tuning, reinforcement learning, or knowledge distillation. However, it remains unclear whether these improvements are driven by major changes in transformer layers or from minor adjustments that leave the relative layer importance structures of the base model largely unchanged. We investigate this question through systematic layer-wise ablation experiments, examining base, instruction-tuned, knowledge-distilled, and reinforcement learning variants on mathematical reasoning benchmarks. Our findings show that mathematical reasoning gives rise to a specific layer importance structure, and this structure persists across all post-training paradigms. Removal of such layers causes accuracy drops of up to 80%. In contrast, non-mathematical tasks like factual recall exhibit no critical layers. This distinction suggests that mathematical reasoning requires specialized layers that emerge during pre-training, while other non-reasoning tasks do not. From an information-theoretic perspective, we also observe that these critical layers are the same layers where major representational transformation occurs.
Authors: Amir Hossein Rahmati, Nathan M. Urban, Byung-Jun Yoon, Xiaoning Qian
Abstract: Machine Learning surrogates have been developed to accelerate solving systems dynamics of complex processes in different science and engineering applications. To faithfully capture governing systems dynamics, these methods rely on large training datasets, hence restricting their applicability in real-world problems. In this work, we propose BayPOD-AL, an active learning framework based on an uncertainty-aware Bayesian proper orthogonal decomposition (POD) approach, which aims to effectively learn reduced-order models from high-fidelity full-order models representing complex systems. Experimental results on predicting the temperature evolution over a rod demonstrate BayPOD-AL's effectiveness in suggesting the informative data and reducing computational cost related to constructing a training dataset compared to other uncertainty-guided active learning strategies. Furthermore, we demonstrate BayPOD-AL's generalizability and efficiency by evaluating its performance on a dataset of higher temporal resolution than the training dataset.
Authors: Andrew F. Ilersich, Prasanth B. Nair
Abstract: The physical sciences are replete with dynamical systems that require the resolution of a wide range of length and time scales. This presents significant computational challenges since direct numerical simulation requires discretization at the finest relevant scales, leading to a high-dimensional state space. In this work, we propose an approach to learn stochastic multiscale models in the form of stochastic differential equations directly from observational data. Our method resolves the state on a coarse mesh while introducing an auxiliary state to capture the effects of unresolved scales. We learn the parameters of the multiscale model using a modern forward-solver-free amortized variational inference method. Our approach draws inspiration from physics-based multiscale modeling approaches, such as large-eddy simulation in fluid dynamics, while learning directly from data. We present numerical studies to demonstrate that our learned multiscale models achieve superior predictive accuracy compared to direct numerical simulation and closure-type models at equivalent resolution.
Authors: Selahattin Akkas, Aditya Devarakonda, Ariful Azad
Abstract: With the growing adoption of graph neural networks (GNNs), explaining their predictions has become increasingly important. However, attributing predictions to specific edges or features remains computationally expensive. For example, classifying a node with 100 neighbors using a 3-layer GNN may involve identifying important edges from millions of candidates contributing to the prediction. To address this challenge, we propose DistShap, a parallel algorithm that distributes Shapley value-based explanations across multiple GPUs. DistShap operates by sampling subgraphs in a distributed setting, executing GNN inference in parallel across GPUs, and solving a distributed least squares problem to compute edge importance scores. DistShap outperforms most existing GNN explanation methods in accuracy and is the first to scale to GNN models with millions of features by using up to 128 GPUs on the NERSC Perlmutter supercomputer.
Authors: Anh Bui, Trang Vu, Trung Le, Junae Kim, Tamas Abraham, Rollin Omari, Amar Kaur, Dinh Phung
Abstract: In this paper, we investigate the semantic collapsing problem in generative personalization, an under-explored topic where the learned visual concept ($V^*$) gradually shifts from its original textual meaning and comes to dominate other concepts in multi-concept input prompts. This issue not only reduces the semantic richness of complex input prompts like "a photo of $V^*$ wearing glasses and playing guitar" into simpler, less contextually rich forms such as "a photo of $V^*$" but also leads to simplified output images that fail to capture the intended concept. We identify the root cause as unconstrained optimisation, which allows the learned embedding $V^*$ to drift arbitrarily in the embedding space, both in direction and magnitude. To address this, we propose a simple yet effective training-free method that adjusts the magnitude and direction of pre-trained embedding at inference time, effectively mitigating the semantic collapsing problem. Our method is broadly applicable across different personalization methods and demonstrates significant improvements in text-image alignment in diverse use cases. Our code is anonymously published at https://anonymous.4open.science/r/Embedding-Adjustment.
URLs: https://anonymous.4open.science/r/Embedding-Adjustment.
Authors: Brian Mak, Jeffrey Flanigan
Abstract: The residual stream acts as a memory bus where transformer layers both store and access features (Elhage et al., 2021). We consider changing the mechanism for retrieving and storing information in the residual stream, and replace the residual stream of the transformer with an outer product memory matrix (Kohonen, 1972, Anderson, 1972). We call this model the Residual Matrix Transformer (RMT). We find that the RMT enjoys a number of attractive properties: 1) the size of the residual stream can be scaled independently of compute and model size, improving performance, 2) the RMT can achieve the same loss as the transformer with 58% fewer FLOPS, 25% fewer parameters, and 41% fewer training tokens tokens, and 3) the RMT outperforms the transformer on downstream evaluations. We theoretically analyze the transformer and the RMT, and show that the RMT allows for more efficient scaling of the residual stream, as well as improved variance propagation properties. Code for this project can be found at https://github.com/bmac3/residual-matrix-transformer.
Authors: Shrenik Jadhav, Birva Sevak, Srijita Das, Akhtar Hussain, Wencong Su, Van-Hai Bui
Abstract: Peer-to-peer (P2P) trading is increasingly recognized as a key mechanism for decentralized market regulation, yet existing approaches often lack robust frameworks to ensure fairness. This paper presents FairMarket-RL, a novel hybrid framework that combines Large Language Models (LLMs) with Reinforcement Learning (RL) to enable fairness-aware trading agents. In a simulated P2P microgrid with multiple sellers and buyers, the LLM acts as a real-time fairness critic, evaluating each trading episode using two metrics: Fairness-To-Buyer (FTB) and Fairness-Between-Sellers (FBS). These fairness scores are integrated into agent rewards through scheduled {\lambda}-coefficients, forming an adaptive LLM-guided reward shaping loop that replaces brittle, rule-based fairness constraints. Agents are trained using Independent Proximal Policy Optimization (IPPO) and achieve equitable outcomes, fulfilling over 90% of buyer demand, maintaining fair seller margins, and consistently reaching FTB and FBS scores above 0.80. The training process demonstrates that fairness feedback improves convergence, reduces buyer shortfalls, and narrows profit disparities between sellers. With its language-based critic, the framework scales naturally, and its extension to a large power distribution system with household prosumers illustrates its practical applicability. FairMarket-RL thus offers a scalable, equity-driven solution for autonomous trading in decentralized energy systems.
Authors: Alexander Theus, Alessandro Cabodi, Sotiris Anagnostidis, Antonio Orvieto, Sidak Pal Singh, Valentina Boeva
Abstract: Understanding the geometry of neural network loss landscapes is a central question in deep learning, with implications for generalization and optimization. A striking phenomenon is linear mode connectivity (LMC), where independently trained models can be connected by low- or zero-loss paths, despite appearing to lie in separate loss basins. However, this is often obscured by symmetries in parameter space -- such as neuron permutations -- which make functionally equivalent models appear dissimilar. Prior work has predominantly focused on neuron re-ordering through permutations, but such approaches are limited in scope and fail to capture the richer symmetries exhibited by modern architectures such as Transformers. In this work, we introduce a unified framework that captures four symmetry classes: permutations, semi-permutations, orthogonal transformations, and general invertible maps -- broadening the set of valid reparameterizations and subsuming many previous approaches as special cases. Crucially, this generalization enables, for the first time, the discovery of low- and zero-barrier linear interpolation paths between independently trained Vision Transformers and GPT-2 models. These results reveal deeper structure in the loss landscape and underscore the importance of symmetry-aware analysis for understanding model space geometry.
Authors: Dujian Ding, Ankur Mallick, Shaokun Zhang, Chi Wang, Daniel Madrigal, Mirian Del Carmen Hipolito Garcia, Menglin Xia, Laks V. S. Lakshmanan, Qingyun Wu, Victor R\"uhle
Abstract: Large language models (LLMs) are powerful tools but are often expensive to deploy at scale. LLM query routing mitigates this by dynamically assigning queries to models of varying cost and quality to obtain a desired trade-off. Prior query routing approaches generate only one response from the selected model and a single response from a small (inexpensive) model was often not good enough to beat a response from a large (expensive) model due to which they end up overusing the large model and missing out on potential cost savings. However, it is well known that for small models, generating multiple responses and selecting the best can enhance quality while remaining cheaper than a single large-model response. We leverage this idea to propose BEST-Route, a novel routing framework that chooses a model and the number of responses to sample from it based on query difficulty and the quality thresholds. Experiments on real-world datasets demonstrate that our method reduces costs by up to 60% with less than 1% performance drop.
Authors: Hao Shu, Jicheng Li, Tianyv Lei, Lijun Sun
Abstract: In real-world scenarios, spatiotemporal traffic data frequently experiences dual degradation from missing values and noise caused by sensor malfunctions and communication failures. Therefore, effective data recovery methods are essential to ensure the reliability of downstream data-driven applications. while classical tensor completion methods have been widely adopted, they are incapable of modeling noise, making them unsuitable for complex scenarios involving simultaneous data missingness and noise interference. Existing Robust Tensor Completion (RTC) approaches offer potential solutions by separately modeling the actual tensor data and noise. However, their effectiveness is often constrained by the over-relaxation of convex rank surrogates and the suboptimal utilization of local consistency, leading to inadequate model accuracy. To address these limitations, we first introduce the tensor L1-L2 norm, a novel non-convex tensor rank surrogate that functions as an effective low-rank representation tool. Leveraging an advanced feature fusion strategy, we further develop the gradient tensor L1-L2 norm by incorporating the tensor L1-L2 norm in the gradient domain. By integrating the gradient tensor nuclear L1-L2 norm into the RTC framework, we propose the Robust Tensor Completion via Gradient Tensor Nuclear L1-L2 Norm (RTC-GTNLN) model, which not only fully exploits both global low-rankness and local consistency without trade-off parameter, but also effectively handles the dual degradation challenges of missing data and noise in traffic data. Extensive experiments conducted on multiple real-world traffic datasets demonstrate that the RTC-GTNLN model consistently outperforms existing state-of-the-art methods in complex recovery scenarios involving simultaneous missing values and noise.
Authors: Jingxiao Ma, Priyadarshini Panda, Sherief Reda
Abstract: Backpropagation has been the cornerstone of neural network training for decades, yet its inefficiencies in time and energy consumption limit its suitability for resource-constrained edge devices. While low-precision neural network quantization has been extensively researched to speed up model inference, its application in training has been less explored. Recently, the Forward-Forward (FF) algorithm has emerged as a promising alternative to backpropagation, replacing the backward pass with an additional forward pass. By avoiding the need to store intermediate activations for backpropagation, FF can reduce memory footprint, making it well-suited for embedded devices. This paper presents an INT8 quantized training approach that leverages FF's layer-by-layer strategy to stabilize gradient quantization. Furthermore, we propose a novel "look-ahead" scheme to address limitations of FF and improve model accuracy. Experiments conducted on NVIDIA Jetson Orin Nano board demonstrate 4.6% faster training, 8.3% energy savings, and 27.0% reduction in memory usage, while maintaining competitive accuracy compared to the state-of-the-art.
Authors: Dibyajyoti Chakraborty, Haiwen Guan, Jason Stock, Troy Arcomano, Guido Cervone, Romit Maulik
Abstract: Score-based diffusion modeling is a generative machine learning algorithm that can be used to sample from complex distributions. They achieve this by learning a score function, i.e., the gradient of the log-probability density of the data, and reversing a noising process using the same. Once trained, score-based diffusion models not only generate new samples but also enable zero-shot conditioning of the generated samples on observed data. This promises a novel paradigm for data and model fusion, wherein the implicitly learned distributions of pretrained score-based diffusion models can be updated given the availability of online data in a Bayesian formulation. In this article, we apply such a concept to the super-resolution of a high-dimensional dynamical system, given the real-time availability of low-resolution and experimentally observed sparse sensor measurements from multimodal data. Additional analysis on how score-based sampling can be used for uncertainty estimates is also provided. Our experiments are performed for a super-resolution task that generates the ERA5 atmospheric dataset given sparse observations from a coarse-grained representation of the same and/or from unstructured experimental observations of the IGRA radiosonde dataset. We demonstrate accurate recovery of the high dimensional state given multiple sources of low-fidelity measurements. We also discover that the generative model can balance the influence of multiple dataset modalities during spatiotemporal reconstructions.
Authors: Hae Jin Song, Laurent Itti
Abstract: Recent breakthroughs and rapid integration of generative models (GMs) have sparked interest in the problem of model attribution and their fingerprints. For instance, service providers need reliable methods of authenticating their models to protect their IP, while users and law enforcement seek to verify the source of generated content for accountability and trust. In addition, a growing threat of model collapse is arising, as more model-generated data are being fed back into sources (e.g., YouTube) that are often harvested for training ("regurgitative training"), heightening the need to differentiate synthetic from human data. Yet, a gap still exists in understanding generative models' fingerprints, we believe, stemming from the lack of a formal framework that can define, represent, and analyze the fingerprints in a principled way. To address this gap, we take a geometric approach and propose a new definition of artifact and fingerprint of GMs using Riemannian geometry, which allows us to leverage the rich theory of differential geometry. Our new definition generalizes previous work (Song et al., 2024) to non-Euclidean manifolds by learning Riemannian metrics from data and replacing the Euclidean distances and nearest-neighbor search with geodesic distances and kNN-based Riemannian center of mass. We apply our theory to a new gradient-based algorithm for computing the fingerprints in practice. Results show that it is more effective in distinguishing a large array of GMs, spanning across 4 different datasets in 2 different resolutions (64 by 64, 256 by 256), 27 model architectures, and 2 modalities (Vision, Vision-Language). Using our proposed definition significantly improves the performance on model attribution, as well as a generalization to unseen datasets, model types, and modalities, suggesting its practical efficacy.
Authors: Cooper Doyle
Abstract: We propose BayesLoRA, a task-specific uncertainty quantification framework that integrates MC-Dropout into Low-Rank Adapters (LoRA). Unlike general-purpose transformer uncertainty methods, BayesLoRA provides guardrails tailored to downstream workflows, enabling agents to introspect and modulate behavior under uncertainty. We demonstrate mathematically and empirically that LoRA adapters exhibit amplified variance outside fine-tuning distributions, yielding reliable confidence estimates for agentic decision-making.
Authors: Thomas Gaskin, Guy J. Abel
Abstract: We present a novel and detailed dataset on origin-destination annual migration flows and stocks between 230 countries and regions, spanning the period from 1990 to the present. Our flow estimates are further disaggregated by country of birth, providing a comprehensive picture of migration over the last 43 years. The estimates are obtained by training a deep recurrent neural network to learn flow patterns from 18 covariates for all countries, including geographic, economic, cultural, societal, and political information. The recurrent architecture of the neural network means that the entire past can influence current migration patterns, allowing us to learn long-range temporal correlations. By training an ensemble of neural networks and additionally pushing uncertainty on the covariates through the trained network, we obtain confidence bounds for all our estimates, allowing researchers to pinpoint the geographic regions most in need of additional data collection. We validate our approach on various test sets of unseen data, demonstrating that it significantly outperforms traditional methods estimating five-year flows while delivering a significant increase in temporal resolution. The model is fully open source: all training data, neural network weights, and training code are made public alongside the migration estimates, providing a valuable resource for future studies of human migration.
Authors: Kamil Faber, Marcin Pietro\'n, Dominik \.Zurek, Roberto Corizzo
Abstract: The recently proposed xLSTM is a powerful model that leverages expressive multiplicative gating and residual connections, providing the temporal capacity needed for long-horizon forecasting and representation learning. This architecture has demonstrated success in time series forecasting, lossless compression, and even large-scale language modeling tasks, where its linear memory footprint and fast inference make it a viable alternative to Transformers. Despite its growing popularity, no prior work has explored xLSTM for anomaly detection. In this work, we fill this gap by proposing xLSTMAD, the first anomaly detection method that integrates a full encoder-decoder xLSTM architecture, purpose-built for multivariate time series data. Our encoder processes input sequences to capture historical context, while the decoder is devised in two separate variants of the method. In the forecasting approach, the decoder iteratively generates forecasted future values xLSTMAD-F, while the reconstruction approach reconstructs the input time series from its encoded counterpart xLSTMAD-R. We investigate the performance of two loss functions: Mean Squared Error (MSE), and Soft Dynamic Time Warping (SoftDTW) to consider local reconstruction fidelity and global sequence alignment, respectively. We evaluate our method on the comprehensive TSB-AD-M benchmark, which spans 17 real-world datasets, using state-of-the-art challenging metrics such as VUS-PR. In our results, xLSTM showcases state-of-the-art accuracy, outperforming 23 popular anomaly detection baselines. Our paper is the first work revealing the powerful modeling capabilities of xLSTM for anomaly detection, paving the way for exciting new developments on this subject. Our code is available at: https://github.com/Nyderx/xlstmad
Authors: Batuhan Hangun, Oguz Altun, Onder Eyecioglu
Abstract: Quantum Neural Networks (QNNs), a prominent approach in Quantum Machine Learning (QML), are emerging as a powerful alternative to classical machine learning methods. Recent studies have focused on the applicability of QNNs to various tasks, such as time-series forecasting, prediction, and classification, across a wide range of applications, including cybersecurity and medical imaging. With the increased use of smart grids driven by the integration of renewable energy systems, machine learning plays an important role in predicting power demand and detecting system disturbances. This study provides an in-depth investigation of QNNs for predicting the power output of a wind turbine. We assess the predictive performance and simulation time of six QNN configurations that are based on the Z Feature Map for data encoding and varying ansatz structures. Through detailed cross-validation experiments and tests on an unseen hold-out dataset, we experimentally demonstrate that QNNs can achieve predictive performance that is competitive with, and in some cases marginally better than, the benchmarked classical approaches. Our results also reveal the effects of dataset size and circuit complexity on predictive performance and simulation time. We believe our findings will offer valuable insights for researchers in the energy domain who wish to incorporate quantum machine learning into their work.
Authors: Shengcai Liu, Hui Ou-yang, Zhiyuan Wang, Cheng Chen, Qijun Cai, Yew-Soon Ong, Ke Tang
Abstract: Learning the structure of Bayesian networks (BNs) from data is challenging, especially for datasets involving a large number of variables. The recently proposed divide-and-conquer (D\&D) strategies present a promising approach for learning large BNs. However, they still face a main issue of unstable learning accuracy across subproblems. In this work, we introduce the idea of employing structure learning ensemble (SLE), which combines multiple BN structure learning algorithms, to consistently achieve high learning accuracy. We further propose an automatic approach called Auto-SLE for learning near-optimal SLEs, addressing the challenge of manually designing high-quality SLEs. The learned SLE is then integrated into a D\&D method. Extensive experiments firmly show the superiority of our method over D\&D methods with single BN structure learning algorithm in learning large BNs, achieving accuracy improvement usually by 30\%$\sim$225\% on datasets involving 10,000 variables. Furthermore, our method generalizes well to datasets with many more (e.g., 30000) variables and different network characteristics than those present in the training data for learning the SLE. These results indicate the significant potential of employing (automatic learning of) SLEs for scalable BN structure learning.
Authors: Homayun Afrabandpey, Hamed Rezazadegan Tavakoli
Abstract: Efficient model distribution is becoming increasingly critical in bandwidth-constrained environments. In this paper, we propose a simple yet effective approach called Progressive Precision Update (P$^2$U) to address this problem. Instead of transmitting the original high-precision model, P$^2$U transmits a lower-bit precision model, coupled with a model update representing the difference between the original high-precision model and the transmitted low precision version. With extensive experiments on various model architectures, ranging from small models ($1 - 6$ million parameters) to a large model (more than $100$ million parameters) and using three different data sets, e.g., chest X-Ray, PASCAL-VOC, and CIFAR-100, we demonstrate that P$^2$U consistently achieves better tradeoff between accuracy, bandwidth usage and latency. Moreover, we show that when bandwidth or startup time is the priority, aggressive quantization (e.g., 4-bit) can be used without severely compromising performance. These results establish P$^2$U as an effective and practical solution for scalable and efficient model distribution in low-resource settings, including federated learning, edge computing, and IoT deployments. Given that P$^2$U complements existing compression techniques and can be implemented alongside any compression method, e.g., sparsification, quantization, pruning, etc., the potential for improvement is even greater.
Authors: Xinyu Chen, Vassilis Digalakis Jr, Lijun Ding, Dingyi Zhuang, Jinhua Zhao
Abstract: Time series autoregression is a classical statistical model for capturing auto-correlations and identifying temporal patterns such as periodicity and seasonality. In this work, we propose a novel sparse autoregression framework from an interpretable machine learning perspective and the model interpretability for periodicity quantification is reinforced by $\ell_0$-norm induced sparsity constraints. On the time-varying time series data, we reformulate the sparse autoregression and convert the involved optimization problem into a mixed-integer optimization (MIO). To accelerate it, we develop a subspace pursuit based decision variable pruning (DVP) strategy to reduce the search space. On the multidimensional time series that involves complicated spatial and temporal dimensions, we propose a spatially- and time-varying sparse autoregression model and resolve the corresponding MIO problem by developing a two-stage optimization scheme. In particular, the proposed scheme makes the model scalable to large problems even with millions of decision variables. Empirically, we conduct extensive experiments to evaluate the proposed models on real-world time series data. First, we demonstrate that the MIO solver can be drastically accelerated through the DVP strategy, while maintaining the same solution quality as a full MIO solver. Applying the time-varying sparse autoregression model to ridesharing trip data, we uncover both daily and weekly periodicities and reveal long-term changes in regularity of human mobility. Second, we demonstrate the spatial patterns of yearly seasonality in climate variable time series such as temperature and precipitation across the past four decades, and our model allows to discover dynamic climate patterns and identify climate phenomena such as El Nino in sea surface temperature.
Authors: Sina Tabakhi, Haiping Lu
Abstract: A key challenge in learning from multimodal biological data is missing modalities, where all data from some modalities are missing for some patients. Current fusion methods address this by excluding patients with missing modalities, imputing missing modalities, or making predictions directly with partial modalities. However, they often struggle with diverse missing-modality patterns and the exponential growth of the number of such patterns as the number of modalities increases. To address these limitations, we propose MAGNET (Missing-modality-Aware Graph neural NETwork) for direct prediction with partial modalities, which introduces a patient-modality multi-head attention mechanism to fuse lower-dimensional modality embeddings based on their importance and missingness. MAGNET's complexity increases linearly with the number of modalities while adapting to missing-pattern variability. To generate predictions, MAGNET further constructs a patient graph with fused multimodal embeddings as node features and the connectivity determined by the modality missingness, followed by a conventional graph neural network. Experiments on three public multiomics datasets for cancer classification, with real-world instead of artificial missingness, show that MAGNET outperforms the state-of-the-art fusion methods. The data and code are available at https://github.com/SinaTabakhi/MAGNET.
Authors: Jaeyun Woo, Jiseok Lee, Brian Kenji Iwana
Abstract: Generative Artificial Intelligence (AI) has rapidly become a powerful tool, capable of generating various types of data, such as images and text. However, despite the significant advancement of generative AI, time series generative AI remains underdeveloped, even though the application of time series is essential in finance, climate, and numerous fields. In this research, we propose a novel method of generating time series conditioned on unstructured natural language descriptions. We use a diffusion model combined with a language model to generate time series from the text. Through the proposed method, we demonstrate that time series generation based on natural language is possible. The proposed method can provide various applications such as custom forecasting, time series manipulation, data augmentation, and transfer learning. Furthermore, we construct and propose a new public dataset for time series generation, consisting of 63,010 time series-description pairs.
Authors: Chen Zhang
Abstract: While deep learning excels in natural image and language processing, its application to high-dimensional data faces computational challenges due to the dimensionality curse. Current large-scale data tools focus on business-oriented descriptive statistics, lacking mathematical statistics support for advanced analysis. We propose a parallel computation architecture based on space completeness, decomposing high-dimensional data into dimension-independent structures for distributed processing. This framework enables seamless integration of data mining and parallel-optimized machine learning methods, supporting scientific computations across diverse data types like medical and natural images within a unified system.
Authors: Liangyu Wang, Huanyi Xie, Xinhai Wang, Tianjin Huang, Mengdi Li, Di Wang
Abstract: Group-based reinforcement learning algorithms such as Group Reward Policy Optimization (GRPO) have proven effective for fine-tuning large language models (LLMs) with human feedback. However, generating and storing multiple responses per prompt incurs substantial memory overhead, especially as the sample group size increases, limiting scalability under constrained hardware. We propose Infinite Sampling, a framework that enables efficient and stable GRPO training by decoupling group size from GPU memory usage. It consists of: (1) micro sampling groups that decompose large groups into memory-feasible rounds; (2) continuous sampling that interleaves generation across groups to improve utilization; and (3) a length-aware scheduler combining token-conditioned sequence length prediction with a two-stage plan: global grouping via FPTAS and runtime refill via SJF. Experiments show that our Micro Sampling Groups reduce peak memory usage by over 50% compared to full-group decoding (e.g., from 21.55 GB to 10.64 GB on Qwen3-1.7B). Building on this, Infinite Sampling improves throughput by over 25% compared to the naive micro sampling group method, reducing decoding steps while maintaining full-length completions and memory usage. Our hybrid scheduling ensures efficient and stable GRPO training with larger groups under realistic GPU memory constraints.
Authors: Prathyush Kumar Reddy Lebaku, Lu Gao, Yunpeng Zhang, Zhixia Li, Yongxin Liu, Tanvir Arafin
Abstract: Anomaly detection in connected autonomous vehicles (CAVs) is crucial for maintaining safe and reliable transportation networks, as CAVs can be susceptible to sensor malfunctions, cyber-attacks, and unexpected environmental disruptions. This study explores an anomaly detection approach by simulating vehicle behavior, generating a dataset that represents typical and atypical vehicular interactions. The dataset includes time-series data of position, speed, and acceleration for multiple connected autonomous vehicles. We utilized machine learning models to effectively identify abnormal driving patterns. First, we applied a stacked Long Short-Term Memory (LSTM) model to capture temporal dependencies and sequence-based anomalies. The stacked LSTM model processed the sequential data to learn standard driving behaviors. Additionally, we deployed a Random Forest model to support anomaly detection by offering ensemble-based predictions, which enhanced model interpretability and performance. The Random Forest model achieved an R2 of 0.9830, MAE of 5.746, and a 95th percentile anomaly threshold of 14.18, while the stacked LSTM model attained an R2 of 0.9998, MAE of 82.425, and a 95th percentile anomaly threshold of 265.63. These results demonstrate the models' effectiveness in accurately predicting vehicle trajectories and detecting anomalies in autonomous driving scenarios.
Authors: Can Hakan Da\u{g}{\i}d{\i}r, Mia Hubert, Peter J. Rousseeuw
Abstract: A new anomaly detection method called kernel outlier detection (KOD) is proposed. It is designed to address challenges of outlier detection in high-dimensional settings. The aim is to overcome limitations of existing methods, such as dependence on distributional assumptions or on hyperparameters that are hard to tune. KOD starts with a kernel transformation, followed by a projection pursuit approach. Its novelties include a new ensemble of directions to search over, and a new way to combine results of different direction types. This provides a flexible and lightweight approach for outlier detection. Our empirical evaluations illustrate the effectiveness of KOD on three small datasets with challenging structures, and on four large benchmark datasets.
Authors: Davide Salaorni, Federico Bianchi, Francesco Trov\`o, Marcello Restelli
Abstract: The increasing integration of renewable energy sources (RESs) is transforming traditional power grid networks, which require new approaches for managing decentralized energy production and consumption. Microgrids (MGs) provide a promising solution by enabling localized control over energy generation, storage, and distribution. This paper presents a novel reinforcement learning (RL)-based methodology for optimizing microgrid energy management. Specifically, we propose an RL agent that learns optimal energy trading and storage policies by leveraging historical data on energy production, consumption, and market prices. A digital twin (DT) is used to simulate the energy storage system dynamics, incorporating degradation factors to ensure a realistic emulation of the analysed setting. Our approach is validated through an experimental campaign using real-world data from a power grid located in the Italian territory. The results indicate that the proposed RL-based strategy outperforms rule-based methods and existing RL benchmarks, offering a robust solution for intelligent microgrid management.
Authors: Jerry Liu, Yasa Baig, Denise Hui Jean Lee, Rajat Vadiraj Dwaraknath, Atri Rudra, Chris R\'e
Abstract: Physics-informed neural networks (PINNs) offer a flexible way to solve partial differential equations (PDEs) with machine learning, yet they still fall well short of the machine-precision accuracy many scientific tasks demand. In this work, we investigate whether the precision ceiling comes from the ill-conditioning of the PDEs or from the typical multi-layer perceptron (MLP) architecture. We introduce the Barycentric Weight Layer (BWLer), which models the PDE solution through barycentric polynomial interpolation. A BWLer can be added on top of an existing MLP (a BWLer-hat) or replace it completely (explicit BWLer), cleanly separating how we represent the solution from how we take derivatives for the PDE loss. Using BWLer, we identify fundamental precision limitations within the MLP: on a simple 1-D interpolation task, even MLPs with O(1e5) parameters stall around 1e-8 RMSE -- about eight orders above float64 machine precision -- before any PDE terms are added. In PDE learning, adding a BWLer lifts this ceiling and exposes a tradeoff between achievable accuracy and the conditioning of the PDE loss. For linear PDEs we fully characterize this tradeoff with an explicit error decomposition and navigate it during training with spectral derivatives and preconditioning. Across five benchmark PDEs, adding a BWLer on top of an MLP improves RMSE by up to 30x for convection, 10x for reaction, and 1800x for wave equations while remaining compatible with first-order optimizers. Replacing the MLP entirely lets an explicit BWLer reach near-machine-precision on convection, reaction, and wave problems (up to 10 billion times better than prior results) and match the performance of standard PINNs on stiff Burgers' and irregular-geometry Poisson problems. Together, these findings point to a practical path for combining the flexibility of PINNs with the precision of classical spectral solvers.
Authors: Tejas Vaidhya, Ayush Kaushal, Vineet Jain, Francis Couture Harpin, Prashant Shishodia, Majid Behbahani, Yuriy Nevmyvaka, Irina Rish
Abstract: Large language models (LLMs) are increasingly used across research and industry applications, yet their inference efficiency remains a significant challenge. As the computational power of modern GPU architectures continuously improves, their memory bandwidth and capacity have not scaled proportionally, creating a critical bottleneck during inference. To address this, we investigate ternary language models (TriLMs) that employ quantization-aware training to significantly reduce memory requirements. We first analyze the scalability of TriLMs by conducting a scaling law analysis, revealing that TriLMs benefit more from increasing training data than from scaling model parameters. Based on this observation, we introduce Spectra-1.1, an open suite of TriLMs trained on up to 1.2 trillion tokens, demonstrating sustained performance gains at scale. Furthermore, to improve inference efficiency, we propose novel 2-bit and 1.6-bit packing schemes for ternary weights, which demonstrate accelerated inference across various CPU architectures. Also, building on the 2-bit packing, we develop a GPU kernel called TriRun that accelerates end-to-end model inference by up to 5 times compared to floating-point baselines. To encourage further exploration and development of TriLMs, we will release the Spectra-1.1 suite and TriRun inference kernels. Overall, our work lays the foundation for building and deploying efficient LLMs, providing a valuable resource for the research community.
Authors: Yash Vardhan Tomar
Abstract: Bias in predictive machine learning (ML) models is a fundamental challenge due to the skewed or unfair outcomes produced by biased models. Existing mitigation strategies rely on either post-hoc corrections or rigid constraints. However, emerging research claims that these techniques can limit scalability and reduce generalizability. To address this, this paper introduces a feature-wise mixing framework to mitigate contextual bias. This was done by redistributing feature representations across multiple contextual datasets. To assess feature-wise mixing's effectiveness, four ML classifiers were trained using cross-validation and evaluated with bias-sensitive loss functions, including disparity metrics and mean squared error (MSE), which served as a standard measure of predictive performance. The proposed method achieved an average bias reduction of 43.35% and a statistically significant decrease in MSE across all classifiers trained on mixed datasets. Additionally, benchmarking against established bias mitigation techniques found that feature-wise mixing consistently outperformed SMOTE oversampling and demonstrated competitive effectiveness without requiring explicit bias attribute identification. Feature-wise mixing efficiently avoids the computational overhead typically associated with fairness-aware learning algorithms. Future work could explore applying feature-wise mixing for real-world fields where accurate predictions are necessary.
Authors: Zain ul Abdeen, Ming Jin
Abstract: This paper explores Reinforcement learning (RL) policy robustness by systematically analyzing network parameters under internal and external stresses. Inspired by synaptic plasticity in neuroscience, synaptic filtering introduces internal stress by selectively perturbing parameters, while adversarial attacks apply external stress through modified agent observations. This dual approach enables the classification of parameters as fragile, robust, or antifragile, based on their influence on policy performance in clean and adversarial settings. Parameter scores are defined to quantify these characteristics, and the framework is validated on PPO-trained agents in Mujoco continuous control environments. The results highlight the presence of antifragile parameters that enhance policy performance under stress, demonstrating the potential of targeted filtering techniques to improve RL policy adaptability. These insights provide a foundation for future advancements in the design of robust and antifragile RL systems.
Authors: Chengyu Dong, Huan Gui, Noveen Sachdeva, Long Jin, Ke Yin, Jingbo Shang, Lichan Hong, Ed H. Chi, Zhe Zhao
Abstract: Knowledge distillation from pretrained visual representation models offers an effective approach to improve small, task-specific production models. However, the effectiveness of such knowledge transfer drops significantly when distilling from strong models that are pretrained in a large scale. In this paper, we address this challenge for pretrained Vision Transformers (ViTs) by exploring methods to fine-tune them for more effective knowledge transfer. Motivated by the connection between mutual information and distillation effectiveness, we propose to employ mutual information-aware optimization during finetuning. For small or highly-imbalanced downstream datasets where such optimization becomes less effective, we introduce a simple yet effective heuristic of reweighting MLP blocks. This approach is inspired by our observation that top MLP blocks are primarily responsible for mutual information loss. Our method enables small student models to benefit from those pretrained models among the strongest.
Authors: Hanlin Dong, Arian Prabowo, Hao Xue, Flora D. Salim
Abstract: Air quality prediction is a challenging forecasting task due to its spatio-temporal complexity and the inherent dynamics as well as uncertainty. Most of the current models handle these two challenges by applying Graph Neural Networks or known physics principles, and quantifying stochasticity through probabilistic networks like Diffusion models. Nevertheless, finding the right balancing point between the certainties and uncertainties remains an open question. Therefore, we propose Double-Diffusion, a novel diffusion probabilistic model that harnesses the power of known physics to guide air quality forecasting with stochasticity. To the best of our knowledge, while precedents have been made of using conditional diffusion models to predict air pollution, this is the first attempt to use physics as a conditional generative approach for air quality prediction. Along with a sampling strategy adopted from image restoration and a new denoiser architecture, Double-Diffusion ranks first in most evaluation scenarios across two real-life datasets compared with other probabilistic models, it also cuts inference time by 50% to 30% while enjoying an increase between 3-12% in Continuous Ranked Probabilistic Score (CRPS).
Authors: Hiro Taiyo Hamada, Ippei Fujisawa, Genji Kawakita, Yuki Yamada
Abstract: Large Language Models (LLMs) such as ChatGPT have shown remarkable abilities in producing human-like text. However, it is unclear how accurately these models internalize concepts that shape human thought and behavior. Here, we developed a quantitative framework to assess concept alignment between LLMs and human psychological dimensions using 43 standardized psychological questionnaires, selected for their established validity in measuring distinct psychological constructs. Our method evaluates how accurately language models reconstruct and classify questionnaire items through pairwise similarity analysis. We compared resulting cluster structures with the original categorical labels using hierarchical clustering. A GPT-4 model achieved superior classification accuracy (66.2\%), significantly outperforming GPT-3.5 (55.9\%) and BERT (48.1\%), all exceeding random baseline performance (31.9\%). We also demonstrated that the estimated semantic similarity from GPT-4 is associated with Pearson's correlation coefficients of human responses in multiple psychological questionnaires. This framework provides a novel approach to evaluate the alignment of the human-LLM concept and identify potential representational biases. Our findings demonstrate that modern LLMs can approximate human psychological constructs with measurable accuracy, offering insights for developing more interpretable AI systems.
Authors: Zhiyu Zhao, Haoxuan Li, Haifeng Zhang, Jun Wang, Francesco Faccio, J\"urgen Schmidhuber, Mengyue Yang
Abstract: When building a world model, a common assumption is that the environment has a single, unchanging underlying causal rule, like applying Newton's laws to every situation. In reality, what appears as a drifting causal mechanism is often the manifestation of a fixed underlying mechanism seen through a narrow observational window. This brings about a problem that, when building a world model, even subtle shifts in policy or environment states can alter the very observed causal mechanisms. In this work, we introduce the \textbf{Meta-Causal Graph} as world models, a minimal unified representation that efficiently encodes the transformation rules governing how causal structures shift across different latent world states. A single Meta-Causal Graph is composed of multiple causal subgraphs, each triggered by meta state, which is in the latent state space. Building on this representation, we introduce a \textbf{Causality-Seeking Agent} whose objectives are to (1) identify the meta states that trigger each subgraph, (2) discover the corresponding causal relationships by agent curiosity-driven intervention policy, and (3) iteratively refine the Meta-Causal Graph through ongoing curiosity-driven exploration and agent experiences. Experiments on both synthetic tasks and a challenging robot arm manipulation task demonstrate that our method robustly captures shifts in causal dynamics and generalizes effectively to previously unseen contexts.
Authors: Shahad Hardan, Darya Taratynova, Abdelmajid Essofi, Karthik Nandakumar, Mohammad Yaqub
Abstract: Privacy preservation in AI is crucial, especially in healthcare, where models rely on sensitive patient data. In the emerging field of machine unlearning, existing methodologies struggle to remove patient data from trained multimodal architectures, which are widely used in healthcare. We propose Forget-MI, a novel machine unlearning method for multimodal medical data, by establishing loss functions and perturbation techniques. Our approach unlearns unimodal and joint representations of the data requested to be forgotten while preserving knowledge from the remaining data and maintaining comparable performance to the original model. We evaluate our results using performance on the forget dataset, performance on the test dataset, and Membership Inference Attack (MIA), which measures the attacker's ability to distinguish the forget dataset from the training dataset. Our model outperforms the existing approaches that aim to reduce MIA and the performance on the forget dataset while keeping an equivalent performance on the test set. Specifically, our approach reduces MIA by 0.202 and decreases AUC and F1 scores on the forget set by 0.221 and 0.305, respectively. Additionally, our performance on the test set matches that of the retrained model, while allowing forgetting. Code is available at https://github.com/BioMedIA-MBZUAI/Forget-MI.git
Authors: Jonathan Schuster, Fabian Transchel
Abstract: In the domain of vehicle telematics the automated recognition of driving maneuvers is used to classify and evaluate driving behaviour. This not only serves as a component to enhance the personalization of insurance policies, but also to increase road safety, reduce accidents and the associated costs as well as to reduce fuel consumption and support environmentally friendly driving. In this context maneuver recognition technically requires a continuous application of time series classification which poses special challenges to the transfer, preprocessing and storage of telematic sensor data, the training of predictive models, and the prediction itself. Although much research has been done in the field of gathering relevant data or regarding the methods to build predictive models for the task of maneuver recognition, there is a practical need for python packages and functions that allow to quickly transform data into the required structure as well as to build and evaluate such models. The maneuverRecognition package was therefore developed to provide the necessary functions for preprocessing, modelling and evaluation and also includes a ready to use LSTM based network structure that can be modified. The implementation of the package is demonstrated using real driving data of three different persons recorded via smartphone sensors.
Authors: David Bossens, Atsushi Nitanda
Abstract: Safety is an essential requirement for reinforcement learning systems. The newly emerging framework of robust constrained Markov decision processes allows learning policies that satisfy long-term constraints while providing guarantees under epistemic uncertainty. This paper presents mirror descent policy optimisation for robust constrained Markov decision processes (RCMDPs), making use of policy gradient techniques to optimise both the policy (as a maximiser) and the transition kernel (as an adversarial minimiser) on the Lagrangian representing a constrained MDP. In the oracle-based RCMDP setting, we obtain an $\mathcal{O}\left(\frac{1}{T}\right)$ convergence rate for the squared distance as a Bregman divergence, and an $\mathcal{O}\left(e^{-T}\right)$ convergence rate for entropy-regularised objectives. In the sample-based RCMDP setting, we obtain an $\tilde{\mathcal{O}}\left(\frac{1}{T^{1/3}}\right)$ convergence rate. Experiments confirm the benefits of mirror descent policy optimisation in constrained and unconstrained optimisation, and significant improvements are observed in robustness tests when compared to baseline policy optimisation algorithms.
Authors: Chen Gong, Bo Liang, Wei Gao, Chenren Xu
Abstract: Generative models have gained significant attention for their ability to produce realistic synthetic data that supplements the quantity of real-world datasets. While recent studies show performance improvements in wireless sensing tasks by incorporating all synthetic data into training sets, the quality of synthetic data remains unpredictable and the resulting performance gains are not guaranteed. To address this gap, we propose tractable and generalizable metrics to quantify quality attributes of synthetic data - affinity and diversity. Our assessment reveals prevalent affinity limitation in current wireless synthetic data, leading to mislabeled data and degraded task performance. We attribute the quality limitation to generative models' lack of awareness of untrained conditions and domain-specific processing. To mitigate these issues, we introduce SynCheck, a quality-guided synthetic data utilization scheme that refines synthetic data quality during task model training. Our evaluation demonstrates that SynCheck consistently outperforms quality-oblivious utilization of synthetic data, and achieves 4.3% performance improvement even when the previous utilization degrades performance by 13.4%.
Authors: Robert Frank, Michael Widrich, Rahmad Akbar, G\"unter Klambauer, Geir Kjetil Sandve, Philippe A. Robert, Victor Greiff
Abstract: Generative machine learning models offer a powerful framework for therapeutic design by efficiently exploring large spaces of biological sequences enriched for desirable properties. Unlike supervised learning methods, which require both positive and negative labeled data, generative models such as LSTMs can be trained solely on positively labeled sequences, for example, high-affinity antibodies. This is particularly advantageous in biological settings where negative data are scarce, unreliable, or biologically ill-defined. However, the lack of attribution methods for generative models has hindered the ability to extract interpretable biological insights from such models. To address this gap, we developed Generative Attribution Metric Analysis (GAMA), an attribution method for autoregressive generative models based on Integrated Gradients. We assessed GAMA using synthetic datasets with known ground truths to characterize its statistical behavior and validate its ability to recover biologically relevant features. We further demonstrated the utility of GAMA by applying it to experimental antibody-antigen binding data. GAMA enables model interpretability and the validation of generative sequence design strategies without the need for negative training data.
Authors: Marco Bressan, Victor Chepoi, Emmanuel Esposito, Maximilian Thiessen
Abstract: Abstract notions of convexity over the vertices of a graph, and corresponding notions of halfspaces, have recently gained attention from the machine learning community. In this work we study monophonic halfspaces, a notion of graph halfspaces defined through closure under induced paths. Our main result is a $2$-satisfiability based decomposition theorem, which allows one to represent monophonic halfspaces as a disjoint union of certain vertex subsets. Using this decomposition, we achieve efficient and (nearly) optimal algorithms for various learning problems, such as teaching, active, and online learning. Most notably, we obtain a polynomial-time algorithm for empirical risk minimization. Independently of the decomposition theorem, we obtain an efficient, stable, and proper sample compression scheme. This makes monophonic halfspaces efficiently learnable with proper learners and linear error rate $1/\varepsilon$ in the realizable PAC setting. Our results answer open questions from the literature, and show a stark contrast with geodesic halfspaces, for which most of the said learning problems are NP-hard.
Authors: Haoran Li, Muhao Guo, Marija Ilic, Yang Weng, Guangchun Ruan
Abstract: Accurate residential load forecasting is critical for power system reliability with rising renewable integration and demand-side flexibility. However, most statistical and machine learning models treat external factors, such as weather, calendar effects, and pricing, as extra input, ignoring their heterogeneity, and thus limiting the extraction of useful external information. We propose a paradigm shift: external data should serve as meta-knowledge to dynamically adapt the forecasting model itself. Based on this idea, we design a meta-representation framework using hypernetworks that modulate selected parameters of a base Deep Learning (DL) model in response to external conditions. This provides both expressivity and adaptability. We further integrate a Mixture-of-Experts (MoE) mechanism to enhance efficiency through selective expert activation, while improving robustness by filtering redundant external inputs. The resulting model, dubbed as a Meta Mixture of Experts for External data (M2oE2), achieves substantial improvements in accuracy and robustness with limited additional overhead, outperforming existing state-of-the-art methods in diverse load datasets. The dataset and source code are publicly available at https://github.com/haorandd/M2oE2\_load\_forecast.git.
URLs: https://github.com/haorandd/M2oE2\_load\_forecast.git.
Authors: Taehwan Yoon, Bongjun Choi
Abstract: Federated learning(FL) is used for distributed scenarios to train artificial intelligence(AI) models while ensuring users' privacy. In federated learning scenario, the server generally never knows about users' data. This type of concept makes the AI training process efficient in terms of data privacy. However, regarding model performance, federated AI models may not sufficiently satisfy AI users' expectations. Furthermore, AI users have a wide range of different needs. It is not easy to satisfy the whole users needs. These types of issues can be addressed through AI model optimization, fine-tuning, or personalization to achieve optimal model performance. To address model optimization challenges, we propose reference model-based federated learning for optimal fine-tuning, which overcomes catastrophic forgetting in each round. This method is derived from Bayesian parameter-efficient transfer learning, which includes an optimal proximal term and enables overcoming the catastrophic forgetting issue in each round by utilizing a reference model that incorporates previous model parameters. As a result, this method achieves both high model performance and low computing cost.
Authors: B\'alint Horv\'ath, Bal\'azs Csan\'ad Cs\'aji
Abstract: The paper proposes a statistical learning approach to the problem of estimating missing pixels of images, crucial for image inpainting and super-resolution problems. One of the main novelties of the method is that it also provides uncertainty quantifications together with the estimated values. Our core assumption is that the underlying data-generating function comes from a Reproducing Kernel Hilbert Space (RKHS). A special emphasis is put on band-limited functions, central to signal processing, which form Paley-Wiener type RKHSs. The proposed method, which we call Simultaneously Guaranteed Kernel Interpolation (SGKI), is an extension and refinement of a recently developed kernel method. An advantage of SGKI is that it not only estimates the missing pixels, but also builds non-asymptotic confidence bands for the unobserved values, which are simultaneously guaranteed for all missing pixels. We also show how to compute these bands efficiently using Schur complements, we discuss a generalization to vector-valued functions, and we present a series of numerical experiments on various datasets containing synthetically generated and benchmark images, as well.
Authors: Yukito Tajima, Nakamasa Inoue, Yusuke Sekikawa, Ikuro Sato, Rio Yokota
Abstract: Gated Linear Units (GLUs) have become essential components in the feed-forward networks of state-of-the-art Large Language Models (LLMs). However, they require twice as many memory reads compared to feed-forward layers without gating, due to the use of separate weight matrices for the gate and value streams. To address this bottleneck, we introduce Masked Gated Linear Units (MGLUs), a novel family of GLUs with an efficient kernel implementation. The core contribution of MGLUs include: (1) the Mixture of Element-wise Gating (MoEG) architecture that learns multiple binary masks, each determining gate or value assignments at the element level on a single shared weight matrix resulting in reduced memory transfer, and (2) FlashMGLU, a hardware-friendly kernel that yields up to a 19.7 $\times$ inference-time speed-up over a naive PyTorch MGLU and is 47% more memory-efficient and 34% faster than standard GLUs despite added architectural complexity on an RTX5090 GPU. In LLM experiments, the Swish-activated variant SwiMGLU preserves its memory advantages while matching - or even surpassing - the downstream accuracy of the SwiGLU baseline.
Authors: Lujun Li, Zhu Qiyuan, Jiacheng Wang, Wei Li, Hao Gu, Sirui Han, Yike Guo
Abstract: Mixture of Experts (MoE) LLMs face significant obstacles due to their massive parameter scale, which imposes memory, storage, and deployment challenges. Although recent expert merging methods promise greater efficiency by consolidating multiple experts, they are fundamentally hindered by parameter conflicts arising from expert specialization. In this paper, we present Sub-MoE, a novel MoE compression framework via Subspace Expert Merging. Our key insight is to perform joint Singular Value Decomposition (SVD) on concatenated expert weights, reducing conflicting parameters by extracting shared $U$-matrices while enabling effective merging of the expert-specific $V$ components. Specifically, Sub-MoE consists of two innovative phases: (1) Adaptive Expert Clustering, which groups functionally coherent experts via K-means clustering based on cosine similarity of expert outputs; and (2) Subspace Expert Merging, which first enforces Experts Union Decomposition to derive the shared $U$-matrix across experts in the same group, then pursues frequency-based merging for individual $V$-matrices, and finalizes expert reconstruction using the merged $V$-matrix. In this way, we align and fuse experts in a shared subspace, and can be extended with intra-expert compression for further inference optimization. Extensive experiments on Mixtral, DeepSeek, and Qwen-1.5|3 MoE LLMs demonstrate that our Sub-MoE significantly outperforms existing expert pruning and merging methods. Notably, our Sub-MoE maintains 96\%|86\% of original performance with 25\%|50\% expert reduction on Mixtral-8x7B in zero-shot benchmarks. Code will be released at https://github.com/lliai/MoERazor.
Authors: Hans Peter Lynsg{\o}e Raaschou-jensen, Constanza Fierro, Anders S{\o}gaard
Abstract: Reasoning models that produce long, hidden chains of thought have emerged as powerful tools for complex, reasoning-intensive tasks\citep{deepseekai2025deepseekr1incentivizingreasoningcapability, openai2024openaio1card}. However, this paradigm introduces a new user experience challenge: users have little insight into how much time the model will spend reasoning before returning an answer. This unpredictability, can lead to user frustration and is likely to compound as LLMs can produce increasingly long tasks asynchronously \citep{kwa2025measuringaiabilitycomplete}. In this paper, we introduce and evaluate methods for both online and offline prediction of model "thinking time," aiming to develop a practical "progress bar for reasoning." We discuss the implications for user interaction and future research directions.
Authors: Chaoqun Du, Yulin Wang, Shiji Song, Gao Huang
Abstract: Bayesian decision theory advocates the Bayes classifier as the optimal approach for minimizing the risk in machine learning problems. Current deep learning algorithms usually solve for the optimal classifier by \emph{implicitly} estimating the posterior probabilities, \emph{e.g.}, by minimizing the Softmax cross-entropy loss. This simple methodology has been proven effective for meticulously balanced academic benchmark datasets. However, it is not applicable to the long-tailed data distributions in the real world, where it leads to the gradient imbalance issue and fails to ensure the Bayes optimal decision rule. To address these challenges, this paper presents a novel approach (BAPE) that provides a more precise theoretical estimation of the data distributions by \emph{explicitly} modeling the parameters of the posterior probabilities and solving them with point estimation. Consequently, our method directly learns the Bayes classifier without gradient descent based on Bayes' theorem, simultaneously alleviating the gradient imbalance and ensuring the Bayes optimal decision rule. Furthermore, we propose a straightforward yet effective \emph{distribution adjustment} technique. This method enables the Bayes classifier trained from the long-tailed training set to effectively adapt to the test data distribution with an arbitrary imbalance factor, thereby enhancing performance without incurring additional computational costs. In addition, we demonstrate the gains of our method are orthogonal to existing learning approaches for long-tailed scenarios, as they are mostly designed under the principle of \emph{implicitly} estimating the posterior probabilities. Extensive empirical evaluations on CIFAR-10-LT, CIFAR-100-LT, ImageNet-LT, and iNaturalist demonstrate that our method significantly improves the generalization performance of popular deep networks, despite its simplicity.
Authors: Alan Jeffares, Mihaela van der Schaar
Abstract: Developing a better understanding of surprising or counterintuitive phenomena has constituted a significant portion of deep learning research in recent years. These include double descent, grokking, and the lottery ticket hypothesis -- among many others. Works in this area often develop ad hoc hypotheses attempting to explain these observed phenomena on an isolated, case-by-case basis. This position paper asserts that, in many prominent cases, there is little evidence to suggest that these phenomena appear in real-world applications and these efforts may be inefficient in driving progress in the broader field. Consequently, we argue against viewing them as isolated puzzles that require bespoke resolutions or explanations. However, despite this, we suggest that deep learning phenomena do still offer research value by providing unique settings in which we can refine our broad explanatory theories of more general deep learning principles. This position is reinforced by analyzing the research outcomes of several prominent examples of these phenomena from the recent literature. We revisit the current norms in the research community in approaching these problems and propose practical recommendations for future research, aiming to ensure that progress on deep learning phenomena is well aligned with the ultimate pragmatic goal of progress in the broader field of deep learning.
Authors: Zelin Zang, WenZhe Li, Fei Chen, Yongjie Xu, Chang Yu, Zhen Lei, Stan Z. Li
Abstract: In single-cell research, tracing and analyzing high-throughput single-cell differentiation trajectories is crucial for understanding complex biological processes. Key to this is the modeling and generation of hierarchical data that represents the intrinsic structure within datasets. Traditional methods face limitations in terms of computational cost, performance, generative capacity, and stability. Recent VAEs based approaches have made strides in addressing these challenges but still require specialized network modules for each tree branch, limiting their stability and ability to capture deep hierarchical relationships. To overcome these challenges, we introduce diffusion-based approach called HDTree. HDTree captures tree relationships within a hierarchical latent space using a unified hierarchical codebook and quantized diffusion processes to model tree node transitions. This method improves stability by eliminating branch-specific modules and enhancing generative capacity through gradual hierarchical changes simulated by the diffusion process. HDTree's effectiveness is demonstrated through comparisons on both general-purpose and single-cell datasets, where it outperforms existing methods in terms of accuracy and performance. These contributions provide a new tool for hierarchical lineage analysis, enabling more accurate and efficient modeling of cellular differentiation paths and offering insights for downstream biological tasks. The code of HDTree is available at anonymous link https://anonymous.4open.science/r/code_HDTree_review-A8DB.
URLs: https://anonymous.4open.science/r/code_HDTree_review-A8DB.
Authors: Malikussaid, Hilal Hudan Nuha
Abstract: Large Language Models (LLMs) demonstrate remarkable potential for scientific discovery, but their application in domains requiring factual accuracy and domain-specific constraints remains challenging. In molecular design for drug discovery, LLMs can suggest creative molecular modifications but often produce chemically invalid or impractical structures. We present VALID-Mol, a systematic framework for integrating chemical validation with LLM-driven molecular design that increases the rate of generating valid chemical structures from 3% to 83%. Our approach combines methodical prompt engineering, automated chemical validation, and a fine-tuned domain-adapted LLM to ensure reliable generation of synthesizable molecules with improved properties. Beyond the specific implementation, we contribute a generalizable methodology for scientifically-constrained LLM applications, with quantifiable reliability improvements. Computational predictions suggest our framework can generate promising candidates for synthesis with up to 17-fold computationally predicted improvements in target affinity while maintaining synthetic accessibility. We provide a detailed analysis of our prompt engineering process, validation architecture, and fine-tuning approach, offering a reproducible blueprint for applying LLMs to other scientific domains where domain-specific validation is essential.
Authors: Keziah Naggita, Julienne LaChance
Abstract: Following the rise in popularity of data-centric machine learning (ML), various data valuation methods have been proposed to quantify the contribution of each datapoint to desired ML model performance metrics (e.g., accuracy). Beyond the technical applications of data valuation methods (e.g., data cleaning, data acquisition, etc.), it has been suggested that within the context of data markets, data buyers might utilize such methods to fairly compensate data owners. Here we demonstrate that data valuation metrics are inherently biased and unstable under simple algorithmic design choices, resulting in both technical and ethical implications. By analyzing 9 tabular classification datasets and 6 data valuation methods, we illustrate how (1) common and inexpensive data pre-processing techniques can drastically alter estimated data values; (2) subsampling via data valuation metrics may increase class imbalance; and (3) data valuation metrics may undervalue underrepresented group data. Consequently, we argue in favor of increased transparency associated with data valuation in-the-wild and introduce the novel Data Valuation Cards (DValCards) framework towards this aim. The proliferation of DValCards will reduce misuse of data valuation metrics, including in data pricing, and build trust in responsible ML systems.
Authors: Pawel Renc, Michal K. Grzeszczyk, Linglong Qian, Nassim Oufattole, Jeff Rasley, Arkadiusz Sitek
Abstract: We present Federated Timeline Synthesis (FTS), a novel framework for training generative foundation models across distributed timeseries data applied to electronic health records (EHR). At its core, FTS represents patient history as tokenized Patient Health Timelines (PHTs), language-agnostic sequences encoding temporal, categorical, and continuous clinical information. Each institution trains an autoregressive transformer on its local PHTs and transmits only model weights to a central server. The server uses the generators to synthesize a large corpus of trajectories and train a Global Generator (GG), enabling zero-shot inference via Monte Carlo simulation of future PHTs. We evaluate FTS on five clinically meaningful prediction tasks using MIMIC-IV data, showing that models trained on synthetic data generated by GG perform comparably to those trained on real data. FTS offers strong privacy guarantees, scalability across institutions, and extensibility to diverse prediction and simulation tasks especially in healthcare, including counterfactual inference, early warning detection, and synthetic trial design.
Authors: Dominik Meier, Sujai Hiremath, Promit Ghosal, Kyra Gan
Abstract: Distinguishing cause and effect from bivariate observational data is a foundational problem in many disciplines, but challenging without additional assumptions. Additive noise models (ANMs) are widely used to enable sample-efficient bivariate causal discovery. However, conventional ANM-based methods fail when unobserved mediators corrupt the causal relationship between variables. This paper makes three key contributions: first, we rigorously characterize why standard ANM approaches break down in the presence of unmeasured mediators. Second, we demonstrate that prior solutions for hidden mediation are brittle in finite sample settings, limiting their practical utility. To address these gaps, we propose Bivariate Denoising Diffusion (BiDD) for causal discovery, a method designed to handle latent noise introduced by unmeasured mediators. Unlike prior methods that infer directionality through mean squared error loss comparisons, our approach introduces a novel independence test statistic: during the noising and denoising processes for each variable, we condition on the other variable as input and evaluate the independence of the predicted noise relative to this input. We prove asymptotic consistency of BiDD under the ANM, and conjecture that it performs well under hidden mediation. Experiments on synthetic and real-world data demonstrate consistent performance, outperforming existing methods in mediator-corrupted settings while maintaining strong performance in mediator-free settings.
Authors: Claudionor Coelho Jr, Yanen Li, Philip Tee
Abstract: Large Language Models (LLMs) have rapidly transformed the landscape of artificial intelligence, enabling natural language interfaces and dynamic orchestration of software components. However, their reliance on probabilistic inference limits their effectiveness in domains requiring strict logical reasoning, discrete decision-making, and robust interpretability. This paper investigates these limitations and proposes a neurosymbolic approach that augments LLMs with logic-based reasoning modules, particularly leveraging Prolog predicates and composable toolsets. By integrating first-order logic and explicit rule systems, our framework enables LLMs to decompose complex queries into verifiable sub-tasks, orchestrate reliable solutions, and mitigate common failure modes such as hallucination and incorrect step decomposition. We demonstrate the practical benefits of this hybrid architecture through experiments on the DABStep benchmark, showing improved precision, coverage, and system documentation in multi-step reasoning tasks. Our results indicate that combining LLMs with modular logic reasoning restores engineering rigor, enhances system reliability, and offers a scalable path toward trustworthy, interpretable AI agents across complex domains.
Authors: Amanda S Barnard
Abstract: Benchmark data sets are a cornerstone of machine learning development and applications, ensuring new methods are robust, reliable and competitive. The relative rarity of benchmark sets in computational science, due to the uniqueness of the problems and the pace of change in the associated domains, makes evaluating new innovations difficult for computational scientists. In this paper a new tool is developed and tested to potentially turn any of the increasing numbers of scientific data sets made openly available into a benchmark accessible to the community. BenchMake uses non-negative matrix factorisation to deterministically identify and isolate challenging edge cases on the convex hull (the smallest convex set that contains all existing data instances) and partitions a required fraction of matched data instances into a testing set that maximises divergence and statistical significance, across tabular, graph, image, signal and textual modalities. BenchMake splits are compared to establish splits and random splits using ten publicly available benchmark sets from different areas of science, with different sizes, shapes, distributions.
Authors: Heitor R. Medeiros, Hossein Sharifi-Noghabi, Gabriel L. Oliveira, Saghar Irandoust
Abstract: Real-world time series often exhibit a non-stationary nature, degrading the performance of pre-trained forecasting models. Test-Time Adaptation (TTA) addresses this by adjusting models during inference, but existing methods typically update the full model, increasing memory and compute costs. We propose PETSA, a parameter-efficient method that adapts forecasters at test time by only updating small calibration modules on the input and output. PETSA uses low-rank adapters and dynamic gating to adjust representations without retraining. To maintain accuracy despite limited adaptation capacity, we introduce a specialized loss combining three components: (1) a robust term, (2) a frequency-domain term to preserve periodicity, and (3) a patch-wise structural term for structural alignment. PETSA improves the adaptability of various forecasting backbones while requiring fewer parameters than baselines. Experimental results on benchmark datasets show that PETSA achieves competitive or better performance across all horizons. Our code is available at: https://github.com/BorealisAI/PETSA
Authors: Mohamed Elbasheer, Adewale Akinfaderin
Abstract: Insider threat detection presents unique challenges due to the authorized status of malicious actors and the subtlety of anomalous behaviors. Existing machine learning methods often treat user activity as isolated events, thereby failing to leverage sequential dependencies in user behavior. In this study, we propose a User-Based Sequencing (UBS) methodology, transforming the CERT insider threat dataset into structured temporal sequences suitable for deep sequential modeling. We deploy a Transformer Encoder architecture to model benign user activity and employ its reconstruction errors as anomaly scores. These scores are subsequently evaluated using three unsupervised outlier detection algorithms: One-Class SVM (OCSVM), Local Outlier Factor (LOF), and Isolation Forest (iForest). Across four rigorously designed test sets, including combinations of multiple CERT dataset releases, our UBS-Transformer pipeline consistently achieves state-of-the-art performance - notably 96.61% accuracy, 99.43% recall, 96.38% F1-score, 95.00% AUROC, and exceptionally low false negative (0.0057) and false positive (0.0571) rates. Comparative analyses demonstrate that our approach substantially outperforms tabular and conventional autoencoder baselines, underscoring the efficacy of sequential user modeling and advanced anomaly detection in the insider threat domain.
Authors: Manaswi Kulahara, Gautam Siddharth Kashyap, Nipun Joshi, Arpita Soni
Abstract: Effective disaster management requires timely and accurate insights, yet traditional methods struggle to integrate multimodal data such as images, weather records, and textual reports. To address this, we propose DisasterNet-LLM, a specialized Large Language Model (LLM) designed for comprehensive disaster analysis. By leveraging advanced pretraining, cross-modal attention mechanisms, and adaptive transformers, DisasterNet-LLM excels in disaster classification. Experimental results demonstrate its superiority over state-of-the-art models, achieving higher accuracy of 89.5%, an F1 score of 88.0%, AUC of 0.92%, and BERTScore of 0.88% in multimodal disaster classification tasks.
Authors: Chunjing Xiao, Jiahui Lu, Xovee Xu, Fan Zhou, Tianshu Xie, Wei Lu, Lifeng Xu
Abstract: Graph anomaly detection is critical in domains such as healthcare and economics, where identifying deviations can prevent substantial losses. Existing unsupervised approaches strive to learn a single model capable of detecting both attribute and structural anomalies. However, they confront the tug-of-war problem between two distinct types of anomalies, resulting in suboptimal performance. This work presents TripleAD, a mutual distillation-based triple-channel graph anomaly detection framework. It includes three estimation modules to identify the attribute, structural, and mixed anomalies while mitigating the interference between different types of anomalies. In the first channel, we design a multiscale attribute estimation module to capture extensive node interactions and ameliorate the over-smoothing issue. To better identify structural anomalies, we introduce a link-enhanced structure estimation module in the second channel that facilitates information flow to topologically isolated nodes. The third channel is powered by an attribute-mixed curvature, a new indicator that encapsulates both attribute and structural information for discriminating mixed anomalies. Moreover, a mutual distillation strategy is introduced to encourage communication and collaboration between the three channels. Extensive experiments demonstrate the effectiveness of the proposed TripleAD model against strong baselines.
Authors: Haolan Guo, Linwei Tao, Haoyang Luo, Minjing Dong, Chang Xu
Abstract: Recent advances in deep learning have significantly improved predictive accuracy. However, modern neural networks remain systematically overconfident, posing risks for deployment in safety-critical scenarios. Current post-hoc calibration methods face a fundamental dilemma: global approaches like Temperature Scaling apply uniform adjustments across all samples, introducing high bias despite computational efficiency, while more expressive methods that operate on full logit distributions suffer from high variance due to noisy high-dimensional inputs and insufficient validation data. To address these challenges, we propose Sample Margin-Aware Recalibration of Temperature (SMART), a lightweight, data-efficient recalibration method that precisely scales logits based on the margin between the top two logits -- termed the logit gap. Specifically, the logit gap serves as a denoised, scalar signal directly tied to decision boundary uncertainty, providing a robust indicator that avoids the noise inherent in high-dimensional logit spaces while preserving model prediction invariance. Meanwhile, SMART employs a novel soft-binned Expected Calibration Error (SoftECE) objective that balances model bias and variance through adaptive binning, enabling stable parameter updates even with extremely limited calibration data. Extensive evaluations across diverse datasets and architectures demonstrate that SMART achieves state-of-the-art calibration performance even with substantially fewer parameters compared to existing parametric methods, offering a principled, robust, and highly efficient solution for practical uncertainty quantification in neural network predictions. The source code is available at: https://anonymous.4open.science/r/SMART-8B11.
Authors: Seung-Wook Kim, Seongyeol Kim, Jiah Kim, Seowon Ji, Se-Ho Lee
Abstract: Federated learning (FL) often suffers from performance degradation due to key challenges such as data heterogeneity and communication constraints. To address these limitations, we present a novel FL framework called FedWSQ, which integrates weight standardization (WS) and the proposed distribution-aware non-uniform quantization (DANUQ). WS enhances FL performance by filtering out biased components in local updates during training, thereby improving the robustness of the model against data heterogeneity and unstable client participation. In addition, DANUQ minimizes quantization errors by leveraging the statistical properties of local model updates. As a result, FedWSQ significantly reduces communication overhead while maintaining superior model accuracy. Extensive experiments on FL benchmark datasets demonstrate that FedWSQ consistently outperforms existing FL methods across various challenging FL settings, including extreme data heterogeneity and ultra-low-bit communication scenarios.
Authors: Kento Imaizumi, Hideaki Iiduka
Abstract: Momentum methods were originally introduced for their superiority to stochastic gradient descent (SGD) in deterministic settings with convex objective functions. However, despite their widespread application to deep neural networks -- a representative case of stochastic nonconvex optimization -- the theoretical justification for their effectiveness in such settings remains limited. Quasi-hyperbolic momentum (QHM) is an algorithm that generalizes various momentum methods and has been studied to better understand the class of momentum-based algorithms as a whole. In this paper, we provide both asymptotic and non-asymptotic convergence results for mini-batch QHM with an increasing batch size. We show that achieving asymptotic convergence requires either a decaying learning rate or an increasing batch size. Since a decaying learning rate adversely affects non-asymptotic convergence, we demonstrate that using mini-batch QHM with an increasing batch size -- without decaying the learning rate -- can be a more effective strategy. Our experiments show that even a finite increase in batch size can provide benefits for training neural networks.
Authors: Jingpu Cheng, Qianxiao Li, Ting Lin, Zuowei Shen
Abstract: We investigate the universal approximation property (UAP) of transformer-type architectures, providing a unified theoretical framework that extends prior results on residual networks to models incorporating attention mechanisms. Our work identifies token distinguishability as a fundamental requirement for UAP and introduces a general sufficient condition that applies to a broad class of architectures. Leveraging an analyticity assumption on the attention layer, we can significantly simplify the verification of this condition, providing a non-constructive approach in establishing UAP for such architectures. We demonstrate the applicability of our framework by proving UAP for transformers with various attention mechanisms, including kernel-based and sparse attention mechanisms. The corollaries of our results either generalize prior works or establish UAP for architectures not previously covered. Furthermore, our framework offers a principled foundation for designing novel transformer architectures with inherent UAP guarantees, including those with specific functional symmetries. We propose examples to illustrate these insights.
Authors: Neta Shaul, Uriel Singer, Itai Gat, Yaron Lipman
Abstract: Diffusion and flow matching models have significantly advanced media generation, yet their design space is well-explored, somewhat limiting further improvements. Concurrently, autoregressive (AR) models, particularly those generating continuous tokens, have emerged as a promising direction for unifying text and media generation. This paper introduces Transition Matching (TM), a novel discrete-time, continuous-state generative paradigm that unifies and advances both diffusion/flow models and continuous AR generation. TM decomposes complex generation tasks into simpler Markov transitions, allowing for expressive non-deterministic probability transition kernels and arbitrary non-continuous supervision processes, thereby unlocking new flexible design avenues. We explore these choices through three TM variants: (i) Difference Transition Matching (DTM), which generalizes flow matching to discrete-time by directly learning transition probabilities, yielding state-of-the-art image quality and text adherence as well as improved sampling efficiency. (ii) Autoregressive Transition Matching (ARTM) and (iii) Full History Transition Matching (FHTM) are partially and fully causal models, respectively, that generalize continuous AR methods. They achieve continuous causal AR generation quality comparable to non-causal approaches and potentially enable seamless integration with existing AR text generation techniques. Notably, FHTM is the first fully causal model to match or surpass the performance of flow-based methods on text-to-image task in continuous domains. We demonstrate these contributions through a rigorous large-scale comparison of TM variants and relevant baselines, maintaining a fixed architecture, training data, and hyperparameters.
Authors: Min-Yeong Park, Won-Jeong Lee, Seong Tae Kim, Gyeong-Moon Park
Abstract: Recently, forecasting future abnormal events has emerged as an important scenario to tackle real-world necessities. However, the solution of predicting specific future time points when anomalies will occur, known as Anomaly Prediction (AP), remains under-explored. Existing methods dealing with time series data fail in AP, focusing only on immediate anomalies or failing to provide precise predictions for future anomalies. To address the AP task, we propose a novel framework called Anomaly to Prompt (A2P), comprised of Anomaly-Aware Forecasting (AAF) and Synthetic Anomaly Prompting (SAP). To enable the forecasting model to forecast abnormal time points, we adopt a strategy to learn the relationships of anomalies. For the robust detection of anomalies, our proposed SAP introduces a learnable Anomaly Prompt Pool (APP) that simulates diverse anomaly patterns using signal adaptive prompt. Comprehensive experiments on multiple real-world datasets demonstrate the superiority of A2P over state-of-the-art methods, showcasing its ability to predict future anomalies. Our implementation code is available at https://github.com/KU-VGI/AP.
Authors: Xin Liao, Bing Yang, Cai Yu
Abstract: The integrity of Water Quality Data (WQD) is critical in environmental monitoring for scientific decision-making and ecological protection. However, water quality monitoring systems are often challenged by large amounts of missing data due to unavoidable problems such as sensor failures and communication delays, which further lead to water quality data becoming High-Dimensional and Sparse (HDS). Traditional data imputation methods are difficult to depict the potential dynamics and fail to capture the deep data features, resulting in unsatisfactory imputation performance. To effectively address the above issues, this paper proposes a Nonlinear Low-rank Representation model (NLR) with Convolutional Neural Networks (CNN) for imputing missing WQD, which utilizes CNNs to implement two ideas: a) fusing temporal features to model the temporal dependence of data between time slots, and b) Extracting nonlinear interactions and local patterns to mine higher-order relationships features and achieve deep fusion of multidimensional information. Experimental studies on three real water quality datasets demonstrate that the proposed model significantly outperforms existing state-of-the-art data imputation models in terms of estimation accuracy. It provides an effective approach for handling water quality monitoring data in complex dynamic environments.
Authors: David Demitri Africa, Sara M. Kapoor, Theo Simon Sorg
Abstract: Modular exponentiation is crucial to number theory and cryptography, yet remains largely unexplored from a mechanistic interpretability standpoint. We train a 4-layer encoder-decoder Transformer model to perform this operation and investigate the emergence of numerical reasoning during training. Utilizing principled sampling strategies, PCA-based embedding analysis, and activation patching, we examine how number-theoretic properties are encoded within the model. We find that reciprocal operand training leads to strong performance gains, with sudden generalization across related moduli. These synchronized accuracy surges reflect grokking-like dynamics, suggesting the model internalizes shared arithmetic structure. We also find a subgraph consisting entirely of attention heads in the final layer sufficient to achieve full performance on the task of regular exponentiation. These results suggest that transformer models learn modular arithmetic through specialized computational circuits, paving the way for more interpretable and efficient neural approaches to modular exponentiation.
Authors: Alex Egg, Martin Iglesias Goyanes, Friso Kingma, Andreu Mora, Leandro von Werra, Thomas Wolf
Abstract: We introduce DABstep, a novel benchmark for evaluating AI agents on realistic multi-step data analysis tasks. DABstep comprises over 450 real-world challenges derived from a financial analytics platform, requiring models to combine code-based data processing with contextual reasoning over heterogeneous documentation. Each task demands an iterative, multi-step problem-solving approach, testing capabilities in data manipulation, cross-referencing multiple sources, and precise result reporting. The benchmark provides a factoid-style answer format with automatic correctness checks for objective scoring at scale. We evaluate leading LLM-based agents, revealing a substantial performance gap: even the best agent achieves only 14.55% accuracy on the hardest tasks. We detail our benchmark's design, dataset composition, task formulation, evaluation protocol, report baseline results and analyze failure modes. DABstep is released with a public leaderboard and toolkit to accelerate research in autonomous data analysis.
Authors: Bartlomiej Sobieski, Matthew Tivnan, Yuang Wang, Siyeop Yoon, Pengfei Jin, Dufan Wu, Quanzheng Li, Przemyslaw Biecek
Abstract: Solving inverse problems -- recovering signals from incomplete or noisy measurements -- is fundamental in science and engineering. Score-based generative models (SGMs) have recently emerged as a powerful framework for this task. Two main paradigms have formed: unsupervised approaches that adapt pretrained generative models to inverse problems, and supervised bridge methods that train stochastic processes conditioned on paired clean and corrupted data. While the former typically assume knowledge of the measurement model, the latter have largely overlooked this structural information. We introduce System embedded Diffusion Bridge Models (SDBs), a new class of supervised bridge methods that explicitly embed the known linear measurement system into the coefficients of a matrix-valued SDE. This principled integration yields consistent improvements across diverse linear inverse problems and demonstrates robust generalization under system misspecification between training and deployment, offering a promising solution to real-world applications.
Authors: Michel Meintz, Jan Dubi\'nski, Franziska Boenisch, Adam Dziedzic
Abstract: Image generative models have become increasingly popular, but training them requires large datasets that are costly to collect and curate. To circumvent these costs, some parties may exploit existing models by using the generated images as training data for their own models. In general, watermarking is a valuable tool for detecting unauthorized use of generated images. However, when these images are used to train a new model, watermarking can only enable detection if the watermark persists through training and remains identifiable in the outputs of the newly trained model - a property known as radioactivity. We analyze the radioactivity of watermarks in images generated by diffusion models (DMs) and image autoregressive models (IARs). We find that existing watermarking methods for DMs fail to retain radioactivity, as watermarks are either erased during encoding into the latent space or lost in the noising-denoising process (during the training in the latent space). Meanwhile, despite IARs having recently surpassed DMs in image generation quality and efficiency, no radioactive watermarking methods have been proposed for them. To overcome this limitation, we propose the first watermarking method tailored for IARs and with radioactivity in mind - drawing inspiration from techniques in large language models (LLMs), which share IARs' autoregressive paradigm. Our extensive experimental evaluation highlights our method's effectiveness in preserving radioactivity within IARs, enabling robust provenance tracking, and preventing unauthorized use of their generated images.
Authors: Dan Yao, Steve McLaughlin, Yoann Altmann
Abstract: In this paper, we propose a unifying message-passing framework for training spiking neural networks (SNNs) using Expectation-Propagation. Our gradient-free method is capable of learning the marginal distributions of network parameters and simultaneously marginalizes nuisance parameters, such as the outputs of hidden layers. This framework allows for the first time, training of discrete and continuous weights, for deterministic and stochastic spiking networks, using batches of training samples. Although its convergence is not ensured, the algorithm converges in practice faster than gradient-based methods, without requiring a large number of passes through the training data. The classification and regression results presented pave the way for new efficient training methods for deep Bayesian networks.
Authors: Jari Peeperkorn, Johannes De Smedt, Jochen De Weerdt
Abstract: Process discovery algorithms automatically extract process models from event logs, but high variability often results in complex and hard-to-understand models. To mitigate this issue, trace clustering techniques group process executions into clusters, each represented by a simpler and more understandable process model. Model-driven trace clustering improves on this by assigning traces to clusters based on their conformity to cluster-specific process models. However, most existing clustering techniques rely on either no process model discovery, or non-stochastic models, neglecting the frequency or probability of activities and transitions, thereby limiting their capability to capture real-world execution dynamics. We propose a novel model-driven trace clustering method that optimizes stochastic process models within each cluster. Our approach uses entropic relevance, a stochastic conformance metric based on directly-follows probabilities, to guide trace assignment. This allows clustering decisions to consider both structural alignment with a cluster's process model and the likelihood that a trace originates from a given stochastic process model. The method is computationally efficient, scales linearly with input size, and improves model interpretability by producing clusters with clearer control-flow patterns. Extensive experiments on public real-life datasets show that our method outperforms existing alternatives in representing process behavior and reveals how clustering performance rankings can shift when stochasticity is considered.
Authors: Xiaoyang Li, Linwei Tao, Haohui Lu, Minjing Dong, Junbin Gao, Chang Xu
Abstract: Graph Neural Networks (GNNs) have demonstrated strong predictive performance on relational data; however, their confidence estimates often misalign with actual predictive correctness, posing significant limitations for deployment in safety-critical settings. While existing graph-aware calibration methods seek to mitigate this limitation, they primarily depend on coarse one-hop statistics, such as neighbor-predicted confidence, or latent node embeddings, thereby neglecting the fine-grained structural heterogeneity inherent in graph topology. In this work, we propose Wavelet-Aware Temperature Scaling (WATS), a post-hoc calibration framework that assigns node-specific temperatures based on tunable heat-kernel graph wavelet features. Specifically, WATS harnesses the scalability and topology sensitivity of graph wavelets to refine confidence estimates, all without necessitating model retraining or access to neighboring logits or predictions. Extensive evaluations across seven benchmark datasets with varying graph structures and two GNN backbones demonstrate that WATS achieves the lowest Expected Calibration Error (ECE) among all compared methods, outperforming both classical and graph-specific baselines by up to 42.3\% in ECE and reducing calibration variance by 17.24\% on average compared with graph-specific methods. Moreover, WATS remains computationally efficient, scaling well across graphs of diverse sizes and densities. Code will be released based on publication.
Authors: Jiongli Zhu, Parjanya Prajakta Prashant, Alex Cloninger, Babak Salimi
Abstract: Training data increasingly shapes not only model accuracy but also regulatory compliance and market valuation of AI assets. Yet existing valuation methods remain inadequate: model-based techniques depend on a single fitted model and inherit its biases, while algorithm-based approaches such as Data Shapley require costly retrainings at web scale. Recent Wasserstein-based model-agnostic methods rely on approximations that misrank examples relative to their true leave-one-out (LOO) utility. We introduce KAIROS, a scalable, model-agnostic valuation framework that assigns each example a distributional influence score: its contribution to the Maximum Mean Discrepancy (MMD) between the empirical training distribution and a clean reference set. Unlike Wasserstein surrogates, our MMD-based influence admits a closed-form solution that faithfully approximates the exact LOO ranking within $O(1/N^2)$ error, requires no retraining, and naturally extends to conditional kernels for unified label- and feature-error detection. Moreover, KAIROS supports efficient online updates: when a new batch of size m arrives, all scores can be updated in $O(mN)$ time, delivering up to 50x speedup without compromising ranking quality. Empirical evaluations on noise, mislabeling, and poisoning benchmarks show that KAIROS consistently outperforms state-of-the-art model-, Shapley-, and Wasserstein-based baselines in both accuracy and runtime. We provide rigorous theoretical guarantees, including symmetry for reproducible rankings and density-separation for interpretable thresholds.
Authors: Chang Qi, Matteo Forasassi, Thomas Lukasiewicz, Tommaso Salvatori
Abstract: Predictive coding networks trained with equilibrium propagation are neural models that perform inference through an iterative energy minimization process. Previous studies have demonstrated their effectiveness in shallow architectures, but show significant performance degradation when depth exceeds five to seven layers. In this work, we show that the reason behind this degradation is due to exponentially imbalanced errors between layers during weight updates, and predictions from the previous layer not being effective in guiding updates in deeper layers. We address the first issue by introducing two novel methods to optimize the latent variables that use precision-weighting to re-balance the distribution of energy among layers during the `relaxation phase', and the second issue by proposing a novel weight update mechanism that reduces error accumulation in deeper layers. Empirically, we test our methods on a large number of image classification tasks, resulting in large improvements in test accuracy across networks with more than seven layers, with performances comparable to those of backprop on similar models. These findings suggest that a better understanding of the relaxation phase is important to train models using equilibrium propagation at scale, and open new possibilities for their application in complex tasks.
Authors: Konstantinos Bourazas, Savvas Papaioannou, Panayiotis Kolios
Abstract: In this work we introduce a novel adaptive anomaly detection framework specifically designed for monitoring sequential random finite set (RFS) observations. Our approach effectively distinguishes between In-Control data (normal) and Out-Of-Control data (anomalies) by detecting deviations from the expected statistical behavior of the process. The primary contributions of this study include the development of an innovative RFS-based framework that not only learns the normal behavior of the data-generating process online but also dynamically adapts to behavioral shifts to accurately identify abnormal point patterns. To achieve this, we introduce a new class of RFS-based posterior distributions, named Power Discounting Posteriors (PD), which facilitate adaptation to systematic changes in data while enabling anomaly detection of point pattern data through a novel predictive posterior density function. The effectiveness of the proposed approach is demonstrated by extensive qualitative and quantitative simulation experiments.
Authors: Dmitry Kovalev
Abstract: In this paper, we revisit stochastic gradient descent (SGD) with AdaGrad-type preconditioning. Our contributions are twofold. First, we develop a unified convergence analysis of SGD with adaptive preconditioning under anisotropic or matrix smoothness and noise assumptions. This allows us to recover state-of-the-art convergence results for several popular adaptive gradient methods, including AdaGrad-Norm, AdaGrad, and ASGO/One-sided Shampoo. In addition, we establish the fundamental connection between two recently proposed algorithms, Scion and DASGO, and provide the first theoretical guarantees for the latter. Second, we show that the convergence of methods like AdaGrad and DASGO can be provably accelerated beyond the best-known rates using Nesterov momentum. Consequently, we obtain the first theoretical justification that AdaGrad-type algorithms can simultaneously benefit from both diagonal preconditioning and momentum, which may provide an ultimate explanation for the practical efficiency of Adam.
Authors: Durgesh Singh, Ahcene Boubekki, Robert Jenssen, Michael C. Kampffmeyer
Abstract: The development of semi-supervised learning (SSL) has in recent years largely focused on the development of new consistency regularization or entropy minimization approaches, often resulting in models with complex training strategies to obtain the desired results. In this work, we instead propose a novel approach that explicitly incorporates the underlying clustering assumption in SSL through extending a recently proposed differentiable clustering module. Leveraging annotated data to guide the cluster centroids results in a simple end-to-end trainable deep SSL approach. We demonstrate that the proposed model improves the performance over the supervised-only baseline and show that our framework can be used in conjunction with other SSL methods to further boost their performance.
Authors: Joris Bekkers
Abstract: Understanding team formations and player positioning is crucial for tactical analysis in football (soccer). This paper presents a flexible method for formation recognition and player position assignment in football using predefined static formation templates and cost minimization from spatiotemporal tracking data, called EFPI. Our approach employs linear sum assignment to optimally match players to positions within a set of template formations by minimizing the total distance between actual player locations and template positions, subsequently selecting the formation with the lowest assignment cost. To improve accuracy, we scale actual player positions to match the dimensions of these formation templates in both width and length. While the method functions effectively on individual frames, it extends naturally to larger game segments such as complete periods, possession sequences or specific intervals (e.g. 10 second intervals, 5 minute intervals etc.). Additionally, we incorporate an optional stability parameter that prevents unnecessary formation changes when assignment costs differ only marginally between time segments. EFPI is available as open-source code through the unravelsports Python package.
Authors: Kenny Peng, Rajiv Movva, Jon Kleinberg, Emma Pierson, Nikhil Garg
Abstract: While sparse autoencoders (SAEs) have generated significant excitement, a series of negative results have added to skepticism about their usefulness. Here, we establish a conceptual distinction that reconciles competing narratives surrounding SAEs. We argue that while SAEs may be less effective for acting on known concepts, SAEs are powerful tools for discovering unknown concepts. This distinction cleanly separates existing negative and positive results, and suggests several classes of SAE applications. Specifically, we outline use cases for SAEs in (i) ML interpretability, explainability, fairness, auditing, and safety, and (ii) social and health sciences.
Authors: Eduard Buss, Till Aust, Heiko Hamann
Abstract: Living plants, while contributing to ecological balance and climate regulation, also function as natural sensors capable of transmitting information about their internal physiological states and surrounding conditions. This rich source of data provides potential for applications in environmental monitoring and precision agriculture. With integration into biohybrid systems, we establish novel channels of physiological signal flow between living plants and artificial devices. We equipped *Hedera helix* with a plant-wearable device called PhytoNode to continuously record the plant's electrophysiological activity. We deployed plants in an uncontrolled outdoor environment to map electrophysiological patterns to environmental conditions. Over five months, we collected data that we analyzed using state-of-the-art and automated machine learning (AutoML). Our classification models achieve high performance, reaching macro F1 scores of up to 95 percent in binary tasks. AutoML approaches outperformed manual tuning, and selecting subsets of statistical features further improved accuracy. Our biohybrid living system monitors the electrophysiology of plants in harsh, real-world conditions. This work advances scalable, self-sustaining, and plant-integrated living biohybrid systems for sustainable environmental monitoring.
Authors: Yuta Sato, Kazuhiko Kawamoto, Hiroshi Kera
Abstract: The chain of thought is fundamental in Transformers, which is to perform step-by-step reasoning. Besides what intermediate steps work, the order of these steps critically affects the difficulty of the reasoning. This study addresses a novel task of unraveling chain of thought - reordering decoder input tokens to a learning-friendly sequence for Transformers to learn arithmetic tasks. The proposed pipeline first trains a Transformer on a mixture of target sequences arranged in different orders and then identifies benign orders as those with fast loss drops in the early stage. As the search space grows factorially with sequence length, we propose a two-stage hierarchical approach for inter- and intra-block reordering. Experiments on four order-sensitive arithmetic tasks show that our method identifies a learning-friendly order out of a few billion candidates. Notably, on the multiplication task, it recovered the reverse-digit order reported in prior studies.
Authors: Miguel Camacho-S\'anchez, Fernando Garc\'ia-Torres, Jesper John Lisegaard, Roc\'io del Amor, Sankhya Mohanty, Valery Naranjo
Abstract: Resin infusion (RI) and resin transfer moulding (RTM) are critical processes for the manufacturing of high-performance fibre-reinforced polymer composites, particularly for large-scale applications such as wind turbine blades. Controlling the resin flow dynamics in these processes is critical to ensure the uniform impregnation of the fibre reinforcements, thereby preventing residual porosities and dry spots that impact the consequent structural integrity of the final component. This paper presents a reinforcement learning (RL) based strategy, established using process simulations, for synchronising the different resin flow fronts in an infusion scenario involving two resin inlets and a single outlet. Using Proximal Policy Optimisation (PPO), our approach addresses the challenge of managing the fluid dynamics in a partially observable environment. The results demonstrate the effectiveness of the RL approach in achieving an accurate flow convergence, highlighting its potential towards improving process control and product quality in composites manufacturing.
Authors: Ikechukwu Ogbonna, Lesley Davidson, Soumya Banerjee, Abhishek Dasgupta, Laurence Kenney, Vikranth Harthikote Nagaraja
Abstract: Millions of people in African countries face barriers to accessing healthcare due to language and literacy gaps. This research tackles this challenge by transforming complex medical documents -- in this case, prosthetic device user manuals -- into accessible formats for underserved populations. This case study in cross-cultural translation is particularly pertinent/relevant for communities that receive donated prosthetic devices but may not receive the accompanying user documentation. Or, if available online, may only be available in formats (e.g., language and readability) that are inaccessible to local populations (e.g., English-language, high resource settings/cultural context). The approach is demonstrated using the widely spoken Pidgin dialect, but our open-source framework has been designed to enable rapid and easy extension to other languages/dialects. This work presents an AI-powered framework designed to process and translate complex medical documents, e.g., user manuals for prosthetic devices, into marginalised languages. The system enables users -- such as healthcare workers or patients -- to upload English-language medical equipment manuals, pose questions in their native language, and receive accurate, localised answers in real time. Technically, the system integrates a Retrieval-Augmented Generation (RAG) pipeline for processing and semantic understanding of the uploaded manuals. It then employs advanced Natural Language Processing (NLP) models for generative question-answering and multilingual translation. Beyond simple translation, it ensures accessibility to device instructions, treatment protocols, and safety information, empowering patients and clinicians to make informed healthcare decisions.
Authors: Mingfei Cheng, Xiaofei Xie, Renzhi Wang, Yuan Zhou, Ming Hu
Abstract: Autonomous Driving Systems (ADSs) continue to face safety-critical risks due to the inherent limitations in their design and performance capabilities. Online repair plays a crucial role in mitigating such limitations, ensuring the runtime safety and reliability of ADSs. Existing online repair solutions enforce ADS compliance by transforming unacceptable trajectories into acceptable ones based on predefined specifications, such as rule-based constraints or training datasets. However, these approaches often lack generalizability, adaptability and tend to be overly conservative, resulting in ineffective repairs that not only fail to mitigate safety risks sufficiently but also degrade the overall driving experience. To address this issue, we propose Adaptive Decision Repair (ADReFT), a novel and effective repair method that identifies safety-critical states through offline learning from failed tests and generates appropriate mitigation actions to improve ADS safety. Specifically, ADReFT incorporates a transformer-based model with two joint heads, State Monitor and Decision Adapter, designed to capture complex driving environment interactions to evaluate state safety severity and generate adaptive repair actions. Given the absence of oracles for state safety identification, we first pretrain ADReFT using supervised learning with coarse annotations, i.e., labeling states preceding violations as positive samples and others as negative samples. It establishes ADReFT's foundational capability to mitigate safety-critical violations, though it may result in somewhat conservative mitigation strategies. Therefore, we subsequently finetune ADReFT using reinforcement learning to improve its initial capability and generate more precise and contextually appropriate repair decisions. Our evaluation results illustrate that ADReFT achieves better repair performance.
Authors: Brandon M. Wood, Misko Dzamba, Xiang Fu, Meng Gao, Muhammed Shuaibi, Luis Barroso-Luque, Kareem Abdelmaqsoud, Vahe Gharakhanyan, John R. Kitchin, Daniel S. Levine, Kyle Michel, Anuroop Sriram, Taco Cohen, Abhishek Das, Ammar Rizvi, Sushree Jagriti Sahoo, Zachary W. Ulissi, C. Lawrence Zitnick
Abstract: The ability to quickly and accurately compute properties from atomic simulations is critical for advancing a large number of applications in chemistry and materials science including drug discovery, energy storage, and semiconductor manufacturing. To address this need, Meta FAIR presents a family of Universal Models for Atoms (UMA), designed to push the frontier of speed, accuracy, and generalization. UMA models are trained on half a billion unique 3D atomic structures (the largest training runs to date) by compiling data across multiple chemical domains, e.g. molecules, materials, and catalysts. We develop empirical scaling laws to help understand how to increase model capacity alongside dataset size to achieve the best accuracy. The UMA small and medium models utilize a novel architectural design we refer to as mixture of linear experts that enables increasing model capacity without sacrificing speed. For example, UMA-medium has 1.4B parameters but only ~50M active parameters per atomic structure. We evaluate UMA models on a diverse set of applications across multiple domains and find that, remarkably, a single model without any fine-tuning can perform similarly or better than specialized models. We are releasing the UMA code, weights, and associated data to accelerate computational workflows and enable the community to continue to build increasingly capable AI models.
Authors: Zain ul Abdeen, Vassilis Kekatos, Ming Jin
Abstract: Certified robustness is a critical property for deploying neural networks (NN) in safety-critical applications. A principle approach to achieving such guarantees is to constrain the global Lipschitz constant of the network. However, accurate methods for Lipschitz-constrained training often suffer from non-convex formulations and poor scalability due to reliance on global semidefinite programs (SDPs). In this letter, we propose a convex training framework that enforces global Lipschitz constraints via semidefinite relaxation. By reparameterizing the NN using loop transformation, we derive a convex admissibility condition that enables tractable and certifiable training. While the resulting formulation guarantees robustness, its scalability is limited by the size of global SDP. To overcome this, we develop a randomized subspace linear matrix inequalities (RS-LMI) approach that decomposes the global constraints into sketched layerwise constraints projected onto low-dimensional subspaces, yielding a smooth and memory-efficient training objective. Empirical results on MNIST, CIFAR-10, and ImageNet demonstrate that the proposed framework achieves competitive accuracy with significantly improved Lipschitz bounds and runtime performance.
Authors: Samuele Marro, Philip Torr
Abstract: While the Internet's core infrastructure was designed to be open and universal, today's application layer is dominated by closed, proprietary platforms. Open and interoperable APIs require significant investment, and market leaders have little incentive to enable data exchange that could erode their user lock-in. We argue that LLM-based agents fundamentally disrupt this status quo. Agents can automatically translate between data formats and interact with interfaces designed for humans: this makes interoperability dramatically cheaper and effectively unavoidable. We name this shift universal interoperability: the ability for any two digital services to exchange data seamlessly using AI-mediated adapters. Universal interoperability undermines monopolistic behaviours and promotes data portability. However, it can also lead to new security risks and technical debt. Our position is that the ML community should embrace this development while building the appropriate frameworks to mitigate the downsides. By acting now, we can harness AI to restore user freedom and competitive markets without sacrificing security.
Authors: Juan Maro\~nas
Abstract: This document shows how to obtain the Jacobian and Hessian matrices of the Kullback-Leibler divergence between two multivariate Gaussian distributions, using the first and second-order differentials. The presented derivations are based on the theory presented by \cite{magnus99}. I've also got great inspiration from some of the derivations in \cite{minka}. Since I pretend to be at most didactic, the document is split into a summary of results and detailed derivations on each of the elements involved, with specific references to the tricks used in the derivations, and to many of the underlying concepts.
Authors: Lijun Sheng, Jian Liang, Ran He, Zilei Wang, Tieniu Tan
Abstract: Test-time adaptation (TTA) methods have gained significant attention for enhancing the performance of vision-language models (VLMs) such as CLIP during inference, without requiring additional labeled data. However, current TTA researches generally suffer from major limitations such as duplication of baseline results, limited evaluation metrics, inconsistent experimental settings, and insufficient analysis. These problems hinder fair comparisons between TTA methods and obscure their practical strengths and weaknesses. To address these challenges, we introduce TTA-VLM, a comprehensive benchmark for evaluating TTA methods on VLMs. Our benchmark implements 8 episodic TTA and 7 online TTA methods within a unified and reproducible framework, and evaluates them across 15 widely used datasets. Unlike prior studies focused solely on CLIP, we extend the evaluation to SigLIP--a model trained with a Sigmoid loss--and include training-time tuning methods such as CoOp, MaPLe, and TeCoA to assess generality. Beyond classification accuracy, TTA-VLM incorporates various evaluation metrics, including robustness, calibration, out-of-distribution detection, and stability, enabling a more holistic assessment of TTA methods. Through extensive experiments, we find that 1) existing TTA methods produce limited gains compared to the previous pioneering work; 2) current TTA methods exhibit poor collaboration with training-time fine-tuning methods; 3) accuracy gains frequently come at the cost of reduced model trustworthiness. We release TTA-VLM to provide fair comparison and comprehensive evaluation of TTA methods for VLMs, and we hope it encourages the community to develop more reliable and generalizable TTA strategies.
Authors: He Wang, Xingyu Xu, Yuejie Chi
Abstract: While Bayesian-based exploration often demonstrates superior empirical performance compared to bonus-based methods in model-based reinforcement learning (RL), its theoretical understanding remains limited for model-free settings. Existing provable algorithms either suffer from computational intractability or rely on stage-wise policy updates which reduce responsiveness and slow down the learning process. In this paper, we propose a novel variant of Q-learning algorithm, refereed to as RandomizedQ, which integrates sampling-based exploration with agile, step-wise, policy updates, for episodic tabular RL. We establish an $\widetilde{O}(\sqrt{H^5SAT})$ regret bound, where $S$ is the number of states, $A$ is the number of actions, $H$ is the episode length, and $T$ is the total number of episodes. In addition, we present a logarithmic regret bound under a mild positive sub-optimality condition on the optimal Q-function. Empirically, RandomizedQ exhibits outstanding performance compared to existing Q-learning variants with both bonus-based and Bayesian-based exploration on standard benchmarks.
Authors: Veronica Lachi, Francesco Ferrini, Antonio Longa, Bruno Lepri, Andrea Passerini, Manfred Jaeger
Abstract: Graph Neural Networks (GNNs) are widely used to compute representations of node pairs for downstream tasks such as link prediction. Yet, theoretical understanding of their expressive power has focused almost entirely on graph-level representations. In this work, we shift the focus to links and provide the first comprehensive study of GNN expressiveness in link representation. We introduce a unifying framework, the $k_\phi$-$k_\rho$-$m$ framework, that subsumes existing message-passing link models and enables formal expressiveness comparisons. Using this framework, we derive a hierarchy of state-of-the-art methods and offer theoretical tools to analyze future architectures. To complement our analysis, we propose a synthetic evaluation protocol comprising the first benchmark specifically designed to assess link-level expressiveness. Finally, we ask: does expressiveness matter in practice? We use a graph symmetry metric that quantifies the difficulty of distinguishing links and show that while expressive models may underperform on standard benchmarks, they significantly outperform simpler ones as symmetry increases, highlighting the need for dataset-aware model selection.
Authors: Gen Li, Yuchen Zhou, Yuting Wei, Yuxin Chen
Abstract: In this paper, we explore provable acceleration of diffusion models without any additional retraining. Focusing on the task of approximating a target data distribution in $\mathbb{R}^d$ to within $\varepsilon$ total-variation distance, we propose a principled, training-free sampling algorithm that requires only the order of $$ d^{1+2/K} \varepsilon^{-1/K} $$ score function evaluations (up to log factor) in the presence of accurate scores, where $K$ is an arbitrarily large fixed integer. This result applies to a broad class of target data distributions, without the need for assumptions such as smoothness or log-concavity. Our theory is robust vis-a-vis inexact score estimation, degrading gracefully as the score estimation error increases -- without demanding higher-order smoothness on the score estimates as assumed in previous work. The proposed algorithm draws insight from high-order ODE solvers, leveraging high-order Lagrange interpolation and successive refinement to approximate the integral derived from the probability flow ODE.
Authors: Paul Wachter, Lukas Niehaus, Julius Sch\"oning
Abstract: Synthetic data has emerged as a cost-effective alternative to real data for training artificial neural networks (ANN). However, the disparity between synthetic and real data results in a domain gap. That gap leads to poor performance and generalization of the trained ANN when applied to real-world scenarios. Several strategies have been developed to bridge this gap, which combine synthetic and real data, known as mixed training using hybrid datasets. While these strategies have been shown to mitigate the domain gap, a systematic evaluation of their generalizability and robustness across various tasks and architectures remains underexplored. To address this challenge, our study comprehensively analyzes two widely used mixing strategies on three prevalent architectures and three distinct hybrid datasets. From these datasets, we sample subsets with varying proportions of synthetic to real data to investigate the impact of synthetic and real components. The findings of this paper provide valuable insights into optimizing the use of synthetic data in the training process of any ANN, contributing to enhancing robustness and efficacy.
Authors: Yuqing Wang, Shangding Gu
Abstract: Data selection plays a crucial role in data-driven decision-making, including in large language models (LLMs), and is typically task-dependent. Properties such as data quality and diversity have been extensively studied and are known to enhance model performance. However, it remains unclear whether there exist other quantitative and general principles of data selection that can consistently improve performance, especially for complex tasks with limited prior knowledge. In this paper, we demonstrate that selecting more uniformly distributed data can improve training efficiency while enhancing performance. Specifically, we establish that more uniform (less biased) distribution leads to a larger minimum pairwise distance between data points, denoted by $h_{\min}$, and prove that a smaller $h_{\min}$ can slow down the training dynamics of gradient descent (GD). Moreover, we theoretically show that the approximation error of neural networks decreases as $h_{\min}$ increases. Our analysis introduces a convergence framework for GD beyond the Neural Tangent Kernel (NTK) regime, applicable to a broad class of architectures, including transformers, without requiring Lipschitz smoothness. This framework further provides theoretical justification for the use of residual connections and function compositions in deep neural architectures. In the end, we conduct comprehensive experiments for supervised fine-tuning across various settings, including different optimization strategies, model sizes, and training datasets. The results consistently demonstrate that selecting data by maximizing pairwise distance significantly accelerates training and achieves comparable or better performance in LLMs across diverse datasets. Code and Datasets are available at the link: https://github.com/SafeRL-Lab/data-uniformity.
Authors: Sixun Dong, Wei Fan, Teresa Wu, Yanjie Fu
Abstract: Time series forecasting traditionally relies on unimodal numerical inputs, which often struggle to capture high-level semantic patterns due to their dense and unstructured nature. While recent approaches have explored representing time series as text using large language models (LLMs), these methods remain limited by the discrete nature of token sequences and lack the perceptual intuition humans typically apply, such as interpreting visual patterns. In this paper, we propose a multimodal contrastive learning framework that transforms raw time series into structured visual and textual perspectives. Rather than using natural language or real-world images, we construct both modalities directly from numerical sequences. We then align these views in a shared semantic space via contrastive learning, enabling the model to capture richer and more complementary representations. Furthermore, we introduce a variate selection module that leverages the aligned representations to identify the most informative variables for multivariate forecasting. Extensive experiments on fifteen short-term and six long-term forecasting benchmarks demonstrate that our approach consistently outperforms strong unimodal and cross-modal baselines, highlighting the effectiveness of multimodal alignment in enhancing time series forecasting. Code is available at: https://github.com/Ironieser/TimesCLIP.
Authors: Louis Bradshaw, Simon Colton
Abstract: We introduce an extensive new dataset of MIDI files, created by transcribing audio recordings of piano performances into their constituent notes. The data pipeline we use is multi-stage, employing a language model to autonomously crawl and score audio recordings from the internet based on their metadata, followed by a stage of pruning and segmentation using an audio classifier. The resulting dataset contains over one million distinct MIDI files, comprising roughly 100,000 hours of transcribed audio. We provide an in-depth analysis of our techniques, offering statistical insights, and investigate the content by extracting metadata tags, which we also provide. Dataset available at https://github.com/loubbrad/aria-midi.
Authors: Sharique Hasan, Alexander Oettl, Sampsa Samila
Abstract: This paper introduces the Generality-Accuracy-Simplicity (GAS) framework to analyze how large language models (LLMs) are reshaping organizations and competitive strategy. We argue that viewing AI as a simple reduction in input costs overlooks two critical dynamics: (a) the inherent trade-offs among generality, accuracy, and simplicity, and (b) the redistribution of complexity across stakeholders. While LLMs appear to defy the traditional trade-off by offering high generality and accuracy through simple interfaces, this user-facing simplicity masks a significant shift of complexity to infrastructure, compliance, and specialized personnel. The GAS trade-off, therefore, does not disappear but is relocated from the user to the organization, creating new managerial challenges, particularly around accuracy in high-stakes applications. We contend that competitive advantage no longer stems from mere AI adoption, but from mastering this redistributed complexity through the design of abstraction layers, workflow alignment, and complementary expertise. This study advances AI strategy by clarifying how scalable cognition relocates complexity and redefines the conditions for technology integration.
Authors: Jens Winkler, Michael Denhard
Abstract: Since weather forecasts are fundamentally uncertain, reliable decision making requires information on the likelihoods of future weather scenarios. We explore the sensitivity of machine learning weather prediction (MLWP) using the 24h Pangu Weather ML model of Huawei to errors in the initial conditions with a specific kind of Singular Vector (SV) perturbations. Our Arnoldi-SV (A-SV) method does not need linear nor adjoint model versions and is applicable to numerical weather prediction (NWP) as well as MLWP. It observes error growth within a given optimization time window by iteratively applying a forecast model to perturbed model states. This creates a Krylov subspace, implicitly based on a matrix operator, which approximates the local error growth. Each iteration adds new dimensions to the Krylov space and its leading right SVs are expected to turn into directions of growing errors. We show that A-SV indeed finds dynamically meaningful perturbation patterns for the 24h Pangu Weather model, which grow right from the beginning of the forecast rollout. These perturbations describe local unstable modes and could be a basis to initialize MLWP ensembles. Since we start A-SV from random noise perturbations, the algorithm transforms noise into perturbations conditioned on a given reference state - a process that is akin to the denoising process of the generic diffusion based ML model of GenCast, therefor we briefly discuss similarities and differences.
Authors: Ana Luiza S. Tavares, Artur Pedro M. Neto, Francinaldo L. Gomes, Paul Rodrigo dos Reis, Arthur G. da Silva, Antonio P. Junior, Bruno D. Gomes
Abstract: Accurate intraoperative localization of the subthalamic nucleus (STN) is essential for the efficacy of Deep Brain Stimulation (DBS) in patients with Parkinson's disease. While microelectrode recordings (MERs) provide rich electrophysiological information during DBS electrode implantation, current localization practices often rely on subjective interpretation of signal features. In this study, we propose a quantitative framework that leverages nonlinear dynamics and entropy-based metrics to classify neural activity recorded inside versus outside the STN. MER data from three patients were preprocessed using a robust artifact correction pipeline, segmented, and labelled based on surgical annotations. A comprehensive set of recurrence quantification analysis, nonlinear, and entropy features were extracted from each segment. Multiple supervised classifiers were trained on every combination of feature domains using stratified 10-fold cross-validation, followed by statistical comparison using paired Wilcoxon signed-rank tests with Holm-Bonferroni correction. The combination of entropy and nonlinear features yielded the highest discriminative power, and the Extra Trees classifier emerged as the best model with a cross-validated F1-score of 0.902+/-0.027 and ROC AUC of 0.887+/-0.055. Final evaluation on a 20% hold-out test set confirmed robust generalization (F1= 0.922, ROC AUC = 0.941). These results highlight the potential of nonlinear and entropy signal descriptors in supporting real-time, data-driven decision-making during DBS surgeries
Authors: Dung Truong, Arnaud Delorme
Abstract: Normalization is a critical yet often overlooked component in the preprocessing pipeline for EEG deep learning applications. The rise of large-scale pretraining paradigms such as self-supervised learning (SSL) introduces a new set of tasks whose nature is substantially different from supervised training common in EEG deep learning applications. This raises new questions about optimal normalization strategies for the applicable task. In this study, we systematically evaluate the impact of normalization granularity (recording vs. window level) and scope (cross-channel vs. within-channel) on both supervised (age and gender prediction) and self-supervised (Contrastive Predictive Coding) tasks. Using high-density resting-state EEG from 2,836 subjects in the Healthy Brain Network dataset, we show that optimal normalization strategies differ significantly between training paradigms. Window-level within-channel normalization yields the best performance in supervised tasks, while minimal or cross-channel normalization at the window level is more effective for SSL. These results underscore the necessity of task-specific normalization choices and challenge the assumption that a universal normalization strategy can generalize across learning settings. Our findings provide practical insights for developing robust EEG deep learning pipelines as the field shifts toward large-scale, foundation model training.
Authors: Wending Heng, Chaoyuan Liang, Yihui Zhao, Zhiqiang Zhang, Glen Cooper, Zhenhong Li
Abstract: Accurately decoding human motion intentions from surface electromyography (sEMG) is essential for myoelectric control and has wide applications in rehabilitation robotics and assistive technologies. However, existing sEMG-based motion estimation methods often rely on subject-specific musculoskeletal (MSK) models that are difficult to calibrate, or purely data-driven models that lack physiological consistency. This paper introduces a novel Physics-Embedded Neural Network (PENN) that combines interpretable MSK forward-dynamics with data-driven residual learning, thereby preserving physiological consistency while achieving accurate motion estimation. The PENN employs a recursive temporal structure to propagate historical estimates and a lightweight convolutional neural network for residual correction, leading to robust and temporally coherent estimations. A two-phase training strategy is designed for PENN. Experimental evaluations on six healthy subjects show that PENN outperforms state-of-the-art baseline methods in both root mean square error (RMSE) and $R^2$ metrics.
Authors: Weizhi Gao, Zhichao Hou, Junqi Yin, Feiyi Wang, Linyu Peng, Xiaorui Liu
Abstract: Diffusion models have emerged as powerful generative models, but their high computation cost in iterative sampling remains a significant bottleneck. In this work, we present an in-depth and insightful study of state-of-the-art acceleration techniques for diffusion models, including caching and quantization, revealing their limitations in computation error and generation quality. To break these limits, this work introduces Modulated Diffusion (MoDiff), an innovative, rigorous, and principled framework that accelerates generative modeling through modulated quantization and error compensation. MoDiff not only inherents the advantages of existing caching and quantization methods but also serves as a general framework to accelerate all diffusion models. The advantages of MoDiff are supported by solid theoretical insight and analysis. In addition, extensive experiments on CIFAR-10 and LSUN demonstrate that MoDiff significant reduces activation quantization from 8 bits to 3 bits without performance degradation in post-training quantization (PTQ). Our code implementation is available at https://github.com/WeizhiGao/MoDiff.
Authors: A. Subedi, S. De, L. Cavuoto, S. Schwaitzberg, M. Hackett, J. Norfleet
Abstract: Objective skill assessment in high-stakes procedural environments requires models that not only decode underlying cognitive and motor processes but also generalize across tasks, individuals, and experimental contexts. While prior work has demonstrated the potential of functional near-infrared spectroscopy (fNIRS) for evaluating cognitive-motor performance, existing approaches are often task-specific, rely on extensive preprocessing, and lack robustness to new procedures or conditions. Here, we introduce an interpretable transformer-based foundation model trained on minimally processed fNIRS signals for cross-procedural skill assessment. Pretrained using self-supervised learning on data from laparoscopic surgical tasks and endotracheal intubation (ETI), the model achieves greater than 88% classification accuracy on all tasks, with Matthews Correlation Coefficient exceeding 0.91 on ETI. It generalizes to a novel emergency airway procedure--cricothyrotomy--using fewer than 30 labeled samples and a lightweight (less than 2k parameter) adapter module, attaining an AUC greater than 87%. Interpretability is achieved via a novel channel attention mechanism--developed specifically for fNIRS--that identifies functionally coherent prefrontal sub-networks validated through ablation studies. Temporal attention patterns align with task-critical phases and capture stress-induced changes in neural variability, offering insight into dynamic cognitive states.
Authors: Krisanu Sarkar
Abstract: We introduce Hindsight-Guided Momentum (HGM), a first-order optimization algorithm that adaptively scales learning rates based on the directional consistency of recent updates. Traditional adaptive methods, such as Adam or RMSprop , adapt learning dynamics using only the magnitude of gradients, often overlooking important geometric cues.Geometric cues refer to directional information, such as the alignment between current gradients and past updates, which reflects the local curvature and consistency of the optimization path. HGM addresses this by incorporating a hindsight mechanism that evaluates the cosine similarity between the current gradient and accumulated momentum. This allows it to distinguish between coherent and conflicting gradient directions, increasing the learning rate when updates align and reducing it in regions of oscillation or noise. The result is a more responsive optimizer that accelerates convergence in smooth regions of the loss surface while maintaining stability in sharper or more erratic areas. Despite this added adaptability, the method preserves the computational and memory efficiency of existing optimizers.By more intelligently responding to the structure of the optimization landscape, HGM provides a simple yet effective improvement over existing approaches, particularly in non-convex settings like that of deep neural network training.
Authors: Mariam Yahya, Aydin Sezgin, Setareh Maghsudi
Abstract: As users in small cell networks increasingly rely on computation-intensive services, cloud-based access often results in high latency. Multi-access edge computing (MEC) mitigates this by bringing computational resources closer to end users, with small base stations (SBSs) serving as edge servers to enable low-latency service delivery. However, limited edge capacity makes it challenging to decide which services to deploy locally versus in the cloud, especially under unknown service demand and dynamic network conditions. To tackle this problem, we model service demand as a linear function of service attributes and formulate the service placement task as a linear bandit problem, where SBSs act as agents and services as arms. The goal is to identify the service that, when placed at the edge, offers the greatest reduction in total user delay compared to cloud deployment. We propose a distributed and adaptive multi-agent best-arm identification (BAI) algorithm under a fixed-confidence setting, where SBSs collaborate to accelerate learning. Simulations show that our algorithm identifies the optimal service with the desired confidence and achieves near-optimal speedup, as the number of learning rounds decreases proportionally with the number of SBSs. We also provide theoretical analysis of the algorithm's sample complexity and communication overhead.
Authors: Xi Fu, Weibang Jiang, Rui Liu, Gernot R. M\"uller-Putz, Cuntai Guan
Abstract: Accurate decoding of lower-limb motion from EEG signals is essential for advancing brain-computer interface (BCI) applications in movement intent recognition and control. However, challenges persist in achieving causal, phase-consistent predictions and in modeling both inter- and intra-subject variability. To address these issues, we propose NeuroDyGait, a domain-generalizable EEG-to-motion decoding framework that leverages structured contrastive representation learning and relational domain modeling. The proposed method employs relative contrastive learning to achieve semantic alignment between EEG and motion embeddings. Furthermore, a multi-cycle gait reconstruction objective is introduced to enforce temporal coherence and maintain biomechanical consistency. To promote inter-session generalization, during fine-tuning, a domain dynamic decoding mechanism adaptively assigns session-specific prediction heads and learns to mix their outputs based on inter-session relationships. NeuroDyGait enables zero-shot motion prediction for unseen individuals without requiring adaptation and achieves superior performance in cross-subject gait decoding on benchmark datasets. Additionally, it demonstrates strong phase-detection capabilities even without explicit phase supervision during training. These findings highlight the potential of relational domain learning in enabling scalable, target-free deployment of BCIs.
Authors: Zhenke Duan, Jiqun Pan, Jiani Tu
Abstract: Accurate detection of ethylene concentrations in mixed gases is crucial in chemical production for safety and health purposes. Traditional methods are hindered by high cost and complexity, limiting their practical application. This study proposes MENGLAN, a Multiscale Enhanced Nonparametric Gas Analyzer that integrates a dual-stream structure, a Hybrid Multi-Head Attention mechanism, and a Feature Reactivation Module to enable real-time, lightweight, and high-precision ethylene concentration prediction. Results show that MENGLAN achieves superior performance, reduced computational demand, and enhanced deployability compared to existing methods.
Authors: Sadia Kamal, Lalu Prasad Yadav Prakash, S M Rafiuddin, Mohammed Rakib, Arunkumar Bagavathi, Atriya Sen, Sagnik Ray Choudhury
Abstract: Political Compass Test (PCT) or similar questionnaires have been used to quantify LLM's political leanings. Building on a recent line of work that examines the validity of PCT tests, we demonstrate that variation in standard generation parameters does not significantly impact the models' PCT scores. However, external factors such as prompt variations and fine-tuning individually and in combination affect the same. Finally, we demonstrate that when models are fine-tuned on text datasets with higher political content than others, the PCT scores are not differentially affected. This calls for a thorough investigation into the validity of PCT and similar tests, as well as the mechanism by which political leanings are encoded in LLMs.
Authors: Shihong Ling, Yue Wan, Xiaowei Jia, Na Du
Abstract: This paper introduces a new framework, DriveBLIP2, built upon the BLIP2-OPT architecture, to generate accurate and contextually relevant explanations for emerging driving scenarios. While existing vision-language models perform well in general tasks, they encounter difficulties in understanding complex, multi-object environments, particularly in real-time applications such as autonomous driving, where the rapid identification of key objects is crucial. To address this limitation, an Attention Map Generator is proposed to highlight significant objects relevant to driving decisions within critical video frames. By directing the model's focus to these key regions, the generated attention map helps produce clear and relevant explanations, enabling drivers to better understand the vehicle's decision-making process in critical situations. Evaluations on the DRAMA dataset reveal significant improvements in explanation quality, as indicated by higher BLEU, ROUGE, CIDEr, and SPICE scores compared to baseline models. These findings underscore the potential of targeted attention mechanisms in vision-language models for enhancing explainability in real-time autonomous driving.
Authors: He-Yang Xu, Hongxiang Gao, Yuwen Li, Xiu-Shen Wei, Chengyu Liu
Abstract: The diagnostic value of electrocardiogram (ECG) lies in its dynamic characteristics, ranging from rhythm fluctuations to subtle waveform deformations that evolve across time and frequency domains. However, supervised ECG models tend to overfit dominant and repetitive patterns, overlooking fine-grained but clinically critical cues, a phenomenon known as Simplicity Bias (SB), where models favor easily learnable signals over subtle but informative ones. In this work, we first empirically demonstrate the presence of SB in ECG analyses and its negative impact on diagnostic performance, while simultaneously discovering that self-supervised learning (SSL) can alleviate it, providing a promising direction for tackling the bias. Following the SSL paradigm, we propose a novel method comprising two key components: 1) Temporal-Frequency aware Filters to capture temporal-frequency features reflecting the dynamic characteristics of ECG signals, and 2) building on this, Multi-Grained Prototype Reconstruction for coarse and fine representation learning across dual domains, further mitigating SB. To advance SSL in ECG analyses, we curate a large-scale multi-site ECG dataset with 1.53 million recordings from over 300 clinical centers. Experiments on three downstream tasks across six ECG datasets demonstrate that our method effectively reduces SB and achieves state-of-the-art performance. Code and dataset will be released publicly.
Authors: Michiel Schepers, Pieter Robberechts, Jan Van Haaren, Jesse Davis
Abstract: Data analysis plays an increasingly important role in soccer, offering new ways to evaluate individual and team performance. One specific application is the evaluation of dribbles: one-on-one situations where an attacker attempts to bypass a defender with the ball. While previous research has primarily relied on 2D positional tracking data, this fails to capture aspects like balance, orientation, and ball control, limiting the depth of current insights. This study explores how pose tracking data (capturing players' posture and movement in three dimensions) can improve our understanding of dribbling skills. We extract novel pose-based features from 1,736 dribbles in the 2022/23 Champions League season and evaluate their impact on dribble success. Our results indicate that features capturing the attacker's balance and the alignment of the orientation between the attacker and defender are informative for predicting dribble success. Incorporating these pose-based features on top of features derived from traditional 2D positional data leads to a measurable improvement in model performance.
Authors: Hassan Baker, Austin J. Brockmeier
Abstract: Detecting brain lesions as abnormalities observed in magnetic resonance imaging (MRI) is essential for diagnosis and treatment. In the search of abnormalities, such as tumors and malformations, radiologists may benefit from computer-aided diagnostics that use computer vision systems trained with machine learning to segment normal tissue from abnormal brain tissue. While supervised learning methods require annotated lesions, we propose a new unsupervised approach (Patch2Loc) that learns from normal patches taken from structural MRI. We train a neural network model to map a patch back to its spatial location within a slice of the brain volume. During inference, abnormal patches are detected by the relatively higher error and/or variance of the location prediction. This generates a heatmap that can be integrated into pixel-wise methods to achieve finer-grained segmentation. We demonstrate the ability of our model to segment abnormal brain tissues by applying our approach to the detection of tumor tissues in MRI on T2-weighted images from BraTS2021 and MSLUB datasets and T1-weighted images from ATLAS and WMH datasets. We show that it outperforms the state-of-the art in unsupervised segmentation. The codebase for this work can be found on our \href{https://github.com/bakerhassan/Patch2Loc}{GitHub page}.
Authors: Hassan Baker, Matthew S. Emigh, Austin J. Brockmeier
Abstract: As a computer vision task, automatic object segmentation remains challenging in specialized image domains without massive labeled data, such as synthetic aperture sonar images, remote sensing, biomedical imaging, etc. In any domain, obtaining pixel-wise segmentation masks is expensive. In this work, we propose a method for training a masking network to perform binary object segmentation using weak supervision in the form of image-wise presence or absence of an object of interest, which provides less information but may be obtained more quickly from manual or automatic labeling. A key step in our method is that the segmented objects can be placed into background-only images to create realistic, images of the objects with counterfactual backgrounds. To create a contrast between the original and counterfactual background images, we propose to first cluster the background-only images, and then during learning create counterfactual images that blend objects segmented from their original source backgrounds to backgrounds chosen from a targeted cluster. One term in the training loss is the divergence between these counterfactual images and the real object images with backgrounds of the target cluster. The other term is a supervised loss for background-only images. While an adversarial critic could provide the divergence, we use sample-based divergences. We conduct experiments on side-scan and synthetic aperture sonar in which our approach succeeds compared to previous unsupervised segmentation baselines that were only tested on natural images. Furthermore, to show generality we extend our experiments to natural images, obtaining reasonable performance with our method that avoids pretrained networks, generative networks, and adversarial critics. The basecode for this work can be found at \href{GitHub}{https://github.com/bakerhassan/WSOS}.
Authors: Mark Wrobel (UCL Centre for Translational Cardiovascular Imaging, University College London), Michele Pascale (UCL Centre for Translational Cardiovascular Imaging, University College London), Tina Yao (UCL Centre for Translational Cardiovascular Imaging, University College London), Ruaraidh Campbell (UCL Centre for Translational Cardiovascular Imaging, University College London), Elena Milano (Great Ormond Street Hospital), Michael Quail (UCL Centre for Translational Cardiovascular Imaging, University College London, Great Ormond Street Hospital), Jennifer Steeden (UCL Centre for Translational Cardiovascular Imaging, University College London), Vivek Muthurangu (UCL Centre for Translational Cardiovascular Imaging, University College London)
Abstract: Background: Conventional cardiovascular magnetic resonance (CMR) in paediatric and congenital heart disease uses 2D, breath-hold, balanced steady state free precession (bSSFP) cine imaging for assessment of function and cardiac-gated, respiratory-navigated, static 3D bSSFP whole-heart imaging for anatomical assessment. Our aim is to concatenate a stack 2D free-breathing real-time cines and use Deep Learning (DL) to create an isotropic a fully segmented 3D cine dataset from these images. Methods: Four DL models were trained on open-source data that performed: a) Interslice contrast correction; b) Interslice respiratory motion correction; c) Super-resolution (slice direction); and d) Segmentation of right and left atria and ventricles (RA, LA, RV, and LV), thoracic aorta (Ao) and pulmonary arteries (PA). In 10 patients undergoing routine cardiovascular examination, our method was validated on prospectively acquired sagittal stacks of real-time cine images. Quantitative metrics (ventricular volumes and vessel diameters) and image quality of the 3D cines were compared to conventional breath hold cine and whole heart imaging. Results: All real-time data were successfully transformed into 3D cines with a total post-processing time of <1 min in all cases. There were no significant biases in any LV or RV metrics with reasonable limits of agreement and correlation. There is also reasonable agreement for all vessel diameters, although there was a small but significant overestimation of RPA diameter. Conclusion: We have demonstrated the potential of creating a 3D-cine data from concatenated 2D real-time cine images using a series of DL models. Our method has short acquisition and reconstruction times with fully segmented data being available within 2 minutes. The good agreement with conventional imaging suggests that our method could help to significantly speed up CMR in clinical practice.
Authors: Yu Zhang, Shanshan Zhao, Bokui Wan, Jinjuan Wang, Xiaodong Yan
Abstract: Detecting a minor average treatment effect is a major challenge in large-scale applications, where even minimal improvements can have a significant economic impact. Traditional methods, reliant on normal distribution-based or expanded statistics, often fail to identify such minor effects because of their inability to handle small discrepancies with sufficient sensitivity. This work leverages a counterfactual outcome framework and proposes a maximum probability-driven two-armed bandit (TAB) process by weighting the mean volatility statistic, which controls Type I error. The implementation of permutation methods further enhances the robustness and efficacy. The established strategic central limit theorem (SCLT) demonstrates that our approach yields a more concentrated distribution under the null hypothesis and a less concentrated one under the alternative hypothesis, greatly improving statistical power. The experimental results indicate a significant improvement in the A/B testing, highlighting the potential to reduce experimental costs while maintaining high statistical power.
Authors: Fabrizio Falasca
Abstract: This work explores key conceptual limitations in data-driven modeling of multiscale dynamical systems, focusing on neural emulators and stochastic climate modeling. A skillful climate model should capture both stationary statistics and responses to external perturbations. While current autoregressive neural models often reproduce the former, they typically struggle with the latter. We begin by analyzing a low-dimensional dynamical system to expose, by analogy, fundamental limitations that persist in high-dimensional settings. Specifically, we construct neural stochastic models under two scenarios: one where the full state vector is observed, and another with only partial observations (i.e. a subset of variables). In the first case, the models accurately capture both equilibrium statistics and forced responses in ensemble mean and variance. In the more realistic case of partial observations, two key challenges emerge: (i) identifying the \textit{proper} variables to model, and (ii) parameterizing the influence of unobserved degrees of freedom. These issues are not specific to neural networks but reflect fundamental limitations of data-driven modeling and the need to target the slow dynamics of the system. We argue that physically grounded strategies -- such as coarse-graining and stochastic parameterizations -- are critical, both conceptually and practically, for the skillful emulation of complex systems like the coupled climate system. Building on these insights, we turn to a more realistic application: a stochastic reduced neural model of the sea surface temperature field and the net radiative flux at the top of the atmosphere, assessing its stationary statistics, response to temperature forcing, and interpretability.
Authors: Callum Duffy, Marcin Jastrzebski
Abstract: In this work, we investigate the phenomenon of spectral bias in quantum machine learning, where, in classical settings, models tend to fit low-frequency components of a target function earlier during training than high-frequency ones, demonstrating a frequency-dependent rate of convergence. We study this effect specifically in parameterised quantum circuits (PQCs). Leveraging the established formulation of PQCs as Fourier series, we prove that spectral bias in this setting arises from the ``redundancy'' of the Fourier coefficients, which denotes the number of terms in the analytical form of the model contributing to the same frequency component. The choice of data encoding scheme dictates the degree of redundancy for a Fourier coefficient. We find that the magnitude of the Fourier coefficients' gradients during training strongly correlates with the coefficients' redundancy. We then further demonstrate this empirically with three different encoding schemes. Additionally, we demonstrate that PQCs with greater redundancy exhibit increased robustness to random perturbations in their parameters at the corresponding frequencies. We investigate how design choices affect the ability of PQCs to learn Fourier sums, focusing on parameter initialization scale and entanglement structure, finding large initializations and low-entanglement schemes tend to slow convergence.
Authors: Boyuan Chen, Minghao Shao, Abdul Basit, Siddharth Garg, Muhammad Shafique
Abstract: The growing capabilities of large language models (LLMs) have exposed them to increasingly sophisticated jailbreak attacks. Among these, obfuscation-based attacks -- which encrypt malicious content to evade detection -- remain highly effective. By leveraging the reasoning ability of advanced LLMs to interpret encrypted prompts, such attacks circumvent conventional defenses that rely on keyword detection or context filtering. These methods are very difficult to defend against, as existing safety mechanisms are not designed to interpret or decode ciphered content. In this work, we propose \textbf{MetaCipher}, a novel obfuscation-based jailbreak framework, along with a reinforcement learning-based dynamic cipher selection mechanism that adaptively chooses optimal encryption strategies from a cipher pool. This approach enhances jailbreak effectiveness and generalizability across diverse task types, victim LLMs, and safety guardrails. Our framework is modular and extensible by design, supporting arbitrary cipher families and accommodating evolving adversarial strategies. We complement our method with a large-scale empirical analysis of cipher performance across multiple victim LLMs. Within as few as 10 queries, MetaCipher achieves over 92\% attack success rate (ASR) on most recent standard malicious prompt benchmarks against state-of-the-art non-reasoning LLMs, and over 74\% ASR against reasoning-capable LLMs, outperforming all existing obfuscation-based jailbreak methods. These results highlight the long-term robustness and adaptability of our approach, making it more resilient than prior methods in the face of advancing safety measures.
Authors: Guan-Horng Liu, Jaemoo Choi, Yongxin Chen, Benjamin Kurt Miller, Ricky T. Q. Chen
Abstract: Computational methods for learning to sample from the Boltzmann distribution -- where the target distribution is known only up to an unnormalized energy function -- have advanced significantly recently. Due to the lack of explicit target samples, however, prior diffusion-based methods, known as diffusion samplers, often require importance-weighted estimation or complicated learning processes. Both trade off scalability with extensive evaluations of the energy and model, thereby limiting their practical usage. In this work, we propose Adjoint Schr\"odinger Bridge Sampler (ASBS), a new diffusion sampler that employs simple and scalable matching-based objectives yet without the need to estimate target samples during training. ASBS is grounded on a mathematical model -- the Schr\"odinger Bridge -- which enhances sampling efficiency via kinetic-optimal transportation. Through a new lens of stochastic optimal control theory, we demonstrate how SB-based diffusion samplers can be learned at scale via Adjoint Matching and prove convergence to the global solution. Notably, ASBS generalizes the recent Adjoint Sampling (Havens et al., 2025) to arbitrary source distributions by relaxing the so-called memoryless condition that largely restricts the design space. Through extensive experiments, we demonstrate the effectiveness of ASBS on sampling from classical energy functions, amortized conformer generation, and molecular Boltzmann distributions.
Authors: Osama Zafar, Mina Namazi, Yuqiao Xu, Youngjin Yoo, Erman Ayday
Abstract: In the current paradigm of digital personalized services, the centralized management of personal data raises significant privacy concerns, security vulnerabilities, and diminished individual autonomy over sensitive information. Despite their efficiency, traditional centralized architectures frequently fail to satisfy rigorous privacy requirements and expose users to data breaches and unauthorized access risks. This pressing challenge calls for a fundamental paradigm shift in methodologies for collecting, storing, and utilizing personal data across diverse sectors, including education, healthcare, and finance. This paper introduces a novel decentralized, privacy-preserving architecture that handles heterogeneous personal information, ranging from educational credentials to health records and financial data. Unlike traditional models, our system grants users complete data ownership and control, allowing them to selectively share information without compromising privacy. The architecture's foundation comprises advanced privacy-enhancing technologies, including secure enclaves and federated learning, enabling secure computation, verification, and data sharing. The system supports diverse functionalities, including local computation, model training, and privacy-preserving data sharing, while ensuring data credibility and robust user privacy.
Authors: Daniel Ciganda, Ignacio Camp\'on, I\~naki Permanyer, Jakob H Macke
Abstract: While age-specific fertility rates (ASFRs) provide the most extensive record of reproductive change, their aggregate nature masks the underlying behavioral mechanisms that ultimately drive fertility trends. To recover these mechanisms, we develop a likelihood-free Bayesian framework that couples an individual-level model of the reproductive process with Sequential Neural Posterior Estimation (SNPE). This allows us to infer eight behavioral and biological parameters from just two aggregate series: ASFRs and the age-profile of planned versus unplanned births. Applied to U.S. National Survey of Family Growth cohorts and to Demographic and Health Survey cohorts from Colombia, the Dominican Republic, and Peru, the method reproduces observed fertility schedules and, critically, predicts out-of-sample micro-level distributions of age at first sex, inter-birth intervals, and family-size ideals, none of which inform the estimation step. Because the fitted model yields complete synthetic life histories, it enables behaviorally explicit population forecasts and supports the construction of demographic digital twins.
Authors: Yuming Ma
Abstract: Extending Buehler et al.'s 2019 Deep Hedging paradigm, we innovatively employ deep neural networks to parameterize convex-risk minimization (CVaR/ES) for the portfolio tail-risk hedging problem. Through comprehensive numerical experiments on crisis-era bootstrap market simulators -- customizable with transaction costs, risk budgets, liquidity constraints, and market impact -- our end-to-end framework not only achieves significant one-day 99% CVaR reduction but also yields practical insights into friction-aware strategy adaptation, demonstrating robustness and operational viability in realistic markets.
Authors: Gabriel M. Mejia, Henry E. Miller, Francis J. A. Leblanc, Bo Wang, Brendan Swain, Lucas Paulo de Lima Camillo
Abstract: Recent benchmarks reveal that models for single-cell perturbation response are often outperformed by simply predicting the dataset mean. We trace this anomaly to a metric artifact: control-referenced deltas and unweighted error metrics reward mode collapse whenever the control is biased or the biological signal is sparse. Large-scale \textit{in silico} simulations and analysis of two real-world perturbation datasets confirm that shared reference shifts, not genuine biological change, drives high performance in these evaluations. We introduce differentially expressed gene (DEG)-aware metrics, weighted mean-squared error (WMSE) and weighted delta $R^{2}$ ($R^{2}_{w}(\Delta)$) with respect to all perturbations, that measure error in niche signals with high sensitivity. We further introduce negative and positive performance baselines to calibrate these metrics. With these improvements, the mean baseline sinks to null performance while genuine predictors are correctly rewarded. Finally, we show that using WMSE as a loss function reduces mode collapse and improves model performance.
Authors: Pedro R. Pires, Tiago A. Almeida
Abstract: Over the past decade, recommender systems have experienced a surge in popularity. Despite notable progress, they grapple with challenging issues, such as high data dimensionality and sparseness. Representing users and items as low-dimensional embeddings learned via neural networks has become a leading solution. However, while recent studies show promising results, many approaches rely on complex architectures or require content data, which may not always be available. This paper presents Interact2Vec, a novel neural network-based model that simultaneously learns distributed embeddings for users and items while demanding only implicit feedback. The model employs state-of-the-art strategies that natural language processing models commonly use to optimize the training phase and enhance the final embeddings. Two types of experiments were conducted regarding the extrinsic and intrinsic quality of the model. In the former, we benchmarked the recommendations generated by Interact2Vec's embeddings in a top-$N$ ranking problem, comparing them with six other recommender algorithms. The model achieved the second or third-best results in 30\% of the datasets, being competitive with other recommenders, and has proven to be very efficient with an average training time reduction of 274\% compared to other embedding-based models. Later, we analyzed the intrinsic quality of the embeddings through similarity tables. Our findings suggest that Interact2Vec can achieve promising results, especially on the extrinsic task, and is an excellent embedding-generator model for scenarios of scarce computing resources, enabling the learning of item and user embeddings simultaneously and efficiently.
Authors: Anamika Lochab, Lu Yan, Patrick Pynadath, Xiangyu Zhang, Ruqi Zhang
Abstract: The rise of API-only access to state-of-the-art LLMs highlights the need for effective black-box jailbreak methods to identify model vulnerabilities in real-world settings. Without a principled objective for gradient-based optimization, most existing approaches rely on genetic algorithms, which are limited by their initialization and dependence on manually curated prompt pools. Furthermore, these methods require individual optimization for each prompt, failing to provide a comprehensive characterization of model vulnerabilities. To address this gap, we introduce VERA: Variational infErence fRamework for jAilbreaking. VERA casts black-box jailbreak prompting as a variational inference problem, training a small attacker LLM to approximate the target LLM's posterior over adversarial prompts. Once trained, the attacker can generate diverse, fluent jailbreak prompts for a target query without re-optimization. Experimental results show that VERA achieves strong performance across a range of target LLMs, highlighting the value of probabilistic inference for adversarial prompt generation.
Authors: Luhuan Wu, Mingzhang Yin, Yixin Wang, John P. Cunningham, David M. Blei
Abstract: Invariant prediction [Peters et al., 2016] analyzes feature/outcome data from multiple environments to identify invariant features - those with a stable predictive relationship to the outcome. Such features support generalization to new environments and help reveal causal mechanisms. Previous methods have primarily tackled this problem through hypothesis testing or regularized optimization. Here we develop Bayesian Invariant Prediction (BIP), a probabilistic model for invariant prediction. BIP encodes the indices of invariant features as a latent variable and recover them by posterior inference. Under the assumptions of Peters et al. [2016], the BIP posterior targets the true invariant features. We prove that the posterior is consistent and that greater environment heterogeneity leads to faster posterior contraction. To handle many features, we design an efficient variational approximation called VI-BIP. In simulations and real data, we find that BIP and VI-BIP are more accurate and scalable than existing methods for invariant prediction.
Authors: Shi Jie Yu
Abstract: This paper studies theoretical lower bounds for estimating the trace of a matrix function, $\text{tr}(f(A))$, focusing on methods that use Hutchinson's method along with Block Krylov techniques. These methods work by approximating matrix-vector products like $f(A)V$ using a Block Krylov subspace. This is closely related to approximating functions with polynomials. We derive theoretical upper bounds on how many Krylov steps are needed for functions such as $A^{-1/2}$ and $A^{-1}$ by analyzing the upper bounds from the polynomial approximation of their scalar equivalent. In addition, we also develop lower limits on the number of queries needed for trace estimation, specifically for $\text{tr}(W^{-p})$ where $W$ is a Wishart matrix. Our study clarifies the connection between the number of steps in Block Krylov methods and the degree of the polynomial used for approximation. This links the total cost of trace estimation to basic limits in polynomial approximation and how much information is needed for the computation.
Authors: Jinliang Shi, Shigang Li, Youxuan Xu, Xueying Wang, Rongtian Fu, Zhi Ma, Tong Wu
Abstract: Sparse matrix multiplication operators (i.e., SpMM and SDDMM) are widely used in deep learning and scientific computing. Modern accelerators are commonly equipped with Tensor cores and CUDA cores to accelerate sparse operators. The former brings superior computing power but only for structured matrix multiplication, while the latter has relatively lower performance but with higher programming flexibility. In this work, we discover that utilizing one resource alone leads to inferior performance for sparse matrix multiplication, due to their respective limitations. To this end, we propose Libra, a systematic approach that enables synergistic computation between CUDA and Tensor cores to achieve the best performance for sparse matrix multiplication. Specifically, we propose a 2D-aware workload distribution strategy to find out the sweet point of task mapping for different sparse operators, leveraging both the high performance of Tensor cores and the low computational redundancy on CUDA cores. In addition, Libra incorporates systematic optimizations for heterogeneous computing, including hybrid load-balancing, finely optimized kernel implementations, and GPU-accelerated preprocessing. Extensive experimental results on H100 and RTX 4090 GPUs show that Libra outperforms the state-of-the-art by on average 3.1x (up to 9.23x) over DTC-SpMM and 2.9x (up to 3.9x) for end-to-end GNN applications. Libra opens up a new perspective for sparse operator acceleration by fully exploiting the heterogeneous computing resources on GPUs.
Authors: Yu Zhang, Xi Zhang, Hualin zhou, Xinyuan Chen, Shang Gao, Hong Jia, Jianfei Yang, Yuankai Qi, Tao Gu
Abstract: Deep learning for human sensing on edge systems offers significant opportunities for smart applications. However, its training and development are hindered by the limited availability of sensor data and resource constraints of edge systems. Current methods that rely on transferring pre-trained models often encounter issues such as modality shift and high resource demands, resulting in substantial accuracy loss, resource overhead, and poor adaptability across different sensing applications. In this paper, we propose XTransfer, a first-of-its-kind method for resource-efficient, modality-agnostic model transfer. XTransfer freely leverages single or multiple pre-trained models and transfers knowledge across different modalities by (i) model repairing that safely repairs modality shift in pre-trained model layers with only few sensor data, and (ii) layer recombining that efficiently searches and recombines layers of interest from source models in a layer-wise manner to create compact models. We benchmark various baselines across diverse human sensing datasets spanning different modalities. Comprehensive results demonstrate that XTransfer achieves state-of-the-art performance on human sensing tasks while significantly reducing the costs of sensor data collection, model training, and edge deployment.
Authors: Honglin Bao, Kai Li
Abstract: Persistence is often regarded as a virtue in science. In this paper, however, we challenge this conventional view by highlighting its contextual nature, particularly how persistence can become a liability during periods of paradigm shift. We focus on the deep learning revolution catalyzed by AlexNet in 2012. Analyzing the 20-year career trajectories of over 5,000 scientists who were active in top machine learning venues during the preceding decade, we examine how their research focus and output evolved. We first uncover a dynamic period in which leading venues increasingly prioritized cutting-edge deep learning developments that displaced relatively traditional statistical learning methods. Scientists responded to these changes in markedly different ways. Those who were previously successful or affiliated with old teams adapted more slowly, experiencing what we term a rigidity penalty - a reluctance to embrace new directions leading to a decline in scientific impact, as measured by citation percentile rank. In contrast, scientists who pursued strategic adaptation - selectively pivoting toward emerging trends while preserving weak connections to prior expertise - reaped the greatest benefits. Taken together, our macro- and micro-level findings show that scientific breakthroughs act as mechanisms that reconfigure power structures within a field.
Authors: Fiona Xiao Jingyi, Lili Liu
Abstract: Forecasting central bank policy decisions remains a persistent challenge for investors, financial institutions, and policymakers due to the wide-reaching impact of monetary actions. In particular, anticipating shifts in the U.S. federal funds rate is vital for risk management and trading strategies. Traditional methods relying only on structured macroeconomic indicators often fall short in capturing the forward-looking cues embedded in central bank communications. This study examines whether predictive accuracy can be enhanced by integrating structured data with unstructured textual signals from Federal Reserve communications. We adopt a multi-modal framework, comparing traditional machine learning models, transformer-based language models, and deep learning architectures in both unimodal and hybrid settings. Our results show that hybrid models consistently outperform unimodal baselines. The best performance is achieved by combining TF-IDF features of FOMC texts with economic indicators in an XGBoost classifier, reaching a test AUC of 0.83. FinBERT-based sentiment features marginally improve ranking but perform worse in classification, especially under class imbalance. SHAP analysis reveals that sparse, interpretable features align more closely with policy-relevant signals. These findings underscore the importance of integrating textual and structured signals transparently. For monetary policy forecasting, simpler hybrid models can offer both accuracy and interpretability, delivering actionable insights for researchers and decision-makers.
Authors: Yanran Wu, Inez Hua, Yi Ding
Abstract: Water consumption is an increasingly critical dimension of computing sustainability, especially as AI workloads rapidly scale. However, current water impact assessment often overlooks where and when water stress is more severe. To fill in this gap, we present SCARF, the first general framework that evaluates water impact of computing by factoring in both spatial and temporal variations in water stress. SCARF calculates an Adjusted Water Impact (AWI) metric that considers both consumption volume and local water stress over time. Through three case studies on LLM serving, datacenters, and semiconductor fabrication plants, we show the hidden opportunities for reducing water impact by optimizing location and time choices, paving the way for water-sustainable computing. The code is available at https://github.com/jojacola/SCARF.
Authors: Minchao Jiang, Shunyu Jia, Jiaming Gu, Xiaoyuan Lu, Guangming Zhu, Anqi Dong, Liang Zhang
Abstract: 3D Gaussian Splatting (3DGS) has become horsepower in high-quality, real-time rendering for novel view synthesis of 3D scenes. However, existing methods focus primarily on geometric and appearance modeling, lacking deeper scene understanding while also incurring high training costs that complicate the originally streamlined differentiable rendering pipeline. To this end, we propose VoteSplat, a novel 3D scene understanding framework that integrates Hough voting with 3DGS. Specifically, Segment Anything Model (SAM) is utilized for instance segmentation, extracting objects, and generating 2D vote maps. We then embed spatial offset vectors into Gaussian primitives. These offsets construct 3D spatial votes by associating them with 2D image votes, while depth distortion constraints refine localization along the depth axis. For open-vocabulary object localization, VoteSplat maps 2D image semantics to 3D point clouds via voting points, reducing training costs associated with high-dimensional CLIP features while preserving semantic unambiguity. Extensive experiments demonstrate effectiveness of VoteSplat in open-vocabulary 3D instance localization, 3D point cloud understanding, click-based 3D object localization, hierarchical segmentation, and ablation studies. Our code is available at https://sy-ja.github.io/votesplat/
Authors: Nuoye Xiong, Anqi Dong, Ning Wang, Cong Hua, Guangming Zhu, Mei Lin, Peiyi Shen, Liang Zhang
Abstract: Recent advances in deep learning have led to increasingly complex models with deeper layers and more parameters, reducing interpretability and making their decisions harder to understand. While many methods explain black-box reasoning, most lack effective interventions or only operate at sample-level without modifying the model itself. To address this, we propose the Concept Bottleneck Model for Enhancing Human-Neural Network Mutual Understanding (CBM-HNMU). CBM-HNMU leverages the Concept Bottleneck Model (CBM) as an interpretable framework to approximate black-box reasoning and communicate conceptual understanding. Detrimental concepts are automatically identified and refined (removed/replaced) based on global gradient contributions. The modified CBM then distills corrected knowledge back into the black-box model, enhancing both interpretability and accuracy. We evaluate CBM-HNMU on various CNN and transformer-based models across Flower-102, CIFAR-10, CIFAR-100, FGVC-Aircraft, and CUB-200, achieving a maximum accuracy improvement of 2.64% and a maximum increase in average accuracy across 1.03%. Source code is available at: https://github.com/XiGuaBo/CBM-HNMU.
Authors: Byung Hyun Lee, Sungjin Lim, Seunggyu Lee, Dong Un Kang, Se Young Chun
Abstract: Remarkable progress in text-to-image diffusion models has brought a major concern about potentially generating images on inappropriate or trademarked concepts. Concept erasing has been investigated with the goals of deleting target concepts in diffusion models while preserving other concepts with minimal distortion. To achieve these goals, recent concept erasing methods usually fine-tune the cross-attention layers of diffusion models. In this work, we first show that merely updating the cross-attention layers in diffusion models, which is mathematically equivalent to adding \emph{linear} modules to weights, may not be able to preserve diverse remaining concepts. Then, we propose a novel framework, dubbed Concept Pinpoint Eraser (CPE), by adding \emph{nonlinear} Residual Attention Gates (ResAGs) that selectively erase (or cut) target concepts while safeguarding remaining concepts from broad distributions by employing an attention anchoring loss to prevent the forgetting. Moreover, we adversarially train CPE with ResAG and learnable text embeddings in an iterative manner to maximize erasing performance and enhance robustness against adversarial attacks. Extensive experiments on the erasure of celebrities, artistic styles, and explicit contents demonstrated that the proposed CPE outperforms prior arts by keeping diverse remaining concepts while deleting the target concepts with robustness against attack prompts. Code is available at https://github.com/Hyun1A/CPE
Authors: Ramya Hebbalaguppe, Tamoghno Kandar, Abhinav Nagpal, Chetan Arora
Abstract: Vision-language models (VLM) have demonstrated impressive performance in image recognition by leveraging self-supervised training on large datasets. Their performance can be further improved by adapting to the test sample using test-time prompt tuning (TPT). Unfortunately, the singular focus of TPT approaches on improving the accuracy suffers from tunnel vision, and leads to degradation in confidence calibration. This limits the applicability of TPT in critical applications. We make three contributions in this work. (1) We posit that random or naive initialization of prompts leads to overfitting on a particular test sample, and is the main reason for miscalibration of the VLM after TPT. To mitigate the problem, we propose careful initialization of test time prompt using prior knowledge about the target label attributes from a large language model (LLM); (2) To further maintain the quality of prompts during \tpt, we propose a novel regularization loss to reduce intraclass distance, and increase inter-class distance between the learnt Through extensive experiments on different CLIP architectures and 15 datasets, we show that our approach can effectively improve the calibration after TPT. We report an average expected calibration error (ECE) of 4.11 with our method, TCA, compared to 11.7 for vanilla TPT, 6.12 for C-TPT (ICLR'24), 6.78 for DiffTPT (CVPR'23), and 8.43 for PromptAlign (NeurIPS'23). The code is publicly accessible at: https://github.com/rhebbalaguppe/TCA_PromptWithoutPanic.
URLs: https://github.com/rhebbalaguppe/TCA_PromptWithoutPanic.
Authors: Arnulf Jentzen, Konrad Kleinberg, Thomas Kruse
Abstract: Discrete time stochastic optimal control problems and Markov decision processes (MDPs) are fundamental models for sequential decision-making under uncertainty and as such provide the mathematical framework underlying reinforcement learning theory. A central tool for solving MDPs is the Bellman equation and its solution, the so-called $Q$-function. In this article, we construct deep neural network (DNN) approximations for $Q$-functions associated to MDPs with infinite time horizon and finite control set $A$. More specifically, we show that if the the payoff function and the random transition dynamics of the MDP can be suitably approximated by DNNs with leaky rectified linear unit (ReLU) activation, then the solutions $Q_d\colon \mathbb R^d\to \mathbb R^{|A|}$, $d\in \mathbb{N}$, of the associated Bellman equations can also be approximated in the $L^2$-sense by DNNs with leaky ReLU activation whose numbers of parameters grow at most polynomially in both the dimension $d\in \mathbb{N}$ of the state space and the reciprocal $1/\varepsilon$ of the prescribed error $\varepsilon\in (0,1)$. Our proof relies on the recently introduced full-history recursive multilevel fixed-point (MLFP) approximation scheme.
Authors: Qilong Xing, Zikai Song, Yuteng Ye, Yuke Chen, Youjia Zhang, Na Feng, Junqing Yu, Wei Yang
Abstract: Segmentation of brain structures from MRI is crucial for evaluating brain morphology, yet existing CNN and transformer-based methods struggle to delineate complex structures accurately. While current diffusion models have shown promise in image segmentation, they are inadequate when applied directly to brain MRI due to neglecting anatomical information. To address this, we propose Collaborative Anatomy Diffusion (CA-Diff), a framework integrating spatial anatomical features to enhance segmentation accuracy of the diffusion model. Specifically, we introduce distance field as an auxiliary anatomical condition to provide global spatial context, alongside a collaborative diffusion process to model its joint distribution with anatomical structures, enabling effective utilization of anatomical features for segmentation. Furthermore, we introduce a consistency loss to refine relationships between the distance field and anatomical structures and design a time adapted channel attention module to enhance the U-Net feature fusion procedure. Extensive experiments show that CA-Diff outperforms state-of-the-art (SOTA) methods.
Authors: Ehsan Pajouheshgar, Yitao Xu, Ali Abbasi, Alexander Mordvintsev, Wenzel Jakob, Sabine S\"usstrunk
Abstract: Neural Cellular Automata (NCAs) are bio-inspired systems in which identical cells self-organize to form complex and coherent patterns by repeatedly applying simple local rules. NCAs display striking emergent behaviors including self-regeneration, generalization and robustness to unseen situations, and spontaneous motion. Despite their success in texture synthesis and morphogenesis, NCAs remain largely confined to low-resolution grids. This limitation stems from (1) training time and memory requirements that grow quadratically with grid size, (2) the strictly local propagation of information which impedes long-range cell communication, and (3) the heavy compute demands of real-time inference at high resolution. In this work, we overcome this limitation by pairing NCA with a tiny, shared implicit decoder, inspired by recent advances in implicit neural representations. Following NCA evolution on a coarse grid, a lightweight decoder renders output images at arbitrary resolution. We also propose novel loss functions for both morphogenesis and texture synthesis tasks, specifically tailored for high-resolution output with minimal memory and computation overhead. Combining our proposed architecture and loss functions brings substantial improvement in quality, efficiency, and performance. NCAs equipped with our implicit decoder can generate full-HD outputs in real time while preserving their self-organizing, emergent properties. Moreover, because each MLP processes cell states independently, inference remains highly parallelizable and efficient. We demonstrate the applicability of our approach across multiple NCA variants (on 2D, 3D grids, and 3D meshes) and multiple tasks, including texture generation and morphogenesis (growing patterns from a seed), showing that with our proposed framework, NCAs seamlessly scale to high-resolution outputs with minimal computational overhead.
Authors: Yunxuan Ma, Siqiang Wang, Zhijian Duan, Yukun Cheng, Xiaotie Deng
Abstract: This paper introduces TEDI (Truthful, Expressive, and Dimension-Insensitive approach), a discretization-free algorithm to learn truthful and utility-maximizing mechanisms. Existing learning-based approaches often rely on discretization of outcome spaces to ensure truthfulness, which leads to inefficiency with increasing problem size. To address this limitation, we formalize the concept of pricing rules, defined as functions that map outcomes to prices. Based on this concept, we propose a novel menu mechanism, which can be equivalent to a truthful direct mechanism under specific conditions. The core idea of TEDI lies in its parameterization of pricing rules using Partial GroupMax Network, a new network architecture designed to universally approximate partial convex functions. To learn optimal pricing rules, we develop novel training techniques, including covariance trick and continuous sampling, to derive unbiased gradient estimators compatible with first-order optimization. Theoretical analysis establishes that TEDI guarantees truthfulness, full expressiveness, and dimension-insensitivity. Experimental evaluation in the studied auction setting demonstrates that TEDI achieves strong performance, competitive with or exceeding state-of-the-art methods. This work presents the first approaches to learn truthful mechanisms without outcome discretization, thereby enhancing algorithmic efficiency. The proposed concepts, network architecture, and learning techniques might offer potential value and provide new insights for automated mechanism design and differentiable economics.
Authors: Marc Bara Iniesta
Abstract: The ambiguity function is fundamental to radar waveform design, characterizing range and Doppler resolution capabilities. However, its traditional formulation involves non-differentiable operations, preventing integration with gradient-based optimization methods and modern machine learning frameworks. This paper presents the first complete mathematical framework and computational implementation for differentiable radar ambiguity functions. Our approach addresses the fundamental technical challenges that have prevented the radar community from leveraging automatic differentiation: proper handling of complex-valued gradients using Wirtinger calculus, efficient computation through parallelized FFT operations, numerical stability throughout cascaded operations, and composability with arbitrary differentiable operations. We term this approach GRAF (Gradient-based Radar Ambiguity Functions), which reformulates the ambiguity function computation to maintain mathematical equivalence while enabling gradient flow through the entire pipeline. The resulting implementation provides a general-purpose differentiable ambiguity function compatible with modern automatic differentiation frameworks, enabling new research directions including neural network-based waveform generation with ambiguity constraints, end-to-end optimization of radar systems, and integration of classical radar theory with modern deep learning. We provide complete implementation details and demonstrate computational efficiency suitable for practical applications. This work establishes the mathematical and computational foundation for applying modern machine learning techniques to radar waveform design, bridging classical radar signal processing with automatic differentiation frameworks.
Authors: Zaydon L. Ali, Wassan Saad Abduljabbar Hayale, Israa Ibraheem Al_Barazanchi, Ravi Sekhar, Pritesh Shah, Sushma Parihar
Abstract: With current advancement in hybermedia knowledges, the privacy of digital information has developed a critical problem. To overawed the susceptibilities of present security protocols, scholars tend to focus mainly on efforts on alternation of current protocols. Over past decade, various proposed encoding models have been shown insecurity, leading to main threats against significant data. Utilizing the suitable encryption model is very vital means of guard against various such, but algorithm is selected based on the dependency of data which need to be secured. Moreover, testing potentiality of the security assessment one by one to identify the best choice can take a vital time for processing. For faster and precisive identification of assessment algorithm, we suggest a security phase exposure model for cipher encryption technique by invoking Support Vector Machine (SVM). In this work, we form a dataset using usual security components like contrast, homogeneity. To overcome the uncertainty in analysing the security and lack of ability of processing data to a risk assessment mechanism. To overcome with such complications, this paper proposes an assessment model for security issues using fuzzy evidential reasoning (ER) approaches. Significantly, the model can be utilised to process and assemble risk assessment data on various aspects in systematic ways. To estimate the performance of our framework, we have various analyses like, recall, F1 score and accuracy.
Authors: Ghufran A. Omran, Wassan Saad Abduljabbar Hayale, Ahmad AbdulQadir AlRababah, Israa Ibraheem Al-Barazanchi, Ravi Sekhar, Pritesh Shah, Sushma Parihar, Harshavardhan Reddy Penubadi
Abstract: Scene categorization (SC) in remotely acquired images is an important subject with broad consequences in different fields, including catastrophe control, ecological observation, architecture for cities, and more. Nevertheless, its several apps, reaching a high degree of accuracy in SC from distant observation data has demonstrated to be difficult. This is because traditional conventional deep learning models require large databases with high variety and high levels of noise to capture important visual features. To address these problems, this investigation file introduces an innovative technique referred to as the Cuttlefish Optimized Bidirectional Recurrent Neural Network (CO- BRNN) for type of scenes in remote sensing data. The investigation compares the execution of CO-BRNN with current techniques, including Multilayer Perceptron- Convolutional Neural Network (MLP-CNN), Convolutional Neural Network-Long Short Term Memory (CNN-LSTM), and Long Short Term Memory-Conditional Random Field (LSTM-CRF), Graph-Based (GB), Multilabel Image Retrieval Model (MIRM-CF), Convolutional Neural Networks Data Augmentation (CNN-DA). The results demonstrate that CO-BRNN attained the maximum accuracy of 97%, followed by LSTM-CRF with 90%, MLP-CNN with 85%, and CNN-LSTM with 80%. The study highlights the significance of physical confirmation to ensure the efficiency of satellite data.
Authors: Ehsan Hallaji, Vaishnavi Shanmugam, Roozbeh Razavi-Far, Mehrdad Saif
Abstract: One of the most difficult challenges in cybersecurity is eliminating Distributed Denial of Service (DDoS) attacks. Automating this task using artificial intelligence is a complex process due to the inherent class imbalance and lack of sufficient labeled samples of real-world datasets. This research investigates the use of Semi-Supervised Learning (SSL) techniques to improve DDoS attack detection when data is imbalanced and partially labeled. In this process, 13 state-of-the-art SSL algorithms are evaluated for detecting DDoS attacks in several scenarios. We evaluate their practical efficacy and shortcomings, including the extent to which they work in extreme environments. The results will offer insight into designing intelligent Intrusion Detection Systems (IDSs) that are robust against class imbalance and handle partially labeled data.
Authors: Kevin Lam, William Daniels, J Maxwell Douglas, Daniel Lai, Samuel Aparicio, Benjamin Bloem-Reddy, Yongjin Park
Abstract: Cancer is a genetic disorder whose clonal evolution can be monitored by tracking noisy genome-wide copy number variants. We introduce the Copy Number Stochastic Block Model (CN-SBM), a probabilistic framework that jointly clusters samples and genomic regions based on discrete copy number states using a bipartite categorical block model. Unlike models relying on Gaussian or Poisson assumptions, CN-SBM respects the discrete nature of CNV calls and captures subpopulation-specific patterns through block-wise structure. Using a two-stage approach, CN-SBM decomposes CNV data into primary and residual components, enabling detection of both large-scale chromosomal alterations and finer aberrations. We derive a scalable variational inference algorithm for application to large cohorts and high-resolution data. Benchmarks on simulated and real datasets show improved model fit over existing methods. Applied to TCGA low-grade glioma data, CN-SBM reveals clinically relevant subtypes and structured residual variation, aiding patient stratification in survival analysis. These results establish CN-SBM as an interpretable, scalable framework for CNV analysis with direct relevance for tumor heterogeneity and prognosis.
Authors: Amir Aghdam, Vincent Tao Hu
Abstract: We address the task of zero-shot fine-grained video classification, where no video examples or temporal annotations are available for unseen action classes. While contrastive vision-language models such as SigLIP demonstrate strong open-set recognition via mean-pooled image-text similarity, they fail to capture the temporal structure critical for distinguishing fine-grained activities. We introduce ActAlign, a zero-shot framework that formulates video classification as sequence alignment. For each class, a large language model generates an ordered sub-action sequence, which is aligned with video frames using Dynamic Time Warping (DTW) in a shared embedding space. Without any video-text supervision or fine-tuning, ActAlign achieves 30.5% accuracy on the extremely challenging ActionAtlas benchmark, where human accuracy is only 61.6%. ActAlign outperforms billion-parameter video-language models while using approximately 8x less parameters. These results demonstrate that structured language priors, combined with classical alignment techniques, offer a scalable and general approach to unlocking the open-set recognition potential of vision-language models for fine-grained video understanding.
Authors: Kesav Kaza, Ramachandran Anantharaman, Rahul Meshram
Abstract: This paper presents a two-timescale hierarchical decentralized architecture for control of Cyber-Physical Systems. The architecture consists of $N$ independent sub-processes, a global controller, and $N$ local controllers, each formulated as a Markov Decision Process (MDP). The global controller, operating at a slower timescale optimizes the infinite-horizon discounted cumulative reward under budget constraints. For the local controllers, operating at a faster timescale, we propose two different optimization frameworks, namely the COpt and FOpt. In the COpt framework, the local controller also optimizes an infinite-horizon MDP, while in the FOpt framework, the local controller optimizes a finite-horizon MDP. The FOpt framework mimics a federal structure, where the local controllers have more autonomy in their decision making. First, the existence of stationary deterministic optimal policies for both these frameworks is established. Then, various relationships between the two frameworks are studied, including a bound on the difference between the two optimal value functions. Additionally, sufficiency conditions are provided such that the two frameworks lead to the same optimal values.
Authors: Asen Dotsinski, Udit Thakur, Marko Ivanov, Mohammad Hafeez Khan, Maria Heuss
Abstract: We present a reproduction study of "Competition of Mechanisms: Tracing How Language Models Handle Facts and Counterfactuals" (Ortu et al., 2024), which investigates competition of mechanisms in language models between factual recall and counterfactual in-context repetition. Our study successfully reproduces their primary findings regarding the localization of factual and counterfactual information, the dominance of attention blocks in mechanism competition, and the specialization of attention heads in handling competing information. We reproduce their results on both GPT-2 (Radford et al., 2019) and Pythia 6.9B (Biderman et al., 2023). We extend their work in three significant directions. First, we explore the generalizability of these findings to even larger models by replicating the experiments on Llama 3.1 8B (Grattafiori et al., 2024), discovering greatly reduced attention head specialization. Second, we investigate the impact of prompt structure by introducing variations where we avoid repeating the counterfactual statement verbatim or we change the premise word, observing a marked decrease in the logit for the counterfactual token. Finally, we test the validity of the authors' claims for prompts of specific domains, discovering that certain categories of prompts skew the results by providing the factual prediction token as part of the subject of the sentence. Overall, we find that the attention head ablation proposed in Ortu et al. (2024) is ineffective for domains that are underrepresented in their dataset, and that the effectiveness varies based on model architecture, prompt structure, domain and task.
Authors: Max Lovig, Tianhao Wang, Zhou Fan
Abstract: Mean-field characterizations of first-order iterative algorithms -- including Approximate Message Passing (AMP), stochastic and proximal gradient descent, and Langevin diffusions -- have enabled a precise understanding of learning dynamics in many statistical applications. For algorithms whose non-linearities have a coordinate-separable form, it is known that such characterizations enjoy a degree of universality with respect to the underlying data distribution. However, mean-field characterizations of non-separable algorithm dynamics have largely remained restricted to i.i.d. Gaussian or rotationally-invariant data. In this work, we initiate a study of universality for non-separable AMP algorithms. We identify a general condition for AMP with polynomial non-linearities, in terms of a Bounded Composition Property (BCP) for their representing tensors, to admit a state evolution that holds universally for matrices with non-Gaussian entries. We then formalize a condition of BCP-approximability for Lipschitz AMP algorithms to enjoy a similar universal guarantee. We demonstrate that many common classes of non-separable non-linearities are BCP-approximable, including local denoisers, spectral denoisers for generic signals, and compositions of separable functions with generic linear maps, implying the universality of state evolution for AMP algorithms employing these non-linearities.
Authors: M. Youssef Abdelhamid, Lennart Vater, Zlatan Ajanovic
Abstract: Developing decision-making algorithms for highly automated driving systems remains challenging, since these systems have to operate safely in an open and complex environments. Reinforcement Learning (RL) approaches can learn comprehensive decision policies directly from experience and already show promising results in simple driving tasks. However, current approaches fail to achieve generalizability for more complex driving tasks and lack learning efficiency. Therefore, we present Scenario-based Automated Driving Reinforcement Learning (SAD-RL), the first framework that integrates Reinforcement Learning (RL) of hierarchical policy in a scenario-based environment. A high-level policy selects maneuver templates that are evaluated and executed by a low-level control logic. The scenario-based environment allows to control the training experience for the agent and to explicitly introduce challenging, but rate situations into the training process. Our experiments show that an agent trained using the SAD-RL framework can achieve safe behaviour in easy as well as challenging situations efficiently. Our ablation studies confirmed that both HRL and scenario diversity are essential for achieving these results.
Authors: Alejandro Romero Amezcua, Mariano Jos\'e Juan Rivera Meraz
Abstract: VisionScores presents a novel proposal being the first system-segmented image score dataset, aiming to offer structure-rich, high information-density images for machine and deep learning tasks. Delimited to two-handed piano pieces, it was built to consider not only certain graphic similarity but also composition patterns, as this creative process is highly instrument-dependent. It provides two scenarios in relation to composer and composition type. The first, formed by 14k samples, considers works from different authors but the same composition type, specifically, Sonatinas. The latter, consisting of 10.8K samples, presents the opposite case, various composition types from the same author, being the one selected Franz Liszt. All of the 24.8k samples are formatted as grayscale jpg images of $128 \times 512$ pixels. VisionScores supplies the users not only the formatted samples but the systems' order and pieces' metadata. Moreover, unsegmented full-page scores and the pre-formatted images are included for further analysis.
Authors: Yu Zheng, Boyang Gong, Fanye Kong, Yueqi Duan, Bingyao Yu, Wenzhao Zheng, Lei Chen, Jiwen Lu, Jie Zhou
Abstract: In this paper, we propose a Counterfactually Decoupled Attention Learning (CDAL) method for open-world model attribution. Existing methods rely on handcrafted design of region partitioning or feature space, which could be confounded by the spurious statistical correlations and struggle with novel attacks in open-world scenarios. To address this, CDAL explicitly models the causal relationships between the attentional visual traces and source model attribution, and counterfactually decouples the discriminative model-specific artifacts from confounding source biases for comparison. In this way, the resulting causal effect provides a quantification on the quality of learned attention maps, thus encouraging the network to capture essential generation patterns that generalize to unseen source models by maximizing the effect. Extensive experiments on existing open-world model attribution benchmarks show that with minimal computational overhead, our method consistently improves state-of-the-art models by large margins, particularly for unseen novel attacks. Source code: https://github.com/yzheng97/CDAL.
Authors: Yuchen Zhou, Jiamin Wu, Zichen Ren, Zhouheng Yao, Weiheng Lu, Kunyu Peng, Qihao Zheng, Chunfeng Song, Wanli Ouyang, Chao Gou
Abstract: Understanding and decoding brain activity from electroencephalography (EEG) signals is a fundamental challenge in neuroscience and AI, with applications in cognition, emotion recognition, diagnosis, and brain-computer interfaces. While recent EEG foundation models advance generalized decoding via unified architectures and large-scale pretraining, they adopt a scale-agnostic dense modeling paradigm inherited from NLP and vision. This design neglects a core property of neural activity: cross-scale spatiotemporal structure. EEG task patterns span a wide range of temporal and spatial scales, from short bursts to slow rhythms, and from localized cortical responses to distributed interactions. Ignoring this diversity leads to suboptimal representations and weak generalization. We propose CSBrain, a Cross-scale Spatiotemporal Brain foundation model for generalized EEG decoding. CSBrain introduces: (i) Cross-scale Spatiotemporal Tokenization (CST), which aggregates multi-scale features from localized temporal windows and anatomical brain regions into compact scale-aware tokens; and (ii) Structured Sparse Attention (SSA), which captures cross-window and cross-region dependencies, enhancing scale diversity while removing spurious correlations. CST and SSA are alternately stacked to progressively integrate multi-scale dependencies. Experiments on 11 EEG tasks across 16 datasets show that CSBrain consistently outperforms task-specific and foundation model baselines. These results establish cross-scale modeling as a key inductive bias and position CSBrain as a robust backbone for future brain-AI research.
Authors: Langming Liu, Wanyu Wang, Chi Zhang, Bo Li, Hongzhi Yin, Xuetao Wei, Wenbo Su, Bo Zheng, Xiangyu Zhao
Abstract: Online advertising in recommendation platforms has gained significant attention, with a predominant focus on channel recommendation and budget allocation strategies. However, current offline reinforcement learning (RL) methods face substantial challenges when applied to sparse advertising scenarios, primarily due to severe overestimation, distributional shifts, and overlooking budget constraints. To address these issues, we propose MTORL, a novel multi-task offline RL model that targets two key objectives. First, we establish a Markov Decision Process (MDP) framework specific to the nuances of advertising. Then, we develop a causal state encoder to capture dynamic user interests and temporal dependencies, facilitating offline RL through conditional sequence modeling. Causal attention mechanisms are introduced to enhance user sequence representations by identifying correlations among causal states. We employ multi-task learning to decode actions and rewards, simultaneously addressing channel recommendation and budget allocation. Notably, our framework includes an automated system for integrating these tasks into online advertising. Extensive experiments on offline and online environments demonstrate MTORL's superiority over state-of-the-art methods.
Authors: Jaime Hieu Do, Trung-Hoang Le, Hady W. Lauw
Abstract: In the online digital realm, recommendation systems are ubiquitous and play a crucial role in enhancing user experience. These systems leverage user preferences to provide personalized recommendations, thereby helping users navigate through the paradox of choice. This work focuses on personalized sequential recommendation, where the system considers not only a user's immediate, evolving session context, but also their cumulative historical behavior to provide highly relevant and timely recommendations. Through an empirical study conducted on diverse real-world datasets, we have observed and quantified the existence and impact of both short-term (immediate and transient) and long-term (enduring and stable) preferences on users' historical interactions. Building on these insights, we propose a framework that combines short- and long-term preferences to enhance recommendation performance, namely Compositions of Variant Experts (CoVE). This novel framework dynamically integrates short- and long-term preferences through the use of different specialized recommendation models (i.e., experts). Extensive experiments showcase the effectiveness of the proposed methods and ablation studies further investigate the impact of variant expert types.
Authors: Tomer Slor, Dean Oren, Shira Baneth, Tom Coen, Haim Suchowski
Abstract: In the rapidly evolving field of optical engineering, precise alignment of multi-lens imaging systems is critical yet challenging, as even minor misalignments can significantly degrade performance. Traditional alignment methods rely on specialized equipment and are time-consuming processes, highlighting the need for automated and scalable solutions. We present two complementary deep learning-based inverse-design methods for diagnosing misalignments in multi-element lens systems using only optical measurements. First, we use ray-traced spot diagrams to predict five-degree-of-freedom (5-DOF) errors in a 6-lens photographic prime, achieving a mean absolute error of 0.031mm in lateral translation and 0.011$^\circ$ in tilt. We also introduce a physics-based simulation pipeline that utilizes grayscale synthetic camera images, enabling a deep learning model to estimate 4-DOF, decenter and tilt errors in both two- and six-lens multi-lens systems. These results show the potential to reshape manufacturing and quality control in precision imaging.
Authors: Gabriel Iturra-Bocaz, Felipe Bravo-Marquez
Abstract: Word embeddings have become essential components in various information retrieval and natural language processing tasks, such as ranking, document classification, and question answering. However, despite their widespread use, traditional word embedding models present a limitation in their static nature, which hampers their ability to adapt to the constantly evolving language patterns that emerge in sources such as social media and the web (e.g., new hashtags or brand names). To overcome this problem, incremental word embedding algorithms are introduced, capable of dynamically updating word representations in response to new language patterns and processing continuous data streams. This paper presents RiverText, a Python library for training and evaluating incremental word embeddings from text data streams. Our tool is a resource for the information retrieval and natural language processing communities that work with word embeddings in streaming scenarios, such as analyzing social media. The library implements different incremental word embedding techniques, such as Skip-gram, Continuous Bag of Words, and Word Context Matrix, in a standardized framework. In addition, it uses PyTorch as its backend for neural network training. We have implemented a module that adapts existing intrinsic static word embedding evaluation tasks for word similarity and word categorization to a streaming setting. Finally, we compare the implemented methods with different hyperparameter settings and discuss the results. Our open-source library is available at https://github.com/dccuchile/rivertext.
Authors: James Hinns, David Martens
Abstract: Deep learning dominates image classification tasks, yet understanding how models arrive at predictions remains a challenge. Much research focuses on local explanations of individual predictions, such as saliency maps, which visualise the influence of specific pixels on a model's prediction. However, reviewing many of these explanations to identify recurring patterns is infeasible, while global methods often oversimplify and miss important local behaviours. To address this, we propose Segment Attribution Tables (SATs), a method for summarising local saliency explanations into (semi-)global insights. SATs take image segments (such as "eyes" in Chihuahuas) and leverage saliency maps to quantify their influence. These segments highlight concepts the model relies on across instances and reveal spurious correlations, such as reliance on backgrounds or watermarks, even when out-of-distribution test performance sees little change. SATs can explain any classifier for which a form of saliency map can be produced, using segmentation maps that provide named segments. SATs bridge the gap between oversimplified global summaries and overly detailed local explanations, offering a practical tool for analysing and debugging image classifiers.
Authors: Yi Li, Hualiang Wang, Xinpeng Ding, Haonan Wang, Xiaomeng Li
Abstract: Multimodal large language models (MLLMs) are broadly empowering various fields. Despite their advancements, the explainability of MLLMs remains less explored, hindering deeper understanding, model credibility, and effective visualization. Unlike conventional vision models (e.g., CNNs, ViTs, CLIP) that produce a single output, MLLMs generate sequences of tokens progressively, where each generated token depends on the previous context. Therefore, earlier context tokens can introduce redundant activations that interfere with the explanation of later tokens beyond their original information. Existing studies often overlook this issue, but our observations reveal that these redundant correlations can significantly hurt the reliability of explanations. To address this, we propose an estimated causal inference method to mitigate the interference of context to achieve high-quality MLLM explanation, with a novel rank Gaussian filter to further reduce activation noises. We term this method Token Activation Map (TAM) to highlight the consideration of interactions between tokens. TAM also indicates that it excels at explaining multiple tokens of MLLM, which is different from the Class Activation Map (CAM) for a single prediction. Our TAM method significantly outperforms existing SoTA methods, showcasing high-quality visualization results that can be utilized for various scenarios, such as object localization, failure case analysis, video visualization, MLLMs visual comparison, and model understanding (e.g., color, shape, action, location, visual reasoning, multi-turn conversation, etc). The code is available atgithub.com/xmed-lab/TAM.
Authors: P. Myles Eugenio
Abstract: We present a neuro-symbolic framework for generative language modeling based on local, event-driven emergent learning. At its core is a hierarchical Hopfield memory chain acting as a compositional short-term memory and dynamic tokenizer (retokenizer). Rather than relying on predefined tokens or supervision, the model builds structure from scratch, learning symbol sequences as multi-scale representations. It constructs projection tensors that bind co-occurring features into hierarchical tokens, introducing redundancy (i.e an emergent gauge structure) and enabling compression of local activations into long-range dependencies. Curiously, we find that the retokenizer can filter natural language patterns from noise, generating synthetic languages with coherent internal morphology -- quantifiably the same as human language. Language is learned in a local (Hebbian) fashion, where model constraints dictate allowed emergent structure, and new information is retained in alignment with this structure. The absence of a global objective enables a form of plasticity not found in conventional language models, allowing the system to generalize beyond its initial inference class -- even without explicit data. We demonstrate that briefly activating a new neuron during inference binds distributed multi-scale token features into a symbolic embedding. These emergent embedding neurons act as long-term memory and support a key-value mechanism for compositional inference and generalization. This architecture provides a methodological foundation for studying how symbolic structure can emerge from local neural learning. It offers a new pathway for building scalable, interpretable neuro-symbolic systems -- where tokens, grammar, and reasoning arise as compressed memory traces within a Hopfield hierarchy. This approach advances the development of neuromorphic architectures for generative language models.
Authors: Perla Mayo, Carolin M. Pirkl, Alin Achim, Bjoern Menze, Mohammad Golbabaee
Abstract: We introduce MRF-DiPh, a novel physics informed denoising diffusion approach for multiparametric tissue mapping from highly accelerated, transient-state quantitative MRI acquisitions like Magnetic Resonance Fingerprinting (MRF). Our method is derived from a proximal splitting formulation, incorporating a pretrained denoising diffusion model as an effective image prior to regularize the MRF inverse problem. Further, during reconstruction it simultaneously enforces two key physical constraints: (1) k-space measurement consistency and (2) adherence to the Bloch response model. Numerical experiments on in-vivo brain scans data show that MRF-DiPh outperforms deep learning and compressed sensing MRF baselines, providing more accurate parameter maps while better preserving measurement fidelity and physical model consistency-critical for solving reliably inverse problems in medical imaging.
Authors: Shouvon Sarker, Xishuang Dong, Lijun Qian
Abstract: Identification of key variables such as medications, diseases, relations from health records and clinical notes has a wide range of applications in the clinical domain. n2c2 2022 provided shared tasks on challenges in natural language processing for clinical data analytics on electronic health records (EHR), where it built a comprehensive annotated clinical data Contextualized Medication Event Dataset (CMED). This study focuses on subtask 2 in Track 1 of this challenge that is to detect and classify medication events from clinical notes through building a novel BERT-based ensemble model. It started with pretraining BERT models on different types of big data such as Wikipedia and MIMIC. Afterwards, these pretrained BERT models were fine-tuned on CMED training data. These fine-tuned BERT models were employed to accomplish medication event classification on CMED testing data with multiple predictions. These multiple predictions generated by these fine-tuned BERT models were integrated to build final prediction with voting strategies. Experimental results demonstrated that BERT-based ensemble models can effectively improve strict Micro-F score by about 5% and strict Macro-F score by about 6%, respectively.
Authors: Norman Knyazev, Harrie Oosterhuis
Abstract: Learning to Rank (LTR) methods generally assume that each document in a top-K ranking is presented in an equal format. However, previous work has shown that users' perceptions of relevance can be changed by varying presentations, i.e., allocating more vertical space to some documents to provide additional textual or image information. Furthermore, presentation length can also redirect attention, as users are more likely to notice longer presentations when scrolling through results. Deciding on the document presentation lengths in a fixed vertical space ranking is an important problem that has not been addressed by existing LTR methods. We address this gap by introducing the variable presentation length ranking task, where simultaneously the ordering of documents and their presentation length is decided. Despite being a generalization of standard ranking, we show that this setting brings significant new challenges: Firstly, the probability ranking principle no longer applies to this setting, and secondly, the problem cannot be divided into separate ordering and length selection tasks. We therefore propose VLPL - a new family of Plackett-Luce list-wise gradient estimation methods for the joint optimization of document ordering and lengths. Our semi-synthetic experiments show that VLPL can effectively balance the expected exposure and attractiveness of all documents, achieving the best performance across different ranking settings. Furthermore, we observe that even simple length-aware methods can achieve significant performance improvements over fixed-length models. Altogether, our theoretical and empirical results highlight the importance and difficulties of combining document presentation with LTR.
Authors: Difeng Cai, Paulina Sep\'ulveda
Abstract: The appearance of singularities in the function of interest constitutes a fundamental challenge in scientific computing. It can significantly undermine the effectiveness of numerical schemes for function approximation, numerical integration, and the solution of partial differential equations (PDEs), etc. The problem becomes more sophisticated if the location of the singularity is unknown, which is often encountered in solving PDEs. Detecting the singularity is therefore critical for developing efficient adaptive methods to reduce computational costs in various applications. In this paper, we consider singularity detection in a purely data-driven setting. Namely, the input only contains given data, such as the vertex set from a mesh. To overcome the limitation of the raw unlabeled data, we propose a self-supervised learning (SSL) framework for estimating the location of the singularity. A key component is a filtering procedure as the pretext task in SSL, where two filtering methods are presented, based on $k$ nearest neighbors and kernel density estimation, respectively. We provide numerical examples to illustrate the potential pathological or inaccurate results due to the use of raw data without filtering. Various experiments are presented to demonstrate the ability of the proposed approach to deal with input perturbation, label corruption, and different kinds of singularities such interior circle, boundary layer, concentric semicircles, etc.
Authors: Tianxing Chen, Kaixuan Wang, Zhaohui Yang, Yuhao Zhang, Zanxin Chen, Baijun Chen, Wanxi Dong, Ziyuan Liu, Dong Chen, Tianshuo Yang, Haibao Yu, Xiaokang Yang, Yusen Qin, Zhiqiang Xie, Yao Mu, Ping Luo, Tian Nian, Weiliang Deng, Yiheng Ge, Yibin Liu, Zixuan Li, Dehui Wang, Zhixuan Liang, Haohui Xie, Rijie Zeng, Yunfei Ge, Peiqing Cong, Guannan He, Zhaoming Han, Ruocheng Yin, Jingxiang Guo, Lunkai Lin, Tianling Xu, Hongzhe Bi, Xuewu Lin, Tianwei Lin, Shujie Luo, Keyu Li, Ziyan Zhao, Ke Fan, Heyang Xu, Bo Peng, Wenlong Gao, Dongjiang Li, Feng Jin, Hui Shen, Jinming Li, Chaowei Cui, Yuchen, Yaxin Peng, Lingdong Zeng, Wenlong Dong, Tengfei Li, Weijie Ke, Jun Chen, Erdemt Bao, Tian Lan, Tenglong Liu, Jin Yang, Huiping Zhuang, Baozhi Jia, Shuai Zhang, Zhengfeng Zou, Fangheng Guan, Tianyi Jia, Ke Zhou, Hongjiu Zhang, Yating Han, Cheng Fang, Yixian Zou, Chongyang Xu, Qinglun Zhang, Shen Cheng, Xiaohe Wang, Ping Tan, Haoqiang Fan, Shuaicheng Liu, Jiaheng Chen, Chuxuan Huang, Chengliang Lin, Kaijun Luo, Boyu Yue, Yi Liu, Jinyu Chen, Zichang Tan, Liming Deng, Shuo Xu, Zijian Cai, Shilong Yin, Hao Wang, Hongshan Liu, Tianyang Li, Long Shi, Ran Xu, Huilin Xu, Zhengquan Zhang, Congsheng Xu, Jinchang Yang, Feng Xu
Abstract: Embodied Artificial Intelligence (Embodied AI) is an emerging frontier in robotics, driven by the need for autonomous systems that can perceive, reason, and act in complex physical environments. While single-arm systems have shown strong task performance, collaborative dual-arm systems are essential for handling more intricate tasks involving rigid, deformable, and tactile-sensitive objects. To advance this goal, we launched the RoboTwin Dual-Arm Collaboration Challenge at the 2nd MEIS Workshop, CVPR 2025. Built on the RoboTwin Simulation platform (1.0 and 2.0) and the AgileX COBOT-Magic Robot platform, the competition consisted of three stages: Simulation Round 1, Simulation Round 2, and a final Real-World Round. Participants totally tackled 17 dual-arm manipulation tasks, covering rigid, deformable, and tactile-based scenarios. The challenge attracted 64 global teams and over 400 participants, producing top-performing solutions like SEM and AnchorDP3 and generating valuable insights into generalizable bimanual policy learning. This report outlines the competition setup, task design, evaluation methodology, key findings and future direction, aiming to support future research on robust and generalizable bimanual manipulation policies. The Challenge Webpage is available at https://robotwin-benchmark.github.io/cvpr-2025-challenge/.
URLs: https://robotwin-benchmark.github.io/cvpr-2025-challenge/.
Authors: Frank Cwitkowitz, Zhiyao Duan
Abstract: Multi-Pitch Estimation (MPE) continues to be a sought after capability of Music Information Retrieval (MIR) systems, and is critical for many applications and downstream tasks involving pitch, including music transcription. However, existing methods are largely based on supervised learning, and there are significant challenges in collecting annotated data for the task. Recently, self-supervised techniques exploiting intrinsic properties of pitch and harmonic signals have shown promise for both monophonic and polyphonic pitch estimation, but these still remain inferior to supervised methods. In this work, we extend the classic supervised MPE paradigm by incorporating several self-supervised objectives based on pitch-invariant and pitch-equivariant properties. This joint training results in a substantial improvement under closed training conditions, which naturally suggests that applying the same objectives to a broader collection of data will yield further improvements. However, in doing so we uncover a phenomenon whereby our model simultaneously overfits to the supervised data while degenerating on data used for self-supervision only. We demonstrate and investigate this and offer our insights on the underlying problem.
Authors: Vikram Rangarajan, Shishira Maiya, Max Ehrlich, Abhinav Shrivastava
Abstract: Implicit Neural Representations (INRs) offer exceptional fidelity for video compression by learning per-video optimized functions, but their adoption is crippled by impractically slow encoding times. Existing attempts to accelerate INR encoding often sacrifice reconstruction quality or crucial coordinate-level control essential for adaptive streaming and transcoding. We introduce SIEDD (Shared-Implicit Encoder with Discrete Decoders), a novel architecture that fundamentally accelerates INR encoding without these compromises. SIEDD first rapidly trains a shared, coordinate-based encoder on sparse anchor frames to efficiently capture global, low-frequency video features. This encoder is then frozen, enabling massively parallel training of lightweight, discrete decoders for individual frame groups, further expedited by aggressive coordinate-space sampling. This synergistic design delivers a remarkable 20-30X encoding speed-up over state-of-the-art INR codecs on HD and 4K benchmarks, while maintaining competitive reconstruction quality and compression ratios. Critically, SIEDD retains full coordinate-based control, enabling continuous resolution decoding and eliminating costly transcoding. Our approach significantly advances the practicality of high-fidelity neural video compression, demonstrating a scalable and efficient path towards real-world deployment. Our codebase is available at https://github.com/VikramRangarajan/SIEDD .
Authors: Kay Giesecke, Enguerrand Horel, Chartsiri Jirachotkulthorn
Abstract: The opacity of many supervised learning algorithms remains a key challenge, hindering scientific discovery and limiting broader deployment -- particularly in high-stakes domains. This paper develops model- and distribution-agnostic significance tests to assess the influence of input features in any regression or classification algorithm. Our method evaluates a feature's incremental contribution to model performance by masking its values across samples. Under the null hypothesis, the distribution of performance differences across a test set has a non-positive median. We construct a uniformly most powerful, randomized sign test for this median, yielding exact p-values for assessing feature significance and confidence intervals with exact coverage for estimating population-level feature importance. The approach requires minimal assumptions, avoids model retraining or auxiliary models, and remains computationally efficient even for large-scale, high-dimensional settings. Experiments on synthetic tasks validate its statistical and computational advantages, and applications to real-world data illustrate its practical utility.
Authors: Jiale Zhang, Zichong Wang, Avash Palikhe, Zhipeng Yin, Wenbin Zhang
Abstract: Fairness benchmarks play a central role in shaping how we evaluate language models, yet surprisingly little attention has been given to examining the datasets that these benchmarks rely on. This survey addresses that gap by presenting a broad and careful review of the most widely used fairness datasets in current language model research, characterizing them along several key dimensions including their origin, scope, content, and intended use to help researchers better appreciate the assumptions and limitations embedded in these resources. To support more meaningful comparisons and analyses, we introduce a unified evaluation framework that reveals consistent patterns of demographic disparities across datasets and scoring methods. Applying this framework to twenty four common benchmarks, we highlight the often overlooked biases that can influence conclusions about model fairness and offer practical guidance for selecting, combining, and interpreting these datasets. We also point to opportunities for creating new fairness benchmarks that reflect more diverse social contexts and encourage more thoughtful use of these tools going forward. All code, data, and detailed results are publicly available at https://github.com/vanbanTruong/Fairness-in-Large-Language-Models/tree/main/datasets to promote transparency and reproducibility across the research community.
URLs: https://github.com/vanbanTruong/Fairness-in-Large-Language-Models/tree/main/datasets
Authors: Menna Taha (Faculty of Engineering at Alexandria University, Alexandria, Egypt), Aya Ahmed (Department of Electronics and Communications Engineering at The American University in Cairo, Egypt), Mohammed Karmoose (Faculty of Engineering at Alexandria University, Alexandria, Egypt, The Wireless Intelligent Networks Center), Yasser Gadallah (Department of Electronics and Communications Engineering at The American University in Cairo, Egypt)
Abstract: Autonomous vehicles (AVs) use object detection models to recognize their surroundings and make driving decisions accordingly. Conventional object detection approaches classify objects into known classes, which limits the AV's ability to detect and appropriately respond to Out-of-Distribution (OOD) objects. This problem is a significant safety concern since the AV may fail to detect objects or misclassify them, which can potentially lead to hazardous situations such as accidents. Consequently, we propose a novel object detection approach that shifts the emphasis from conventional class-based classification to object harmfulness determination. Instead of object detection by their specific class, our method identifies them as either 'harmful' or 'harmless' based on whether they pose a danger to the AV. This is done based on the object position relative to the AV and its trajectory. With this metric, our model can effectively detect previously unseen objects to enable the AV to make safer real-time decisions. Our results demonstrate that the proposed model effectively detects OOD objects, evaluates their harmfulness, and classifies them accordingly, thus enhancing the AV decision-making effectiveness in dynamic environments.
Authors: Yingyuan Li, Aokun Wang, Zhongjian Wang
Abstract: In this work, we propose a novel machine learning approach to compute the optimal transport map between two continuous distributions from their unpaired samples, based on the DeepParticle methods. The proposed method leads to a min-min optimization during training and does not impose any restriction on the network structure. Theoretically we establish a weak convergence guarantee and a quantitative error bound between the learned map and the optimal transport map. Our numerical experiments validate the theoretical results and the effectiveness of the new approach, particularly on real-world tasks.
Authors: Zhen Zhang, Xin Liu, Shaoli Wang, Jiaye Teng
Abstract: Covariate shift occurs when the distribution of input features differs between the training and testing phases. In covariate shift, estimating an unknown function's moment is a classical problem that remains under-explored, despite its common occurrence in real-world scenarios. In this paper, we investigate the minimax lower bound of the problem when the source and target distributions are known. To achieve the minimax optimal bound (up to a logarithmic factor), we propose a two-stage algorithm. Specifically, it first trains an optimal estimator for the function under the source distribution, and then uses a likelihood ratio reweighting procedure to calibrate the moment estimator. In practice, the source and target distributions are typically unknown, and estimating the likelihood ratio may be unstable. To solve this problem, we propose a truncated version of the estimator that ensures double robustness and provide the corresponding upper bound. Extensive numerical studies on synthetic examples confirm our theoretical findings and further illustrate the effectiveness of our proposed method.
Authors: William Gay, William He, Nicholas Kocurek, Ryan O'Donnell
Abstract: Certain tasks in high-dimensional statistics become easier when the underlying distribution satisfies a local-to-global property called approximate tensorization of entropy (ATE). For example, the Glauber dynamics Markov chain of an ATE distribution mixes fast and can produce approximate samples in a small amount of time, since such a distribution satisfies a modified log-Sobolev inequality. Moreover, identity-testing for an ATE distribution requires few samples if the tester is given coordinate conditional access to the unknown distribution, as shown by Blanca, Chen, \v{S}tefankovi\v{c}, and Vigoda (COLT 2023). A natural class of distributions that do not satisfy ATE consists of mixtures of (few) distributions that do satisfy ATE. We study the complexity of identity-testing and sampling for these distributions. Our main results are the following: 1. We show fast mixing of Glauber dynamics from a data-based initialization, with optimal sample complexity, for mixtures of distributions satisfying modified log-Sobolev inequalities. This extends work of Huang, Koehler, Lee, Mohanty, Rajaraman, Vuong, and Wu (STOC 2025, COLT 2025) for mixtures of distributions satisfying Poincar\'e inequalities. 2. Answering an open question posed by Blanca et al., we give efficient identity-testers for mixtures of ATE distributions in the coordinate-conditional sampling access model. We also give some simplifications and improvements to the original algorithm of Blanca et al.
Authors: Xiaoxiao Yang, Chao Feng, Jiancheng Chen
Abstract: Portable and wearable consumer-grade electroencephalography (EEG) devices, like Muse headbands, offer unprecedented mobility for daily brain-computer interface (BCI) applications, including cognitive load detection. However, the exacerbated non-stationarity in portable EEG signals constrains data fidelity and decoding accuracy, creating a fundamental trade-off between portability and performance. To mitigate such limitation, we propose MuseCogNet (Muse-based Cognitive Network), a unified joint learning framework integrating self-supervised and supervised training paradigms. In particular, we introduce an EEG-grounded self-supervised reconstruction loss based on average pooling to capture robust neurophysiological patterns, while cross-entropy loss refines task-specific cognitive discriminants. This joint learning framework resembles the bottom-up and top-down attention in humans, enabling MuseCogNet to significantly outperform state-of-the-art methods on a publicly available Muse dataset and establish an implementable pathway for neurocognitive monitoring in ecological settings.
Authors: Chenlang Yi, Zizhan Xiong, Qi Qi, Xiyuan Wei, Girish Bathla, Ching-Long Lin, Bobak Jack Mortazavi, Tianbao Yang
Abstract: Contrastive Language-Image Pre-training (CLIP) models have demonstrated superior performance across various visual tasks including medical image classification. However, fairness concerns, including demographic biases, have received limited attention for CLIP models. This oversight leads to critical issues, particularly those related to race and gender, resulting in disparities in diagnostic outcomes and reduced reliability for underrepresented groups. To address these challenges, we introduce AdFair-CLIP, a novel framework employing adversarial feature intervention to suppress sensitive attributes, thereby mitigating spurious correlations and improving prediction fairness. We conduct comprehensive experiments on chest X-ray (CXR) datasets, and show that AdFair-CLIP significantly enhances both fairness and diagnostic accuracy, while maintaining robust generalization in zero-shot and few-shot scenarios. These results establish new benchmarks for fairness-aware learning in CLIP-based medical diagnostic models, particularly for CXR analysis.
Authors: Haoshu Xu, Hongzhe Li
Abstract: We propose a novel test for assessing partial effects in Frechet regression on Bures Wasserstein manifolds. Our approach employs a sample splitting strategy: the first subsample is used to fit the Frechet regression model, yielding estimates of the covariance matrices and their associated optimal transport maps, while the second subsample is used to construct the test statistic. We prove that this statistic converges in distribution to a weighted mixture of chi squared components, where the weights correspond to the eigenvalues of an integral operator defined by an appropriate RKHS kernel. We establish that our procedure achieves the nominal asymptotic size and demonstrate that its worst-case power converges uniformly to one. Through extensive simulations and a real data application, we illustrate the test's finite-sample accuracy and practical utility.
Authors: Jisu Han, Jihee Park, Dongyoon Han, Wonjun Hwang
Abstract: Training on test-time data enables deep learning models to adapt to dynamic environmental changes, enhancing their practical applicability. Online adaptation from source to target domains is promising but it remains highly reliant on the performance of source pretrained model. In this paper, we investigate whether test-time adaptation (TTA) methods can continuously improve models trained via self-supervised learning (SSL) without relying on source pretraining. We introduce a self-supervised TTA protocol after observing that existing TTA approaches struggle when directly applied to self-supervised models with low accuracy on the source domain. Furthermore, we propose a collaborative learning framework that integrates SSL and TTA models, leveraging contrastive learning and knowledge distillation for stepwise representation refinement. We validate our method on diverse self-supervised models, including DINO, MoCo, and iBOT, across TTA benchmarks. Extensive experiments validate the effectiveness of our approach in SSL, showing that it achieves competitive performance even without source pretraining.
Authors: Jefferson Hernandez, Ruozhen He, Guha Balakrishnan, Alexander C. Berg, Vicente Ordonez
Abstract: We introduce GVIT, a classification framework that abandons conventional pixel or patch grid input representations in favor of a compact set of learnable 2D Gaussians. Each image is encoded as a few hundred Gaussians whose positions, scales, orientations, colors, and opacities are optimized jointly with a ViT classifier trained on top of these representations. We reuse the classifier gradients as constructive guidance, steering the Gaussians toward class-salient regions while a differentiable renderer optimizes an image reconstruction loss. We demonstrate that by 2D Gaussian input representations coupled with our GVIT guidance, using a relatively standard ViT architecture, closely matches the performance of a traditional patch-based ViT, reaching a 76.9% top-1 accuracy on Imagenet-1k using a ViT-B architecture.
Authors: Yuhao Huang, Yueyue Xu, Haoran Dou, Jiaxiao Deng, Xin Yang, Hongyu Zheng, Dong Ni
Abstract: Congenital uterine anomalies (CUAs) can lead to infertility, miscarriage, preterm birth, and an increased risk of pregnancy complications. Compared to traditional 2D ultrasound (US), 3D US can reconstruct the coronal plane, providing a clear visualization of the uterine morphology for assessing CUAs accurately. In this paper, we propose an intelligent system for simultaneous automated plane localization and CUA diagnosis. Our highlights are: 1) we develop a denoising diffusion model with local (plane) and global (volume/text) guidance, using an adaptive weighting strategy to optimize attention allocation to different conditions; 2) we introduce a reinforcement learning-based framework with unsupervised rewards to extract the key slice summary from redundant sequences, fully integrating information across multiple planes to reduce learning difficulty; 3) we provide text-driven uncertainty modeling for coarse prediction, and leverage it to adjust the classification probability for overall performance improvement. Extensive experiments on a large 3D uterine US dataset show the efficacy of our method, in terms of plane localization and CUA diagnosis. Code is available at https://github.com/yuhoo0302/CUA-US.
Authors: Zhendong Yu, Weizhong Huang, Haiping Huang
Abstract: Fixed points of recurrent neural networks can be leveraged to store and generate information. These fixed points can be captured by the Boltzmann-Gibbs measure, which leads to neural Langevin dynamics that can be used for sampling and learning a real dataset. We call this type of generative model neural Langevin machine, which is interpretable due to its analytic form of distribution and is simple to train. Moreover, the learning process is derived as a local asymmetric plasticity rule, bearing biological relevance. Therefore, one can realize a continuous sampling of creative dynamics in a neural network, mimicking an imagination process in brain circuits. This neural Langevin machine may be another promising generative model, at least in its strength in circuit-based sampling and biologically plausible learning rule.
Authors: Huai-Chih Wang, Hsiang-Chun Chuang, Hsi-Chun Cheng, Dai-Jie Wu, Shao-Hua Sun
Abstract: Effective coordination among artificial agents in dynamic and uncertain environments remains a significant challenge in multi-agent systems. Existing approaches, such as self-play and population-based methods, either generalize poorly to unseen partners or require extensive training. To overcome these limitations, we propose Coordination Transformers (CooT), a novel in-context coordination framework that uses recent interaction histories to adapt to unseen partners rapidly. Unlike previous approaches that primarily aim to increase the diversity of training partners, CooT explicitly focuses on adapting to new partner behaviors by predicting actions aligned with observed partner interactions. Trained on interaction trajectories collected from diverse pairs of agents with complementary behaviors, CooT quickly learns effective coordination strategies without explicit supervision or fine-tuning. Evaluations on the Overcooked benchmark demonstrate that CooT significantly outperforms baseline methods in coordination tasks involving previously unseen partners. Human evaluations further confirm CooT as the most effective collaborative partner, while extensive ablations highlight its robustness, flexibility, and sensitivity to context in multi-agent scenarios.
Authors: Ryui Kaneko, Shimpei Goto
Abstract: We find an efficient approach to approximately convert matrix product states (MPSs) into restricted Boltzmann machine wave functions consisting of a multinomial hidden unit through a canonical polyadic (CP) decomposition of the MPSs. This method allows us to generate well-behaved initial neural network quantum states for quantum many-body ground-state calculations in polynomial time of the number of variational parameters and systematically shorten the distance between the initial states and the ground states with increasing the rank of the CP decomposition. We demonstrate the efficiency of our method by taking the transverse-field Ising model as an example and discuss possible applications of our method to more general quantum many-body systems in which the ground-state wave functions possess complex nodal structures.
Authors: Luigi Sigillo, Renato Giamba, Danilo Comminiello
Abstract: The acquisition of high-resolution satellite imagery is often constrained by the spatial and temporal limitations of satellite sensors, as well as the high costs associated with frequent observations. These challenges hinder applications such as environmental monitoring, disaster response, and agricultural management, which require fine-grained and high-resolution data. In this paper, we propose MWT-Diff, an innovative framework for satellite image super-resolution (SR) that combines latent diffusion models with wavelet transforms to address these challenges. At the core of the framework is a novel metadata-, wavelet-, and time-aware encoder (MWT-Encoder), which generates embeddings that capture metadata attributes, multi-scale frequency information, and temporal relationships. The embedded feature representations steer the hierarchical diffusion dynamics, through which the model progressively reconstructs high-resolution satellite imagery from low-resolution inputs. This process preserves critical spatial characteristics including textural patterns, boundary discontinuities, and high-frequency spectral components essential for detailed remote sensing analysis. The comparative analysis of MWT-Diff across multiple datasets demonstrated favorable performance compared to recent approaches, as measured by standard perceptual quality metrics including FID and LPIPS.
Authors: Siddhartha Mondal, Avik Mitra, Chayan Sarkar
Abstract: The deployment of robot assistants in large indoor spaces has seen significant growth, with escorting tasks becoming a key application. However, most current escorting robots primarily rely on navigation-focused strategies, assuming that the person being escorted will follow without issue. In crowded environments, this assumption often falls short, as individuals may struggle to keep pace, become obstructed, get distracted, or need to stop unexpectedly. As a result, conventional robotic systems are often unable to provide effective escorting services due to their limited understanding of human movement dynamics. To address these challenges, an effective escorting robot must continuously detect and interpret human actions during the escorting process and adjust its movement accordingly. However, there is currently no existing dataset designed specifically for human action detection in the context of escorting. Given that escorting often occurs in crowded environments, where other individuals may enter the robot's camera view, the robot also needs to identify the specific human it is escorting (the subject) before predicting their actions. Since no existing model performs both person re-identification and action prediction in real-time, we propose a novel neural network architecture that can accomplish both tasks. This enables the robot to adjust its speed dynamically based on the escortee's movements and seamlessly resume escorting after any disruption. In comparative evaluations against strong baselines, our system demonstrates superior efficiency and effectiveness, showcasing its potential to significantly improve robotic escorting services in complex, real-world scenarios.
Authors: Xiao Li, Yiming Zhu, Yifan Huang, Wei Zhang, Yingzhe He, Jie Shi, Xiaolin Hu
Abstract: Object detection plays a crucial role in many security-sensitive applications. However, several recent studies have shown that object detectors can be easily fooled by physically realizable attacks, \eg, adversarial patches and recent adversarial textures, which pose realistic and urgent threats. Adversarial Training (AT) has been recognized as the most effective defense against adversarial attacks. While AT has been extensively studied in the $l_\infty$ attack settings on classification models, AT against physically realizable attacks on object detectors has received limited exploration. Early attempts are only performed to defend against adversarial patches, leaving AT against a wider range of physically realizable attacks under-explored. In this work, we consider defending against various physically realizable attacks with a unified AT method. We propose PBCAT, a novel Patch-Based Composite Adversarial Training strategy. PBCAT optimizes the model by incorporating the combination of small-area gradient-guided adversarial patches and imperceptible global adversarial perturbations covering the entire image. With these designs, PBCAT has the potential to defend against not only adversarial patches but also unseen physically realizable attacks such as adversarial textures. Extensive experiments in multiple settings demonstrated that PBCAT significantly improved robustness against various physically realizable attacks over state-of-the-art defense methods. Notably, it improved the detection accuracy by 29.7\% over previous defense methods under one recent adversarial texture attack.
Authors: Marvin Xhemrishi, Alexandre Graell i Amat, Bal\'azs Pej\'o
Abstract: Federated learning with secure aggregation enables private and collaborative learning from decentralised data without leaking sensitive client information. However, secure aggregation also complicates the detection of malicious client behaviour and the evaluation of individual client contributions to the learning. To address these challenges, QI (Pejo et al.) and FedGT (Xhemrishi et al.) were proposed for contribution evaluation (CE) and misbehaviour detection (MD), respectively. QI, however, lacks adequate MD accuracy due to its reliance on the random selection of clients in each training round, while FedGT lacks the CE ability. In this work, we combine the strengths of QI and FedGT to achieve both robust MD and accurate CE. Our experiments demonstrate superior performance compared to using either method independently.
Authors: Guillaume Coqueret, Martial Laguerre
Abstract: This paper investigates the impact of posterior drift on out-of-sample forecasting accuracy in overparametrized machine learning models. We document the loss in performance when the loadings of the data generating process change between the training and testing samples. This matters crucially in settings in which regime changes are likely to occur, for instance, in financial markets. Applied to equity premium forecasting, our results underline the sensitivity of a market timing strategy to sub-periods and to the bandwidth parameters that control the complexity of the model. For the average investor, we find that focusing on holding periods of 15 years can generate very heterogeneous returns, especially for small bandwidths. Large bandwidths yield much more consistent outcomes, but are far less appealing from a risk-adjusted return standpoint. All in all, our findings tend to recommend cautiousness when resorting to large linear models for stock market predictions.
Authors: Roham Maiti, Debasmita Bhoumik
Abstract: Brain plays a crucial role in regulating body functions and cognitive processes, with brain tumors posing significant risks to human health. Precise and prompt detection is a key factor in proper treatment and better patient outcomes. Traditional methods for detecting brain tumors, that include biopsies, MRI, and CT scans often face challenges due to their high costs and the need for specialized medical expertise. Recent developments in machine learning (ML) and deep learning (DL) has exhibited strong capabilities in automating the identification and categorization of brain tumors from medical images, especially MRI scans. However, these classical ML models have limitations, such as high computational demands, the need for large datasets, and long training times, which hinder their accessibility and efficiency. Our research uses MobileNET model for efficient detection of these tumors. The novelty of this project lies in building an accurate tumor detection model which use less computing re-sources and runs in less time followed by efficient decision making through the use of image processing technique for accurate results. The suggested method attained an average accuracy of 98.5%.
Authors: Ximeng Liu, Shizhen Zhao, Xinbing Wang
Abstract: Recently, researchers have explored ML-based Traffic Engineering (TE), leveraging neural networks to solve TE problems traditionally addressed by optimization. However, existing ML-based TE schemes remain impractical: they either fail to handle topology changes or suffer from poor scalability due to excessive computational and memory overhead. To overcome these limitations, we propose Geminet, a lightweight and scalable ML-based TE framework that can handle changing topologies. Geminet is built upon two key insights: (i) a methodology that decouples neural networks from topology by learning an iterative gradient-descent-based adjustment process, as the update rule of gradient descent is topology-agnostic, relying only on a few gradient-related quantities; (ii) shifting optimization from path-level routing weights to edge-level dual variables, reducing memory consumption by leveraging the fact that edges are far fewer than paths. Evaluations on WAN and data center datasets show that Geminet significantly improves scalability. Its neural network size is only 0.04% to 7% of existing schemes, while handling topology variations as effectively as HARP, a state-of-the-art ML-based TE approach, without performance degradation. When trained on large-scale topologies, Geminet consumes under 10 GiB of memory, more than eight times less than the 80-plus GiB required by HARP, while achieving 5.45 times faster convergence speed, demonstrating its potential for large-scale deployment.
Authors: Mario Koddenbrock, Rudolf Hoffmann, David Brodmann, Erik Rodner
Abstract: In real-world vision-language applications, practitioners increasingly rely on large, pretrained foundation models rather than custom-built solutions, despite limited transparency regarding their training data and processes. While these models achieve impressive performance on general benchmarks, their effectiveness can decline notably under specialized domain shifts, such as unique imaging conditions or environmental variations. In this work, we introduce Deepbench, a framework designed to assess domain-specific robustness of vision-language models (VLMs). Deepbench leverages a large language model (LLM) to generate realistic, context-aware image corruptions tailored to specific deployment domains without requiring labeled data. We evaluate a range of contrastive vision-language architectures and architectural variants across six real-world domains and observe substantial variability in robustness, highlighting the need for targeted, domain-aware evaluation. Deepbench is released as open-source software to support further research into domain-aware robustness assessment.
Authors: Xingting Yao, Qinghao Hu, Fei Zhou, Tielong Liu, Gang Li, Peisong Wang, Jian Cheng
Abstract: Multi-bit spiking neural networks (SNNs) have recently become a heated research spot, pursuing energy-efficient and high-accurate AI. However, with more bits involved, the associated memory and computation demands escalate to the point where the performance improvements become disproportionate. Based on the insight that different layers demonstrate different importance and extra bits could be wasted and interfering, this paper presents an adaptive bit allocation strategy for direct-trained SNNs, achieving fine-grained layer-wise allocation of memory and computation resources. Thus, SNN's efficiency and accuracy can be improved. Specifically, we parametrize the temporal lengths and the bit widths of weights and spikes, and make them learnable and controllable through gradients. To address the challenges caused by changeable bit widths and temporal lengths, we propose the refined spiking neuron, which can handle different temporal lengths, enable the derivation of gradients for temporal lengths, and suit spike quantization better. In addition, we theoretically formulate the step-size mismatch problem of learnable bit widths, which may incur severe quantization errors to SNN, and accordingly propose the step-size renewal mechanism to alleviate this issue. Experiments on various datasets, including the static CIFAR and ImageNet and the dynamic CIFAR-DVS and DVS-GESTURE, demonstrate that our methods can reduce the overall memory and computation cost while achieving higher accuracy. Particularly, our SEWResNet-34 can achieve a 2.69\% accuracy gain and 4.16$\times$ lower bit budgets over the advanced baseline work on ImageNet. This work will be fully open-sourced.
Authors: Gijs Luijten, Roberto Maria Scardigno, Lisle Faray de Paiva, Peter Hoyer, Jens Kleesiek, Domenico Buongiorno, Vitoantonio Bevilacqua, Jan Egger
Abstract: Ultrasound (US) is widely accessible and radiation-free but has a steep learning curve due to its dynamic nature and non-standard imaging planes. Additionally, the constant need to shift focus between the US screen and the patient poses a challenge. To address these issues, we integrate deep learning (DL)-based semantic segmentation for real-time (RT) automated kidney volumetric measurements, which are essential for clinical assessment but are traditionally time-consuming and prone to fatigue. This automation allows clinicians to concentrate on image interpretation rather than manual measurements. Complementing DL, augmented reality (AR) enhances the usability of US by projecting the display directly into the clinician's field of view, improving ergonomics and reducing the cognitive load associated with screen-to-patient transitions. Two AR-DL-assisted US pipelines on HoloLens-2 are proposed: one streams directly via the application programming interface for a wireless setup, while the other supports any US device with video output for broader accessibility. We evaluate RT feasibility and accuracy using the Open Kidney Dataset and open-source segmentation models (nnU-Net, Segmenter, YOLO with MedSAM and LiteMedSAM). Our open-source GitHub pipeline includes model implementations, measurement algorithms, and a Wi-Fi-based streaming solution, enhancing US training and diagnostics, especially in point-of-care settings.
Authors: Xue Wen Tan, Stanley Kok
Abstract: Every publicly traded U.S. company files an annual 10-K report containing critical insights into financial health and risk. We propose Tiny eXplainable Risk Assessor (TinyXRA), a lightweight and explainable transformer-based model that automatically assesses company risk from these reports. Unlike prior work that relies solely on the standard deviation of excess returns (adjusted for the Fama-French model), which indiscriminately penalizes both upside and downside risk, TinyXRA incorporates skewness, kurtosis, and the Sortino ratio for more comprehensive risk assessment. We leverage TinyBERT as our encoder to efficiently process lengthy financial documents, coupled with a novel dynamic, attention-based word cloud mechanism that provides intuitive risk visualization while filtering irrelevant terms. This lightweight design ensures scalable deployment across diverse computing environments with real-time processing capabilities for thousands of financial documents which is essential for production systems with constrained computational resources. We employ triplet loss for risk quartile classification, improving over pairwise loss approaches in existing literature by capturing both the direction and magnitude of risk differences. Our TinyXRA achieves state-of-the-art predictive accuracy across seven test years on a dataset spanning 2013-2024, while providing transparent and interpretable risk assessments. We conduct comprehensive ablation studies to evaluate our contributions and assess model explanations both quantitatively by systematically removing highly attended words and sentences, and qualitatively by examining explanation coherence. The paper concludes with findings, practical implications, limitations, and future research directions.
Authors: Shiao Wang, Ju Huang, Qingchuan Ma, Jinfeng Gao, Chunyi Xu, Xiao Wang, Lan Chen, Bo Jiang
Abstract: Combining traditional RGB cameras with bio-inspired event cameras for robust object tracking has garnered increasing attention in recent years. However, most existing multimodal tracking algorithms depend heavily on high-complexity Vision Transformer architectures for feature extraction and fusion across modalities. This not only leads to substantial computational overhead but also limits the effectiveness of cross-modal interactions. In this paper, we propose an efficient RGB-Event object tracking framework based on the linear-complexity Vision Mamba network, termed Mamba-FETrack V2. Specifically, we first design a lightweight Prompt Generator that utilizes embedded features from each modality, together with a shared prompt pool, to dynamically generate modality-specific learnable prompt vectors. These prompts, along with the modality-specific embedded features, are then fed into a Vision Mamba-based FEMamba backbone, which facilitates prompt-guided feature extraction, cross-modal interaction, and fusion in a unified manner. Finally, the fused representations are passed to the tracking head for accurate target localization. Extensive experimental evaluations on multiple RGB-Event tracking benchmarks, including short-term COESOT dataset and long-term datasets, i.e., FE108 and FELT V2, demonstrate the superior performance and efficiency of the proposed tracking framework. The source code and pre-trained models will be released on https://github.com/Event-AHU/Mamba_FETrack
Authors: Parosh Aziz Abdulla, Mohamed Faouzi Atig, Julie Cailler, Chencheng Liang, Philipp R\"ummer
Abstract: Nielsen transformation is a standard approach for solving word equations: by repeatedly splitting equations and applying simplification steps, equations are rewritten until a solution is reached. When solving a conjunction of word equations in this way, the performance of the solver will depend considerably on the order in which equations are processed. In this work, the use of Graph Neural Networks (GNNs) for ranking word equations before and during the solving process is explored. For this, a novel graph-based representation for word equations is presented, preserving global information across conjuncts, enabling the GNN to have a holistic view during ranking. To handle the variable number of conjuncts, three approaches to adapt a multi-classification task to the problem of ranking equations are proposed. The training of the GNN is done with the help of minimum unsatisfiable subsets (MUSes) of word equations. The experimental results show that, compared to state-of-the-art string solvers, the new framework solves more problems in benchmarks where each variable appears at most once in each equation.
Authors: Anton Andreychuk, Konstantin Yakovlev, Aleksandr Panov, Alexey Skrynnik
Abstract: Multi-agent pathfinding (MAPF) is a common abstraction of multi-robot trajectory planning problems, where multiple homogeneous robots simultaneously move in the shared environment. While solving MAPF optimally has been proven to be NP-hard, scalable, and efficient, solvers are vital for real-world applications like logistics, search-and-rescue, etc. To this end, decentralized suboptimal MAPF solvers that leverage machine learning have come on stage. Building on the success of the recently introduced MAPF-GPT, a pure imitation learning solver, we introduce MAPF-GPT-DDG. This novel approach effectively fine-tunes the pre-trained MAPF model using centralized expert data. Leveraging a novel delta-data generation mechanism, MAPF-GPT-DDG accelerates training while significantly improving performance at test time. Our experiments demonstrate that MAPF-GPT-DDG surpasses all existing learning-based MAPF solvers, including the original MAPF-GPT, regarding solution quality across many testing scenarios. Remarkably, it can work with MAPF instances involving up to 1 million agents in a single environment, setting a new milestone for scalability in MAPF domains.
Authors: Alexander Tyurin
Abstract: We consider centralized distributed optimization in the classical federated learning setup, where $n$ workers jointly find an $\varepsilon$-stationary point of an $L$-smooth, $d$-dimensional nonconvex function $f$, having access only to unbiased stochastic gradients with variance $\sigma^2$. Each worker requires at most $h$ seconds to compute a stochastic gradient, and the communication times from the server to the workers and from the workers to the server are $\tau_{s}$ and $\tau_{w}$ seconds per coordinate, respectively. One of the main motivations for distributed optimization is to achieve scalability with respect to $n$. For instance, it is well known that the distributed version of SGD has a variance-dependent runtime term $\frac{h \sigma^2 L \Delta}{n \varepsilon^2},$ which improves with the number of workers $n,$ where $\Delta = f(x^0) - f^*,$ and $x^0 \in R^d$ is the starting point. Similarly, using unbiased sparsification compressors, it is possible to reduce both the variance-dependent runtime term and the communication runtime term. However, once we account for the communication from the server to the workers $\tau_{s}$, we prove that it becomes infeasible to design a method using unbiased random sparsification compressors that scales both the server-side communication runtime term $\tau_{s} d \frac{L \Delta}{\varepsilon}$ and the variance-dependent runtime term $\frac{h \sigma^2 L \Delta}{\varepsilon^2},$ better than poly-logarithmically in $n$, even in the homogeneous (i.i.d.) case, where all workers access the same distribution. To establish this result, we construct a new "worst-case" function and develop a new lower bound framework that reduces the analysis to the concentration of a random sum, for which we prove a concentration bound. These results reveal fundamental limitations in scaling distributed optimization, even under the homogeneous assumption.
Authors: Travis Dick, Alessandro Epasto, Adel Javanmard, Josh Karlin, Andres Munoz Medina, Vahab Mirrokni, Sergei Vassilvitskii, Peilin Zhong
Abstract: The analysis of the privacy properties of Privacy-Preserving Ads APIs is an area of research that has received strong interest from academics, industry, and regulators. Despite this interest, the empirical study of these methods is hindered by the lack of publicly available data. Reliable empirical analysis of the privacy properties of an API, in fact, requires access to a dataset consisting of realistic API outputs; however, privacy concerns prevent the general release of such data to the public. In this work, we develop a novel methodology to construct synthetic API outputs that are simultaneously realistic enough to enable accurate study and provide strong privacy protections. We focus on one Privacy-Preserving Ads APIs: the Topics API, part of Google Chrome's Privacy Sandbox. We developed a methodology to generate a differentially-private dataset that closely matches the re-identification risk properties of the real Topics API data. The use of differential privacy provides strong theoretical bounds on the leakage of private user information from this release. Our methodology is based on first computing a large number of differentially-private statistics describing how output API traces evolve over time. Then, we design a parameterized distribution over sequences of API traces and optimize its parameters so that they closely match the statistics obtained. Finally, we create the synthetic data by drawing from this distribution. Our work is complemented by an open-source release of the anonymized dataset obtained by this methodology. We hope this will enable external researchers to analyze the API in-depth and replicate prior and future work on a realistic large-scale dataset. We believe that this work will contribute to fostering transparency regarding the privacy properties of Privacy-Preserving Ads APIs.
Authors: Louis Bradshaw, Honglu Fan, Alexander Spangher, Stella Biderman, Simon Colton
Abstract: We study the capabilities of generative autoregressive transformer models trained on large amounts of symbolic solo-piano transcriptions. After first pretraining on approximately 60,000 hours of music, we use a comparatively smaller, high-quality subset, to finetune models to produce musical continuations, perform symbolic classification tasks, and produce general-purpose contrastive MIDI embeddings by adapting the SimCLR framework to symbolic music. When evaluating piano continuation coherence, our generative model outperforms leading symbolic generation techniques and remains competitive with proprietary audio generation models. On MIR classification benchmarks, frozen representations from our contrastive model achieve state-of-the-art results in linear probe experiments, while direct finetuning demonstrates the generalizability of pretrained representations, often requiring only a few hundred labeled examples to specialize to downstream tasks.
Authors: Yuexuan Kong, Gabriel Meseguer-Brocal, Vincent Lostanlen, Mathieu Lagrange, Romain Hennequin
Abstract: In music information retrieval (MIR), contrastive self-supervised learning for general-purpose representation models is effective for global tasks such as automatic tagging. However, for local tasks such as chord estimation, it is widely assumed that contrastively trained general-purpose self-supervised models are inadequate and that more sophisticated SSL is necessary; e.g., masked modeling. Our paper challenges this assumption by revealing the potential of contrastive SSL paired with a transformer in local MIR tasks. We consider a lightweight vision transformer with one-dimensional patches in the time--frequency domain (ViT-1D) and train it with simple contrastive SSL through normalized temperature-scaled cross-entropy loss (NT-Xent). Although NT-Xent operates only over the class token, we observe that, potentially thanks to weight sharing, informative musical properties emerge in ViT-1D's sequence tokens. On global tasks, the temporal average of class and sequence tokens offers a performance increase compared to the class token alone, showing useful properties in the sequence tokens. On local tasks, sequence tokens perform unexpectedly well, despite not being specifically trained for. Furthermore, high-level musical features such as onsets emerge from layer-wise attention maps and self-similarity matrices show different layers capture different musical dimensions. Our paper does not focus on improving performance but advances the musical interpretation of transformers and sheds light on some overlooked abilities of contrastive SSL paired with transformers for sequence modeling in MIR.
Authors: Reihaneh Zohrabi, Hosein Hasani, Mahdieh Soleymani Baghshah, Anna Rohrbach, Marcus Rohrbach, Mohammad Hossein Rohban
Abstract: Out-of-distribution (OOD) detection is crucial for ensuring the reliability and safety of machine learning models in real-world applications, where they frequently face data distributions unseen during training. Despite progress, existing methods are often vulnerable to spurious correlations that mislead models and compromise robustness. To address this, we propose SPROD, a novel prototype-based OOD detection approach that explicitly addresses the challenge posed by unknown spurious correlations. Our post-hoc method refines class prototypes to mitigate bias from spurious features without additional data or hyperparameter tuning, and is broadly applicable across diverse backbones and OOD detection settings. We conduct a comprehensive spurious correlation OOD detection benchmarking, comparing our method against existing approaches and demonstrating its superior performance across challenging OOD datasets, such as CelebA, Waterbirds, UrbanCars, Spurious Imagenet, and the newly introduced Animals MetaCoCo. On average, SPROD improves AUROC by 4.7% and FPR@95 by 9.3% over the second best.
Authors: Andr\'as Gy\"orgy, Tor Lattimore, Nevena Lazi\'c, Csaba Szepesv\'ari
Abstract: Sound deductive reasoning -- the ability to derive new knowledge from existing facts and rules -- is an indisputably desirable aspect of general intelligence. Despite the major advances of AI systems in areas such as math and science, especially since the introduction of transformer architectures, it is well-documented that even the most advanced frontier systems regularly and consistently falter on easily-solvable deductive reasoning tasks. Hence, these systems are unfit to fulfill the dream of achieving artificial general intelligence capable of sound deductive reasoning. We argue that their unsound behavior is a consequence of the statistical learning approach powering their development. To overcome this, we contend that to achieve reliable deductive reasoning in learning-based AI systems, researchers must fundamentally shift from optimizing for statistical performance against distributions on reasoning problems and algorithmic tasks to embracing the more ambitious exact learning paradigm, which demands correctness on all inputs. We argue that exact learning is both essential and possible, and that this ambitious objective should guide algorithm design.
Authors: David B\'alik, Martin Jure\v{c}ek, Mark Stamp
Abstract: This work addresses the challenge of malware classification using machine learning by developing a novel dataset labeled at both the malware type and family levels. Raw binaries were collected from sources such as VirusShare, VX Underground, and MalwareBazaar, and subsequently labeled with family information parsed from binary names and type-level labels integrated from ClarAVy. The dataset includes 14 malware types and 17 malware families, and was processed using a unified feature extraction pipeline based on static analysis, particularly extracting features from Portable Executable headers, to support advanced classification tasks. The evaluation was focused on three key classification tasks. In the binary classification of malware versus benign samples, Random Forest and XGBoost achieved high accuracy on the full datasets, reaching 98.5% for type-based detection and 98.98% for family-based detection. When using truncated datasets of 1,000 samples to assess performance under limited data conditions, both models still performed strongly, achieving 97.6% for type-based detection and 98.66% for family-based detection. For interclass classification, which distinguishes between malware types or families, the models reached up to 97.5% accuracy on type-level tasks and up to 93.7% on family-level tasks. In the multiclass classification setting, which assigns samples to the correct type or family, SVM achieved 81.1% accuracy on type labels, while Random Forest and XGBoost reached approximately 73.4% on family labels. The results highlight practical trade-offs between accuracy and computational cost, and demonstrate that labeling at both the type and family levels enables more fine-grained and insightful malware classification. The work establishes a robust foundation for future research on advanced malware detection and classification.
Authors: Evan Bell, Daniel A. Serino, Ben S. Southworth, Trevor Wilcox, Marc L. Klasky
Abstract: Estimating physical parameters or material properties from experimental observations is a common objective in many areas of physics and material science. In many experiments, especially in shock physics, radiography is the primary means of observing the system of interest. However, radiography does not provide direct access to key state variables, such as density, which prevents the application of traditional parameter estimation approaches. Here we focus on flyer plate impact experiments on porous materials, and resolving the underlying parameterized equation of state (EoS) and crush porosity model parameters given radiographic observation(s). We use machine learning as a tool to demonstrate with high confidence that using only high impact velocity data does not provide sufficient information to accurately infer both EoS and crush model parameters, even with fully resolved density fields or a dynamic sequence of images. We thus propose an observable data set consisting of low and high impact velocity experiments/simulations that capture different regimes of compaction and shock propagation, and proceed to introduce a generative machine learning approach which produces a posterior distribution of physical parameters directly from radiographs. We demonstrate the effectiveness of the approach in estimating parameters from simulated flyer plate impact experiments, and show that the obtained estimates of EoS and crush model parameters can then be used in hydrodynamic simulations to obtain accurate and physically admissible density reconstructions. Finally, we examine the robustness of the approach to model mismatches, and find that the learned approach can provide useful parameter estimates in the presence of out-of-distribution radiographic noise and previously unseen physics, thereby promoting a potential breakthrough in estimating material properties from experimental radiographic images.
Authors: Germans Savcisens, Tina Eliassi-Rad
Abstract: We often attribute human characteristics to large language models (LLMs) and claim that they "know" certain things. LLMs have an internal probabilistic knowledge that represents information retained during training. How can we assess the veracity of this knowledge? We examine two common methods for probing the veracity of LLMs and discover several assumptions that are flawed. To address these flawed assumptions, we introduce sAwMIL (short for Sparse Aware Multiple-Instance Learning), a probing method that utilizes the internal activations of LLMs to separate statements into true, false, and neither. sAwMIL is based on multiple-instance learning and conformal prediction. We evaluate sAwMIL on 5 validity criteria across 16 open-source LLMs, including both default and chat-based variants, as well as on 3 new datasets. Among the insights we provide are: (1) the veracity signal is often concentrated in the third quarter of an LLM's depth; (2) truth and falsehood signals are not always symmetric; (3) linear probes perform better on chat models than on default models; (4) nonlinear probes may be required to capture veracity signals for some LLMs with reinforcement learning from human feedback or knowledge distillation; and (5) LLMs capture a third type of signal that is distinct from true and false and is neither true nor false. These findings provide a reliable method for verifying what LLMs "know" and how certain they are of their probabilistic internal knowledge.
Authors: Junping Wang, Bicheng Wang, Yibo Xuea, Yuan Xie
Abstract: Resilience non-equilibrium measurement, the ability to maintain fundamental functionality amidst failures and errors, is crucial for scientific management and engineering applications of industrial chain. The problem is particularly challenging when the number or types of multiple co-evolution of resilience (for example, randomly placed) are extremely chaos. Existing end-to-end deep learning ordinarily do not generalize well to unseen full-feld reconstruction of spatiotemporal co-evolution structure, and predict resilience of network topology, especially in multiple chaos data regimes typically seen in real-world applications. To address this challenge, here we propose industrial brain, a human-like autonomous cognitive decision-making and planning framework integrating higher-order activity-driven neuro network and CT-OODA symbolic reasoning to autonomous plan resilience directly from observational data of global variable. The industrial brain not only understands and model structure of node activity dynamics and network co-evolution topology without simplifying assumptions, and reveal the underlying laws hidden behind complex networks, but also enabling accurate resilience prediction, inference, and planning. Experimental results show that industrial brain significantly outperforms resilience prediction and planning methods, with an accurate improvement of up to 10.8\% over GoT and OlaGPT framework and 11.03\% over spectral dimension reduction. It also generalizes to unseen topologies and dynamics and maintains robust performance despite observational disturbances. Our findings suggest that industrial brain addresses an important gap in resilience prediction and planning for industrial chain.
Authors: Xiangchen Li, Saeid Ghafouri, Bo Ji, Hans Vandierendonck, Deepu John, Dimitrios S. Nikolopoulos
Abstract: As machine learning inferences increasingly move to edge devices, adapting to diverse computational capabilities, hardware, and memory constraints becomes more critical. Instead of relying on a pre-trained model fixed for all future inference queries across diverse edge devices, we argue that planning an inference pattern with a request-specific model tailored to the device's computational capacity, accuracy requirements, and time constraints is more cost-efficient and robust to diverse scenarios. To this end, we propose an accuracy-aware and workload-balanced inference system that integrates joint model quantization and inference partitioning. In this approach, the server dynamically responds to inference queries by sending a quantized model and adaptively sharing the inference workload with the device. Meanwhile, the device's computational power, channel capacity, and accuracy requirements are considered when deciding. Furthermore, we introduce a new optimization framework for the inference system, incorporating joint model quantization and partitioning. Our approach optimizes layer-wise quantization bit width and partition points to minimize time consumption and cost while accounting for varying accuracy requirements of tasks through an accuracy degradation metric in our optimization model. To our knowledge, this work represents the first exploration of optimizing quantization layer-wise bit-width in the inference serving system, by introducing theoretical measurement of accuracy degradation. Simulation results demonstrate a substantial reduction in overall time and power consumption, with computation payloads decreasing by over 80% and accuracy degradation kept below 1%.
Authors: Stefan Buijsman, Sarah Carter, Juan Pablo Berm\'udez
Abstract: AI systems increasingly support human decision-making across domains of professional, skill-based, and personal activity. While previous work has examined how AI might affect human autonomy globally, the effects of AI on domain-specific autonomy -- the capacity for self-governed action within defined realms of skill or expertise -- remain understudied. We analyze how AI decision-support systems affect two key components of domain-specific autonomy: skilled competence (the ability to make informed judgments within one's domain) and authentic value-formation (the capacity to form genuine domain-relevant values and preferences). By engaging with prior investigations and analyzing empirical cases across medical, financial, and educational domains, we demonstrate how the absence of reliable failure indicators and the potential for unconscious value shifts can erode domain-specific autonomy both immediately and over time. We then develop a constructive framework for autonomy-preserving AI support systems. We propose specific socio-technical design patterns -- including careful role specification, implementation of defeater mechanisms, and support for reflective practice -- that can help maintain domain-specific autonomy while leveraging AI capabilities. This framework provides concrete guidance for developing AI systems that enhance rather than diminish human agency within specialized domains of action.
Authors: Hongyu H\`e, Minhao Jin, Maria Apostolaki
Abstract: Network data conforms to a wide range of rules that arise from protocols, design principles, and deployment decisions (e.g., a packet's queuing delay must be less than its end-to-end delay). Formalizing such rules as logic constraints can (i) improve the quality of synthetic data, (ii) reduce the brittleness of machine learning (ML) models, and (iii) improve semantic understanding of network measurements. However, these benefits remain out of reach if rule extraction is manual or solely reliant on ML, as both approaches yield incomplete, unreliable, and/or inaccurate rules. This paper formulates rule extraction as a constraint modeling problem and introduces NetNomos that learns propositional logic constraints directly from raw network measurements. Constraint modeling in this domain is uniquely challenging due to the scale of the data, the inherent learning complexity and passive environment, and the lack of ground truth supervision. NetNomos addresses these challenges via a lattice-based search structured by constraint specificity and succinctness. Our approach reduces learning complexity from superquadratic to logarithmic and enables efficient traversal in combinatorial search space. Our evaluations on diverse network datasets show that NetNomos learns all benchmark rules, including those associated with as little as 0.01% of data points, in under three hours. In contrast, baseline methods discover less than 25% of the rules and require several days to run. Through three case studies, we show that: NetNomos (i) finds rule violations in the outputs of all seven synthetic traffic generators, hence can be used to assess and guide their generation process; (ii) detects semantic differences in traffic, hence can be used for anomaly detection; and (iii) automatically finds rules used for telemetry imputation, hence can support monitoring through inference.
Authors: Dustin Wright
Abstract: Scientific information expresses human understanding of nature. This knowledge is largely disseminated in different forms of text, including scientific papers, news articles, and discourse among people on social media. While important for accelerating our pursuit of knowledge, not all scientific text is faithful to the underlying science. As the volume of this text has burgeoned online in recent years, it has become a problem of societal importance to be able to identify the faithfulness of a given piece of scientific text automatically. This thesis is concerned with the cultivation of datasets, methods, and tools for machine understanding of scientific language, in order to analyze and understand science communication at scale. To arrive at this, I present several contributions in three areas of natural language processing and machine learning: automatic fact checking, learning with limited data, and scientific text processing. These contributions include new methods and resources for identifying check-worthy claims, adversarial claim generation, multi-source domain adaptation, learning from crowd-sourced labels, cite-worthiness detection, zero-shot scientific fact checking, detecting exaggerated scientific claims, and modeling degrees of information change in science communication. Critically, I demonstrate how the research outputs of this thesis are useful for effectively learning from limited amounts of scientific text in order to identify misinformative scientific statements and generate new insights into the science communication process
Authors: Masahiro Kato
Abstract: This study investigates adaptive experimental design for treatment choice, also known as fixed-budget best-arm identification. We consider an adaptive procedure consisting of a treatment-allocation phase followed by a treatment-choice phase, and we design an adaptive experiment for this setup to efficiently identify the best treatment arm, defined as the one with the highest expected outcome. In our designed experiment, the treatment-allocation phase consists of two stages. The first stage is a pilot phase, where we allocate each treatment arm uniformly with equal proportions to eliminate clearly suboptimal arms and estimate outcome variances. In the second stage, we allocate treatment arms in proportion to the variances estimated in the first stage. After the treatment-allocation phase, the procedure enters the treatment-choice phase, where we choose the treatment arm with the highest sample mean as our estimate of the best treatment arm. We prove that this single design is simultaneously asymptotically minimax and Bayes optimal for the simple regret, with upper bounds that match our lower bounds up to exact constants. Therefore, our designed experiment achieves the sharp efficiency limits without requiring separate tuning for minimax and Bayesian objectives.
Authors: Nicolas Heintz, Tom Francart, Alexander Bertrand
Abstract: Auditory attention decoding (AAD) algorithms exploit brain signals, such as electroencephalography (EEG), to identify which speaker a listener is focusing on in a multi-speaker environment. While state-of-the-art AAD algorithms can identify the attended speaker on short time windows, their predictions are often too inaccurate for practical use. In this work, we propose augmenting AAD with a hidden Markov model (HMM) that models the temporal structure of attention. More specifically, the HMM relies on the fact that a subject is much less likely to switch attention than to keep attending the same speaker at any moment in time. We show how a HMM can significantly improve existing AAD algorithms in both causal (real-time) and non-causal (offline) settings. We further demonstrate that HMMs outperform existing postprocessing approaches in both accuracy and responsiveness, and explore how various factors such as window length, switching frequency, and AAD accuracy influence overall performance. The proposed method is computationally efficient, intuitive to use and applicable in both real-time and offline settings.
Authors: Alexis Melot, Sean U. N. Wood, Yannick Coffinier, Pierre Yger, Fabien Alibart
Abstract: Spike sorting is a crucial step in decoding multichannel extracellular neural signals, enabling the identification of individual neuronal activity. A key challenge in brain-machine interfaces (BMIs) is achieving real-time, low-power spike sorting at the edge while keeping high neural decoding performance. This study introduces the Neuromorphic Sparse Sorter (NSS), a compact two-layer spiking neural network optimized for efficient spike sorting. NSS leverages the Locally Competitive Algorithm (LCA) for sparse coding to extract relevant features from noisy events with reduced computational demands. NSS learns to sort detected spike waveforms in an online fashion and operates entirely unsupervised. To exploit multi-bit spike coding capabilities of neuromorphic platforms like Intel's Loihi 2, a custom neuron model was implemented, enabling flexible power-performance trade-offs via adjustable spike bit-widths. Evaluations on simulated and real-world tetrode signals with biological drift showed NSS outperformed established pipelines such as WaveClus3 and PCA+KMeans. With 2-bit graded spikes, NSS on Loihi 2 outperformed NSS implemented with leaky integrate-and-fire neuron and achieved an F1-score of 77% (+10% improvement) while consuming 8.6mW (+1.65mW) when tested on a drifting recording, with a computational processing time of 0.25ms (+60 us) per inference.
Authors: Xinming Wei, Jiahao Zhang, Haoran Li, Jiayu Chen, Rui Qu, Maoliang Li, Xiang Chen, Guojie Luo
Abstract: The proliferation of agentic Large Language Models (LLMs) on personal devices introduces a new class of workloads characterized by a dichotomy of objectives. Reactive tasks, initiated by users, demand immediate, low-latency responses, while proactive tasks operate invisibly and prioritize throughput. Existing on-device LLM engines, designed for isolated inferences, fail to efficiently manage these concurrent and conflicting requests on consumer-grade heterogeneous SoCs with CPU, integrated GPU, and NPU. This paper introduces Agent.xpu, an efficient serving system for agentic LLM workloads on memory-unified heterogeneous SoCs. With dedicated offline profiling, Agent.xpu first constructs a heterogeneous execution graph, which fuses and chunks model kernels for affinity-guided, elastic accelerator mapping with predictive kernel annotation. At runtime, its online scheduler enables fine-grained, kernel-level preemption to guarantee the responsiveness of reactive tasks. To maximize SoC utilization, it adopts slack-aware kernel backfill to opportunistically append proactive tasks, and mitigates NPU-iGPU contention via bandwidth-aware dispatch. Evaluation on an Intel Core Ultra SoC shows that Agent.xpu achieves 4.6$\times$ lower latency for reactive tasks and sustains 1.6$\times$-6.8$\times$ higher throughput for proactive tasks compared to state-of-the-art inference engines.
Authors: Tim Roith, Leon Bungert, Philipp Wacker
Abstract: Consensus-based optimization (CBO) has established itself as an efficient gradient-free optimization scheme, with attractive mathematical properties, such as mean-field convergence results for non-convex loss functions. In this work, we study CBO in the context of closed-box adversarial attacks, which are imperceptible input perturbations that aim to fool a classifier, without accessing its gradient. Our contribution is to establish a connection between the so-called consensus hopping as introduced by Riedl et al. and natural evolution strategies (NES) commonly applied in the context of adversarial attacks and to rigorously relate both methods to gradient-based optimization schemes. Beyond that, we provide a comprehensive experimental study that shows that despite the conceptual similarities, CBO can outperform NES and other evolutionary strategies in certain scenarios.
Authors: Tung-Ling Li, Hongliang Liu
Abstract: We introduce logit-gap steering, a fast jailbreak framework that casts the refusal-affirmation gap of RLHF-aligned language models as a single pass over the vocabulary. A forward-computable score blends gap reduction with lightweight proxies for KL penalty and reward shift, allowing a "sort-sum-stop" sweep to complete in under a second and return a short suffix--two orders of magnitude fewer model calls than beam or gradient attacks. The same suffix generalises to unseen prompts and scales from 0.5 B to 70 B checkpoints, lifting one-shot attack success from baseline levels to 80-100% while preserving topical coherence. Beyond efficiency, these suffixes expose sentence-boundary reward cliffs and other alignment artefacts, offering a lightweight probe into how safety tuning reshapes internal representations.
Authors: Rahul Kumar, Wenqi Wei, Ying Mao, Junaid Farooq, Ying Wang, Juntao Chen
Abstract: We propose a circuit-level attack, SQUASH, a SWAP-Based Quantum Attack to sabotage Hybrid Quantum Neural Networks (HQNNs) for classification tasks. SQUASH is executed by inserting SWAP gate(s) into the variational quantum circuit of the victim HQNN. Unlike conventional noise-based or adversarial input attacks, SQUASH directly manipulates the circuit structure, leading to qubit misalignment and disrupting quantum state evolution. This attack is highly stealthy, as it does not require access to training data or introduce detectable perturbations in input states. Our results demonstrate that SQUASH significantly degrades classification performance, with untargeted SWAP attacks reducing accuracy by up to 74.08\% and targeted SWAP attacks reducing target class accuracy by up to 79.78\%. These findings reveal a critical vulnerability in HQNN implementations, underscoring the need for more resilient architectures against circuit-level adversarial interventions.
Authors: Shai Yehezkel, Omer Dahary, Andrey Voynov, Daniel Cohen-Or
Abstract: Denoising diffusion models excel at generating high-quality images conditioned on text prompts, yet their effectiveness heavily relies on careful guidance during the sampling process. Classifier-Free Guidance (CFG) provides a widely used mechanism for steering generation by setting the guidance scale, which balances image quality and prompt alignment. However, the choice of the guidance scale has a critical impact on the convergence toward a visually appealing and prompt-adherent image. In this work, we propose an annealing guidance scheduler which dynamically adjusts the guidance scale over time based on the conditional noisy signal. By learning a scheduling policy, our method addresses the temperamental behavior of CFG. Empirical results demonstrate that our guidance scheduler significantly enhances image quality and alignment with the text prompt, advancing the performance of text-to-image generation. Notably, our novel scheduler requires no additional activations or memory consumption, and can seamlessly replace the common classifier-free guidance, offering an improved trade-off between prompt alignment and quality.
Authors: Bo Liu, Leon Guertler, Simon Yu, Zichen Liu, Penghui Qi, Daniel Balcells, Mickel Liu, Cheston Tan, Weiyan Shi, Min Lin, Wee Sun Lee, Natasha Jaques
Abstract: Recent advances in reinforcement learning have shown that language models can develop sophisticated reasoning through training on tasks with verifiable rewards, but these approaches depend on human-curated problem-answer pairs and domain-specific reward engineering. We introduce SPIRAL, a self-play framework where models learn by playing multi-turn, zero-sum games against continuously improving versions of themselves, eliminating the need for human supervision. Through self-play, SPIRAL generates an infinite curriculum of progressively challenging problems as models must constantly adapt to stronger opponents. To enable this self-play training at scale, We implement a fully online, multi-turn, multi-agent reinforcement learning system for LLMs and propose role-conditioned advantage estimation (RAE) to stabilize multi-agent training. Using SPIRAL, self-play on zero-sum games produces reasoning capabilities that transfer broadly. Training Qwen3-4B-Base on Kuhn Poker alone achieves 8.6% improvement on math and 8.4% on general reasoning, outperforming SFT on 25,000 expert game trajectories. Analysis reveals that this transfer occurs through three cognitive patterns: systematic decomposition, expected value calculation, and case-by-case analysis. Multi-game training (TicTacToe, Kuhn Poker, Simple Negotiation) further enhances performance as each game develops distinct reasoning strengths. Applying SPIRAL to a strong reasoning model (DeepSeek-R1-Distill-Qwen-7B) can still lead to 2.0% average improvement. These results demonstrate that zero-sum games naturally develop transferable reasoning capabilities, highlighting a promising direction for autonomous reasoning development.
Authors: Keqin Liu, Tianshuo Zheng, Zhi-Hua Zhou
Abstract: The multi-armed bandit (MAB) problem is a widely studied model in the field of operations research for sequential decision making and reinforcement learning. This paper mainly considers the classical MAB model with the heavy-tailed reward distributions. We introduce the extended robust UCB policy, which is an extension of the pioneering UCB policies proposed by Bubeck et al. [5] and Lattimore [22]. The previous UCB policies require some strict conditions on the reward distributions, which can be hard to guarantee in practical scenarios. Our extended robust UCB generalizes Lattimore's seminary work (for moments of orders $p=4$ and $q=2$) to arbitrarily chosen $p>q>1$ as long as the two moments have a known controlled relationship, while still achieving the optimal regret growth order $O(log T)$, thus providing a broadened application area of the UCB policies for the heavy-tailed reward distributions. Furthermore, we achieve a near-optimal regret order without any knowledge of the reward distributions as long as their $p$-th moments exist for some $p>1$.
Authors: Xiaoteng Ma, Junyao Chen, Li Xia, Jun Yang, Qianchuan Zhao, Zhengyuan Zhou
Abstract: We present Distributional Soft Actor-Critic (DSAC), a distributional reinforcement learning (RL) algorithm that combines the strengths of distributional information of accumulated rewards and entropy-driven exploration from Soft Actor-Critic (SAC) algorithm. DSAC models the randomness in both action and rewards, surpassing baseline performances on various continuous control tasks. Unlike standard approaches that solely maximize expected rewards, we propose a unified framework for risk-sensitive learning, one that optimizes the risk-related objective while balancing entropy to encourage exploration. Extensive experiments demonstrate DSAC's effectiveness in enhancing agent performances for both risk-neutral and risk-sensitive control tasks.
Authors: Lingzhi Gao, Zhenyuan Zhang, Chao Wu
Abstract: While existing federated learning approaches primarily focus on aggregating local models to construct a global model, in realistic settings, some clients may be reluctant to share their private models due to the inclusion of privacy-sensitive information. Knowledge distillation, which can extract model knowledge without accessing model parameters, is well-suited for this federated scenario. However, most distillation methods in federated learning (federated distillation) require a proxy dataset, which is difficult to obtain in the real world. Therefore, in this paper, we introduce a distributed three-player Generative Adversarial Network (GAN) to implement data-free mutual distillation and propose an effective method called FedDTG. We confirmed that the fake samples generated by GAN can make federated distillation more efficient and robust. Additionally, the distillation process between clients can deliver good individual client performance while simultaneously acquiring global knowledge and protecting data privacy. Our extensive experiments on benchmark vision datasets demonstrate that our method outperforms other federated distillation algorithms in terms of generalization.
Authors: Haosen Ge, Hamsa Bastani, Osbert Bastani
Abstract: Existing approaches to algorithmic fairness aim to ensure equitable outcomes if human decision-makers comply perfectly with algorithmic decisions. However, perfect compliance with the algorithm is rarely a reality or even a desirable outcome in human-AI collaboration. Yet, recent studies have shown that selective compliance with fair algorithms can amplify discrimination relative to the prior human policy. As a consequence, ensuring equitable outcomes requires fundamentally different algorithmic design principles that ensure robustness to the decision-maker's (a priori unknown) compliance pattern. We define the notion of compliance-robustly fair algorithmic recommendations that are guaranteed to (weakly) improve fairness in decisions, regardless of the human's compliance pattern. We propose a simple optimization strategy to identify the best performance-improving compliance-robustly fair policy. However, we show that it may be infeasible to design algorithmic recommendations that are simultaneously fair in isolation, compliance-robustly fair, and more accurate than the human policy; thus, if our goal is to improve the equity and accuracy of human-AI collaboration, it may not be desirable to enforce traditional algorithmic fairness constraints. We illustrate the value of our approach on criminal sentencing data before and after the introduction of an algorithmic risk assessment tool in Virginia.
Authors: Taejin Kim, Jiarui Li, Shubhranshu Singh, Nikhil Madaan, Carlee Joe-Wong
Abstract: The delicate equilibrium between user privacy and the ability to unleash the potential of distributed data is an important concern. Federated learning, which enables the training of collaborative models without sharing of data, has emerged as a privacy-centric solution. This approach brings forth security challenges, notably poisoning and backdoor attacks where malicious entities inject corrupted data into the training process, as well as evasion attacks that aim to induce misclassifications at test time. Our research investigates the intersection of adversarial training, a common defense method against evasion attacks, and backdoor attacks within federated learning. We introduce Adversarial Robustness Unhardening (ARU), which is employed by a subset of adversarial clients to intentionally undermine model robustness during federated training, rendering models susceptible to a broader range of evasion attacks. We present extensive experiments evaluating ARU's impact on adversarial training and existing robust aggregation defenses against poisoning and backdoor attacks. Our results show that ARU can substantially undermine adversarial training's ability to harden models against test-time evasion attacks, and that adversaries employing ARU can even evade robust aggregation defenses that often neutralize poisoning or backdoor attacks.
Authors: Xing Shen, Hengguan Huang, Brennan Nichyporuk, Tal Arbel
Abstract: Once deployed, medical image analysis methods are often faced with unexpected image corruptions and noise perturbations. These unknown covariate shifts present significant challenges to deep learning based methods trained on "clean" images. This often results in unreliable predictions and poorly calibrated confidence, hence hindering clinical applicability. While recent methods have been developed to address specific issues such as confidence calibration or adversarial robustness, no single framework effectively tackles all these challenges simultaneously. To bridge this gap, we propose LaDiNE, a novel ensemble learning method combining the robustness of Vision Transformers with diffusion-based generative models for improved reliability in medical image classification. Specifically, transformer encoder blocks are used as hierarchical feature extractors that learn invariant features from images for each ensemble member, resulting in features that are robust to input perturbations. In addition, diffusion models are used as flexible density estimators to estimate member densities conditioned on the invariant features, leading to improved modeling of complex data distributions while retaining properly calibrated confidence. Extensive experiments on tuberculosis chest X-rays and melanoma skin cancer datasets demonstrate that LaDiNE achieves superior performance compared to a wide range of state-of-the-art methods by simultaneously improving prediction accuracy and confidence calibration under unseen noise, adversarial perturbations, and resolution degradation.
Authors: Lukas Sch\"afer, Logan Jones, Anssi Kanervisto, Yuhan Cao, Tabish Rashid, Raluca Georgescu, Dave Bignell, Siddhartha Sen, Andrea Trevi\~no Gavito, Sam Devlin
Abstract: Video games have served as useful benchmarks for the decision-making community, but going beyond Atari games towards modern games has been prohibitively expensive for the vast majority of the research community. Prior work in modern video games typically relied on game-specific integration to obtain game features and enable online training, or on existing large datasets. An alternative approach is to train agents using imitation learning to play video games purely from images. However, this setting poses a fundamental question: which visual encoders obtain representations that retain information critical for decision making? To answer this question, we conduct a systematic study of imitation learning with publicly available pre-trained visual encoders compared to the typical task-specific end-to-end training approach in Minecraft, Counter-Strike: Global Offensive, and Minecraft Dungeons. Our results show that end-to-end training can be effective with comparably low-resolution images and only minutes of demonstrations, but significant improvements can be gained by utilising pre-trained encoders such as DINOv2 depending on the game. In addition to enabling effective decision making, we show that pre-trained encoders can make decision-making research in video games more accessible by significantly reducing the cost of training.
Authors: Qilong Yan, Yufeng Zhang, Jinghao Zhang, Jingpu Duan, Jian Yin
Abstract: Graph Neural Networks (GNNs) have made significant advancements in node classification, but their success relies on sufficient labeled nodes per class in the training data. Real-world graph data often exhibits a long-tail distribution with sparse labels, emphasizing the importance of GNNs' ability in few-shot node classification, which entails categorizing nodes with limited data. Traditional episodic meta-learning approaches have shown promise in this domain, but they face an inherent limitation: it might lead the model to converge to suboptimal solutions because of random and uniform task assignment, ignoring task difficulty levels. This could lead the meta-learner to face complex tasks too soon, hindering proper learning. Ideally, the meta-learner should start with simple concepts and advance to more complex ones, like human learning. So, we introduce CPT, a novel two-stage curriculum learning method that aligns task difficulty with the meta-learner's progressive competence, enhancing overall performance. Specifically, in CPT's initial stage, the focus is on simpler tasks, fostering foundational skills for engaging with complex tasks later. Importantly, the second stage dynamically adjusts task difficulty based on the meta-learner's growing competence, aiming for optimal knowledge acquisition. Extensive experiments on popular node classification datasets demonstrate significant improvements of our strategy over existing methods.
Authors: Yancheng Wang, Yingzhen Yang
Abstract: Graph Neural Networks (GNNs) have achieved remarkable success in learning node representations and have shown strong performance in tasks such as node classification. However, recent findings indicate that the presence of noise in real-world graph data can substantially impair the effectiveness of GNNs. To address this challenge, we introduce a robust and innovative node representation learning method named Graph Contrastive Learning with Low-Rank Regularization, or GCL-LRR, which follows a two-stage transductive learning framework for node classification. In the first stage, the GCL-LRR encoder is optimized through prototypical contrastive learning while incorporating a low-rank regularization objective. In the second stage, the representations generated by GCL-LRR are employed by a linear transductive classifier to predict the labels of unlabeled nodes within the graph. Our GCL-LRR is inspired by the Low Frequency Property (LFP) of the graph data and its labels, and it is also theoretically motivated by our sharp generalization bound for transductive learning. To the best of our knowledge, our theoretical result is among the first to theoretically demonstrate the advantage of low-rank regularization in transductive learning, which is also supported by strong empirical results. To further enhance the performance of GCL-LRR, we present an improved model named GCL-LR-Attention, which incorporates a novel LR-Attention layer into GCL-LRR. GCL-LR-Attention reduces the kernel complexity of GCL-LRR and contributes to a tighter generalization bound, leading to improved performance. Extensive evaluations on standard benchmark datasets evidence the effectiveness and robustness of both GCL-LRR and GCL-LR-Attention in learning meaningful node representations. The code is available at https://github.com/Statistical-Deep-Learning/GCL-LR-Attention.
URLs: https://github.com/Statistical-Deep-Learning/GCL-LR-Attention.
Authors: H{\aa}kon Hanisch Kj{\ae}rnli, Lluis Mas-Ribas, Hans Jakob H{\aa}land, Vegard Sj{\aa}vik, Aida Ashrafi, Helge Langseth, Odd Erik Gundersen
Abstract: When deploying time series forecasting models based on machine learning to real world settings, one often encounter situations where the data distribution drifts. Such drifts expose the forecasting models to out-of-distribution (OOD) data, and machine learning models lack robustness in these settings. Robustness can be improved by using deep generative models or genetic algorithms to augment time series datasets, but these approaches lack interpretability and are computationally expensive. In this work, we develop an interpretable and simple framework for generating time series. Our method combines time-series decompositions with analytic functions, and is able to generate time series with characteristics matching both in- and out-of-distribution data. This approach allows users to generate new time series in an interpretable fashion, which can be used to augment the dataset and improve forecasting robustness. We demonstrate our framework through EXPRTS, a visual analytics tool designed for univariate time series forecasting models and datasets. Different visualizations of the data distribution, forecasting errors and single time series instances enable users to explore time series datasets, apply transformations, and evaluate forecasting model robustness across diverse scenarios. We show how our framework can generate meaningful OOD time series that improve model robustness, and we validate EXPRTS effectiveness and usability through three use-cases and a user study.
Authors: Junhoo Lee, Hyunho Lee, Kyomin Hwang, Nojun Kwak
Abstract: Deep learning has achieved tremendous success. However, unlike SVMs, which provide direct decision criteria and can be trained with a small dataset, it still has significant weaknesses due to its requirement for massive datasets during training and the black-box characteristics on decision criteria. This paper addresses these issues by identifying support vectors in deep learning models. To this end, we propose the DeepKKT condition, an adaptation of the traditional Karush-Kuhn-Tucker (KKT) condition for deep learning models, and confirm that generated Deep Support Vectors (DSVs) using this condition exhibit properties similar to traditional support vectors. This allows us to apply our method to few-shot dataset distillation problems and alleviate the black-box characteristics of deep learning models. Additionally, we demonstrate that the DeepKKT condition can transform conventional classification models into generative models with high fidelity, particularly as latent generative models using class labels as latent variables. We validate the effectiveness of DSVs using common datasets (ImageNet, CIFAR10 and CIFAR100) on the general architectures (ResNet and ConvNet), proving their practical applicability.
Authors: Archer Amon, Zhipeng Yin, Zichong Wang, Avash Palikhe, Wenbin Zhang
Abstract: Generative AI is becoming increasingly prevalent in creative fields, sparking urgent debates over how current copyright laws can keep pace with technological innovation. Recent controversies of AI models generating near-replicas of copyrighted material highlight the need to adapt current legal frameworks and develop technical methods to mitigate copyright infringement risks. This task requires understanding the intersection between computational concepts such as large-scale data scraping and probabilistic content generation, legal definitions of originality and fair use, and economic impacts on IP rights holders. However, most existing research on copyright in AI takes a purely computer science or law-based approach, leaving a gap in coordinating these approaches that only multidisciplinary efforts can effectively address. To bridge this gap, our survey adopts a comprehensive approach synthesizing insights from law, policy, economics, and computer science. It begins by discussing the foundational goals and considerations that should be applied to copyright in generative AI, followed by methods for detecting and assessing potential violations in AI system outputs. Next, it explores various regulatory options influenced by legal, policy, and economic frameworks to manage and mitigate copyright concerns associated with generative AI and reconcile the interests of IP rights holders with that of generative AI producers. The discussion then introduces techniques to safeguard individual creative works from unauthorized replication, such as watermarking and cryptographic protections. Finally, it describes advanced training strategies designed to prevent AI models from reproducing protected content. In doing so, we highlight key opportunities for action and offer actionable strategies that creators, developers, and policymakers can use in navigating the evolving copyright landscape.
Authors: Luigi Palmieri, Chiara Boldrini, Lorenzo Valerio, Andrea Passarella, Marco Conti, J\'anos Kert\'esz
Abstract: In the vibrant landscape of AI research, decentralised learning is gaining momentum. Decentralised learning allows individual nodes to keep data locally where they are generated and to share knowledge extracted from local data among themselves through an interactive process of collaborative refinement. This paradigm supports scenarios where data cannot leave local nodes due to privacy or sovereignty reasons or real-time constraints imposing proximity of models to locations where inference has to be carried out. The distributed nature of decentralised learning implies significant new research challenges with respect to centralised learning. Among them, in this paper, we focus on robustness issues. Specifically, we study the effect of nodes' disruption on the collective learning process. Assuming a given percentage of "central" nodes disappear from the network, we focus on different cases, characterised by (i) different distributions of data across nodes and (ii) different times when disruption occurs with respect to the start of the collaborative learning task. Through these configurations, we are able to show the non-trivial interplay between the properties of the network connecting nodes, the persistence of knowledge acquired collectively before disruption or lack thereof, and the effect of data availability pre- and post-disruption. Our results show that decentralised learning processes are remarkably robust to network disruption. As long as even minimum amounts of data remain available somewhere in the network, the learning process is able to recover from disruptions and achieve significant classification accuracy. This clearly varies depending on the remaining connectivity after disruption, but we show that even nodes that remain completely isolated can retain significant knowledge acquired before the disruption.
Authors: Yinsong Chen, Samson S. Yu, Zhong Li, Chee Peng Lim
Abstract: In recent years, inconsistency in Bayesian deep learning has attracted significant attention. Tempered or generalized posterior distributions are frequently employed as direct and effective solutions. Nonetheless, the underlying mechanisms and the effectiveness of generalized posteriors remain active research topics. In this work, we interpret posterior tempering as a correction for model misspecification via adjustments to the joint probability, and as a recalibration of priors by reducing aleatoric uncertainty. We also identify a unique property of the Laplace approximation: the generalized normalizing constant remains invariant, in contrast to general Bayesian learning, where this constant typically depends on model parameters after generalization. Leveraging this property, we introduce the generalized Laplace approximation, which requires only a simple modification to the Hessian calculation of the regularized loss. This approach provides a flexible and scalable framework for high-quality posterior inference. We evaluate the proposed method on state-of-the-art neural networks and real-world datasets, demonstrating that the generalized Laplace approximation enhances predictive performance.
Authors: Sergei Krutikov (Booking.com), Bulat Khaertdinov (Maastricht University), Rodion Kiriukhin (Booking.com), Shubham Agrawal (Booking.com), Mozhdeh Ariannezhad (Booking.com), Kees Jan De Vries (Booking.com)
Abstract: Transformer-based neural networks, empowered by Self-Supervised Learning (SSL), have demonstrated unprecedented performance across various domains. However, related literature suggests that tabular Transformers may struggle to outperform classical Machine Learning algorithms, such as Gradient Boosted Decision Trees (GBDT). In this paper, we aim to challenge GBDTs with tabular Transformers on a typical task faced in e-commerce, namely fraud detection. Our study is additionally motivated by the problem of selection bias, often occurring in real-life fraud detection systems. It is caused by the production system affecting which subset of traffic becomes labeled. This issue is typically addressed by sampling randomly a small part of the whole production data, referred to as a Control Group. This subset follows a target distribution of production data and therefore is usually preferred for training classification models with standard ML algorithms. Our methodology leverages the capabilities of Transformers to learn transferable representations using all available data by means of SSL, giving it an advantage over classical methods. Furthermore, we conduct large-scale experiments, pre-training tabular Transformers on vast amounts of data instances and fine-tuning them on smaller target datasets. The proposed approach outperforms heavily tuned GBDTs by a considerable margin of the Average Precision (AP) score in offline evaluations. Finally, we report the results of an online A/B experiment. Experimental results confirm the superiority of tabular Transformers compared to GBDTs in production, demonstrated by a statistically significant improvement in our business metric.
Authors: Yash Sinha, Murari Mandal, Mohan Kankanhalli
Abstract: User data spread across multiple modalities has popularized multi-modal recommender systems (MMRS). They recommend diverse content such as products, social media posts, TikTok reels, etc., based on a user-item interaction graph. With rising data privacy demands, recent methods propose unlearning private user data from uni-modal recommender systems (RS). However, methods for unlearning item data related to outdated user preferences, revoked licenses, and legally requested removals are still largely unexplored. Previous RS unlearning methods are unsuitable for MMRS due to the incompatibility of their matrix-based representation with the multi-modal user-item interaction graph. Moreover, their data partitioning step degrades performance on each shard due to poor data heterogeneity and requires costly performance aggregation across shards. This paper introduces MMRecUn, the first approach known to us for unlearning in MMRS and unlearning item data. Given a trained RS model, MMRecUn employs a novel Reverse Bayesian Personalized Ranking (BPR) objective to enable the model to forget marked data. The reverse BPR attenuates the impact of user-item interactions within the forget set, while the forward BPR reinforces the significance of user-item interactions within the retain set. Our experiments demonstrate that MMRecUn outperforms baseline methods across various unlearning requests when evaluated on benchmark MMRS datasets. MMRecUn achieves recall performance improvements of up to 49.85% compared to baseline methods and is up to 1.3x faster than the Gold model, which is trained on retain set from scratch. MMRecUn offers significant advantages, including superiority in removing target interactions, preserving retained interactions, and zero overhead costs compared to previous methods. Code: https://github.com/MachineUnlearn/MMRecUN Extended version: arXiv:2405.15328
Authors: Hong Liu, Xiuxiu Qiu, Yiming Shi, Miao Xu, Zelin Zang, Zhen Lei
Abstract: Unsupervised fault detection in multivariate time series plays a vital role in ensuring the stable operation of complex systems. Traditional methods often assume that normal data follow a single Gaussian distribution and identify anomalies as deviations from this distribution. {\color{black} However, this simplified assumption fails to capture the diversity and structural complexity of real-world time series, which can lead to misjudgments and reduced detection performance in practical applications. To address this issue, we propose a new method that combines a neighborhood-driven data augmentation strategy with a multi-manifold representation learning framework.} By incorporating information from local neighborhoods, the augmentation module can simulate contextual variations of normal data, enhancing the model's adaptability to distributional changes. In addition, we design a structure-aware feature learning approach that encourages natural clustering of similar patterns in the feature space while maintaining sufficient distinction between different operational states. Extensive experiments on several public benchmark datasets demonstrate that our method achieves superior performance in terms of both accuracy and robustness, showing strong potential for generalization and real-world deployment.
Authors: Nienke Nijkamp, June Sallou, Niels van der Heijden, Lu\'is Cruz
Abstract: Integrating Artificial Intelligence (AI) into software systems has significantly enhanced their capabilities while escalating energy demands. Ensemble learning, combining predictions from multiple models to form a single prediction, intensifies this problem due to cumulative energy consumption. This paper presents a novel approach to model selection that addresses the challenge of balancing the accuracy of AI models with their energy consumption in a live AI ensemble system. We explore how reducing the number of models or improving the efficiency of model usage within an ensemble during inference can reduce energy demands without substantially sacrificing accuracy. This study introduces and evaluates two model selection strategies, Static and Dynamic, for optimizing ensemble learning systems performance while minimizing energy usage. Our results demonstrate that the Static strategy improves the F1 score beyond the baseline, reducing average energy usage from 100% from the full ensemble to 62%. The Dynamic strategy further enhances F1 scores, using on average 76% compared to 100% of the full ensemble. Moreover, we propose an approach that balances accuracy with resource consumption, significantly reducing energy usage without substantially impacting accuracy. This method decreased the average energy usage of the Static strategy from approximately 62% to 14%, and for the Dynamic strategy, from around 76% to 57%. Our field study of Green AI using an operational AI system developed by a large professional services provider shows the practical applicability of adopting energy-conscious model selection strategies in live production environments.
Authors: Renato M. Silva, Gregorio F. Azevedo, Matheus V. V. Berto, Jean R. Rocha, Eduardo C. Fidelis, Matheus V. Nogueira, Pedro H. Lisboa, Tiago A. Almeida
Abstract: Traffic incidents involving vulnerable road users (VRUs) constitute a significant proportion of global road accidents. Advances in traffic communication ecosystems, coupled with sophisticated signal processing and machine learning techniques, have facilitated the utilization of data from diverse sensors. Despite these advancements and the availability of extensive datasets, substantial progress is required to mitigate traffic casualties. This paper provides a comprehensive survey of state-of-the-art technologies and methodologies to enhance the safety of VRUs. The study investigates the communication networks between vehicles and VRUs, emphasizing the integration of advanced sensors and the availability of relevant datasets. It explores preprocessing techniques and data fusion methods to enhance sensor data quality. Furthermore, our study assesses critical simulation environments essential for developing and testing VRU safety systems. Our research also highlights recent advances in VRU detection and classification algorithms, addressing challenges such as variable environmental conditions. Additionally, we cover cutting-edge research in predicting VRU intentions and behaviors, which is mandatory for proactive collision avoidance strategies. Through this survey, we aim to provide a comprehensive understanding of the current landscape of VRU safety technologies, identifying areas of progress and areas needing further research and development.
Authors: Adrian Pekar, Richard Jozsa
Abstract: This study investigates the efficacy of machine learning models in network security threat detection through the critical lens of partial versus complete flow information, addressing a common gap between research settings and real-time operational needs. We systematically evaluate how a standard benchmark model, Random Forest, performs under varying training and testing conditions (complete/complete, partial/partial, complete/partial), quantifying the performance impact when dealing with the incomplete data typical in real-time environments. Our findings demonstrate a significant performance difference, with precision and recall dropping by up to 30% under certain conditions when models trained on complete flows are tested against partial flows. The study also reveals that, for the evaluated dataset and model, a minimum threshold around 7 packets in the test set appears necessary for maintaining reliable detection rates, providing valuable, quantified insights for developing more realistic real-time detection strategies.
Authors: Shanghao Shi, Md Shahedul Haque, Abhijeet Parida, Chaoyu Zhang, Marius George Linguraru, Y. Thomas Hou, Syed Muhammad Anwar, Wenjing Lou
Abstract: Federated learning (FL) allows participants to collaboratively train machine learning models while keeping their data local, making it ideal for collaborations among healthcare institutions on sensitive data. However, in this paper, we propose a novel privacy attack called MedLeak, which allows a malicious FL server to recover high-quality site-specific private medical data from the client model updates. MedLeak works by introducing an adversarially crafted model during the FL training process. Honest clients, unaware of the insidious changes in the published models, continue to send back their updates as per the standard FL protocol. Leveraging a novel analytical method, MedLeak can efficiently recover private client data from the aggregated parameter updates, eliminating costly optimization. In addition, the scheme relies solely on the aggregated updates, thus rendering secure aggregation protocols ineffective, as they depend on the randomization of intermediate results for security while leaving the final aggregated results unaltered. We implement MedLeak on medical image datasets (MedMNIST, COVIDx CXR-4, and Kaggle Brain Tumor MRI), as well as a medical text dataset (MedAbstract). The results demonstrate that our attack achieves high recovery rates and strong quantitative scores on both image and text datasets. We also thoroughly evaluate MedLeak across different attack parameters, providing insights into key factors that influence attack performance and potential defenses. Furthermore, we demonstrate that the recovered data can support downstream tasks such as disease classification with minimal performance loss. Our findings validate the need for enhanced privacy measures in FL systems, particularly for safeguarding sensitive medical data against powerful model inversion attacks.
Authors: Yi Ren, Danica J. Sutherland
Abstract: Learning dynamics, which describes how the learning of specific training examples influences the model's predictions on other examples, gives us a powerful tool for understanding the behavior of deep learning systems. We study the learning dynamics of large language models during different types of finetuning, by analyzing the step-wise decomposition of how influence accumulates among different potential responses. Our framework allows a uniform interpretation of many interesting observations about the training of popular algorithms for both instruction tuning and preference tuning. In particular, we propose a hypothetical explanation of why specific types of hallucination are strengthened after finetuning, e.g., the model might use phrases or facts in the response for question B to answer question A, or the model might keep repeating similar simple phrases when generating responses. We also extend our framework and highlight a unique "squeezing effect" to explain a previously observed phenomenon in off-policy direct preference optimization (DPO), where running DPO for too long makes even the desired outputs less likely. This framework also provides insights into where the benefits of on-policy DPO and other variants come from. The analysis not only provides a novel perspective of understanding LLM's finetuning but also inspires a simple, effective method to improve alignment performance.
Authors: Liu Ziyin, Yizhou Xu, Isaac Chuang
Abstract: When symmetry is present in the loss function, the model is likely to be trapped in a low-capacity state that is sometimes known as a "collapse". Being trapped in these low-capacity states can be a major obstacle to training across many scenarios where deep learning technology is applied. We first prove two concrete mechanisms through which symmetries lead to reduced capacities and ignored features during training and inference. We then propose a simple and theoretically justified algorithm, syre, to remove almost all symmetry-induced low-capacity states in neural networks. When this type of entrapment is especially a concern, removing symmetries with the proposed method is shown to correlate well with improved optimization or performance. A remarkable merit of the proposed method is that it is model-agnostic and does not require any knowledge of the symmetry.
Authors: Qitian Wu, David Wipf, Junchi Yan
Abstract: Learning representations for structured data with certain geometries (e.g., observed or unobserved) is a fundamental challenge, wherein message passing neural networks (MPNNs) have become a de facto class of model solutions. In this paper, inspired by physical systems, we propose an energy-constrained diffusion model, which integrates the inductive bias of diffusion on manifolds with layer-wise constraints of energy minimization. We identify that the diffusion operators have a one-to-one correspondence with the energy functions implicitly descended by the diffusion process, and the finite-difference iteration for solving the energy-constrained diffusion system induces the propagation layers of various types of MPNNs operating on observed or latent structures. This leads to a unified mathematical framework for common neural architectures whose computational flows can be cast as message passing (or its special case), including MLPs, GNNs, and Transformers. Building on these insights, we devise a new class of neural message passing models, dubbed diffusion-inspired Transformers (DIFFormer), whose global attention layers are derived from the principled energy-constrained diffusion framework. Across diverse datasets ranging from real-world networks to images, texts, and physical particles, we demonstrate that the new model achieves promising performance in scenarios where the data structures are observed (as a graph), partially observed, or entirely unobserved.
Authors: Xingrui Gu, Chuyi Jiang, Erte Wang, Qiang Cui, Leimin Tian, Lianlong Wu, Siyang Song, Chuang Yu
Abstract: Traditional machine learning methods for movement recognition often struggle with limited model interpretability and a lack of insight into human movement dynamics. This study introduces a novel representation learning framework based on causal inference to address these challenges. Our two-stage approach combines the Peter-Clark (PC) algorithm and Kullback-Leibler (KL) divergence to identify and quantify causal relationships between human joints. By capturing joint interactions, the proposed causal Graph Convolutional Network (GCN) produces interpretable and robust representations. Experimental results on the EmoPain dataset demonstrate that the causal GCN outperforms traditional GCNs in accuracy, F1 score, and recall, particularly in detecting protective behaviors. This work contributes to advancing human motion analysis and lays a foundation for adaptive and intelligent healthcare solutions.
Authors: Xavier F. Cadet, Anastasia Borovykh, Mohammad Malekzadeh, Sara Ahmadi-Abhari, Hamed Haddadi
Abstract: Machine unlearning (MU) aims to remove the influence of particular data points from the learnable parameters of a trained machine learning model. This is a crucial capability in light of data privacy requirements, trustworthiness, and safety in deployed models. MU is particularly challenging for deep neural networks (DNNs), such as convolutional nets or vision transformers, as such DNNs tend to memorize a notable portion of their training dataset. Nevertheless, the community lacks a rigorous and multifaceted study that looks into the success of MU methods for DNNs. In this paper, we investigate 18 state-of-the-art MU methods across various benchmark datasets and models, with each evaluation conducted over 10 different initializations, a comprehensive evaluation involving MU over 100K models. We show that, with the proper hyperparameters, Masked Small Gradients (MSG) and Convolution Transpose (CT), consistently perform better in terms of model accuracy and run-time efficiency across different models, datasets, and initializations, assessed by population-based membership inference attacks (MIA) and per-sample unlearning likelihood ratio attacks (U-LiRA). Furthermore, our benchmark highlights the fact that comparing a MU method only with commonly used baselines, such as Gradient Ascent (GA) or Successive Random Relabeling (SRL), is inadequate, and we need better baselines like Negative Gradient Plus (NG+) with proper hyperparameter selection.
Authors: Sedjro Salomon Hotegni, Sebastian Peitz
Abstract: Deep neural networks (DNNs) are vulnerable to small adversarial perturbations, which are tiny changes to the input data that appear insignificant but cause the model to produce drastically different outputs. Many defense methods require modifying model architectures during evaluation or performing test-time data purification. This not only introduces additional complexity but is often architecture-dependent. We show, however, that robust feature learning during training can significantly enhance DNN robustness. We propose MOREL, a multi-objective approach that aligns natural and adversarial features using cosine similarity and multi-positive contrastive losses to encourage similar features for same-class inputs. Extensive experiments demonstrate that MOREL significantly improves robustness against both white-box and black-box attacks. Our code is available at https://github.com/salomonhotegni/MOREL
Authors: Zelin Zang, Yuhao Wang, Jinlin Wu, Hong Liu, Yue Shen, Zhen Lei, Stan. Z Li
Abstract: Dimensionality reduction (DR) plays a crucial role in various fields, including data engineering and visualization, by simplifying complex datasets while retaining essential information. However, achieving both high DR accuracy and strong explainability remains a fundamental challenge, especially for users dealing with high-dimensional data. Traditional DR methods often face a trade-off between precision and transparency, where optimizing for performance can lead to reduced explainability, and vice versa. This limitation is especially prominent in real-world applications such as image, tabular, and text data analysis, where both accuracy and explainability are critical. To address these challenges, this work introduces the MOE-based Explainable Deep Manifold Transformation (DMT-ME). The proposed approach combines hyperbolic embeddings, which effectively capture complex hierarchical structures, with Mixture of Experts (MOE) models, which dynamically allocate tasks based on input features. DMT-ME enhances DR accuracy by leveraging hyperbolic embeddings to represent the hierarchical nature of data, while also improving explainability by explicitly linking input data, embedding outcomes, and key features through the MOE structure. Extensive experiments demonstrate that DMT-ME consistently achieves superior performance in both DR accuracy and model explainability, making it a robust solution for complex data analysis. The code is available at https://github.com/zangzelin/code_dmtme
Authors: Kaustubh Kislay, Shlok Singh, Soham Joshi, Rohan Dutta, Jay Shim, George Flint, Kevin Zhu
Abstract: Symbolic Regression remains an NP-Hard problem, with extensive research focusing on AI models for this task. Transformer models have shown promise in Symbolic Regression, but performance suffers with smaller datasets. We propose applying k-fold cross-validation to a transformer-based symbolic regression model trained on a significantly reduced dataset (15,000 data points, down from 500,000). This technique partitions the training data into multiple subsets (folds), iteratively training on some while validating on others. Our aim is to provide an estimate of model generalization and mitigate overfitting issues associated with smaller datasets. Results show that this process improves the model's output consistency and generalization by a relative improvement in validation loss of 53.31%. Potentially enabling more efficient and accessible symbolic regression in resource-constrained environments.
Authors: Fang Sun, Zijie Huang, Haixin Wang, Huacong Tang, Xiao Luo, Wei Wang, Yizhou Sun
Abstract: Accurately predicting long-horizon molecular dynamics (MD) trajectories remains a significant challenge, as existing deep learning methods often struggle to retain fidelity over extended simulations. We hypothesize that one key factor limiting accuracy is the difficulty of capturing interactions that span distinct spatial and temporal scales, ranging from high-frequency local vibrations to low-frequency global conformational changes. To address these limitations, we propose Graph Fourier Neural ODEs (GF-NODE), integrating a graph Fourier transform for spatial frequency decomposition with a Neural ODE framework for continuous-time evolution. Specifically, GF-NODE first decomposes molecular configurations into multiple spatial frequency modes using the graph Laplacian, then evolves the frequency components in time via a learnable Neural ODE module that captures both local and global dynamics, and finally reconstructs the updated molecular geometry through an inverse graph Fourier transform. By explicitly modeling high- and low-frequency phenomena in this unified pipeline, GF-NODE captures long-range correlations and local fluctuations more effectively. We provide theoretical insight through heat equation analysis on a simplified diffusion model, demonstrating how graph Laplacian eigenvalues can determine temporal dynamics scales, and crucially validate this correspondence through comprehensive empirical analysis on real molecular dynamics trajectories showing quantitative spatial-temporal correlations across diverse molecular systems. Experimental results on challenging MD benchmarks demonstrate that GF-NODE achieves state-of-the-art accuracy while preserving essential geometrical features over extended simulations. These findings highlight the promise of bridging spectral decomposition with continuous-time modeling to improve the robustness and predictive power of MD simulations.
Authors: Gionni Marchetti
Abstract: A data-driven approach based on unsupervised machine learning is proposed to infer the intrinsic dimension $m^{\ast}$ of the high-dimensional trajectories of the Fermi-Pasta-Ulam-Tsingou (FPUT) model. Principal component analysis (PCA) is applied to trajectory data consisting of $n_s = 4,000,000$ datapoints, of the FPUT $\beta$ model with $N = 32$ coupled oscillators, revealing a critical relationship between $m^{\ast}$ and the model's nonlinear strength. By estimating the intrinsic dimension $m^{\ast}$ using multiple methods (participation ratio, Kaiser rule, and the Kneedle algorithm), it is found that $m^{\ast}$ increases with the model nonlinearity. Interestingly, in the weakly nonlinear regime, for trajectories initialized by exciting the first mode, the participation ratio estimates $m^{\ast} = 2, 3$, strongly suggesting that quasi-periodic motion on a low-dimensional Riemannian manifold underlies the characteristic energy recurrences observed in the FPUT model.
Authors: Yuqi Luo, Chenyang Song, Xu Han, Yingfa Chen, Chaojun Xiao, Xiaojun Meng, Liqun Deng, Jiansheng Wei, Zhiyuan Liu, Maosong Sun
Abstract: Activation sparsity denotes the existence of substantial weakly-contributed elements within activation outputs that can be eliminated, benefiting many important applications concerned with large language models (LLMs). Although promoting greater activation sparsity within LLMs deserves deep studies, existing works lack comprehensive and quantitative research on the correlation between activation sparsity and potentially influential factors. In this paper, we present a comprehensive study on the quantitative scaling properties and influential factors of the activation sparsity within decoder-only Transformer-based LLMs. Specifically, we propose PPL-$p\%$ sparsity, a precise and performance-aware activation sparsity metric that is applicable to any activation function. Through extensive experiments, we find several important phenomena. Firstly, different activation functions exhibit comparable performance but opposite training-time sparsity trends. The activation ratio (i.e., $1-\mathrm{sparsity\ ratio}$) evolves as a convergent increasing power-law and decreasing logspace power-law with the amount of training data for SiLU-activated and ReLU-activated LLMs, respectively. These demonstrate that ReLU is more efficient as the activation function than SiLU and can leverage more training data to improve activation sparsity. Secondly, the activation ratio linearly increases with the width-depth ratio below a certain bottleneck point, indicating the potential advantage of a deeper architecture at a fixed parameter scale. Finally, at similar width-depth ratios, we surprisingly find that the limit value of activation sparsity varies weakly with the parameter scale, i.e., the activation patterns within LLMs are insensitive to the parameter scale. These empirical laws towards LLMs with greater activation sparsity have important implications for making LLMs more efficient and interpretable.
Authors: Franklin Y. Ruan, Aiwei Zhang, Jenny Y. Oh, SouYoung Jin, Nicholas C. Jacobson
Abstract: Pretrained foundation models and transformer architectures have driven the success of large language models (LLMs) and other modern AI breakthroughs. However, similar advancements in health data modeling remain limited due to the need for innovative adaptations. Wearable movement data offers a valuable avenue for exploration, as it's a core feature in nearly all commercial smartwatches, well established in clinical and mental health research, and the sequential nature of the data shares similarities to language. We introduce the Pretrained Actigraphy Transformer (PAT), the first open source foundation model designed for time-series wearable movement data. Leveraging transformer-based architectures and novel techniques, such as patch embeddings, and pretraining on data from 29,307 participants in a national U.S. sample, PAT achieves state-of-the-art performance in several mental health prediction tasks. PAT is also lightweight and easily interpretable, making it a robust tool for mental health research. GitHub: https://github.com/njacobsonlab/Pretrained-Actigraphy-Transformer/
URLs: https://github.com/njacobsonlab/Pretrained-Actigraphy-Transformer/
Authors: Chuan Liu, Huanran Chen, Yichi Zhang, Yinpeng Dong, Jun Zhu
Abstract: Adversarial examples usually exhibit good cross-model transferability, enabling attacks on black-box models with limited information about their architectures and parameters, which are highly threatening in commercial black-box scenarios. Model ensembling is an effective strategy to improve the transferability of adversarial examples by attacking multiple surrogate models. However, since prior studies usually adopt few models in the ensemble, there remains an open question of whether scaling the number of models can further improve black-box attacks. Inspired by the scaling law of large foundation models, we investigate the scaling laws of black-box adversarial attacks in this work. Through theoretical analysis and empirical evaluations, we conclude with clear scaling laws that using more surrogate models enhances adversarial transferability. Comprehensive experiments verify the claims on standard image classifiers, diverse defended models and multimodal large language models using various adversarial attack methods. Specifically, by scaling law, we achieve 90%+ transfer attack success rate on even proprietary models like GPT-4o. Further visualization indicates that there is also a scaling law on the interpretability and semantics of adversarial perturbations.
Authors: Haomin Zhuang, Yihua Zhang, Kehan Guo, Jinghan Jia, Gaowen Liu, Sijia Liu, Xiangliang Zhang
Abstract: Recent advancements in LLMs unlearning have shown remarkable success in removing unwanted data-model influences while preserving the model's utility for legitimate knowledge. Despite these strides, sparse Mixture-of-Experts (MoE) LLMs--a key subset of the LLM family--have remained unexplored in the context of unlearning. As MoE LLMs are celebrated for their exceptional performance, we ask:How can unlearning be performed effectively and efficiently on MoE LLMs? Our pilot study shows that the dynamic routing nature of MoE LLMs introduces unique challenges, leading to excessive forgetting, uncontrolled knowledge erasure and substantial utility drops when existing unlearning methods are applied. To address this, we propose a novel Selected-Expert Unlearning Framework (SEUF). Through expert attribution, unlearning is concentrated on the most actively engaged experts for the specified knowledge. Concurrently, an anchor loss is applied to the router to stabilize the active state of this targeted expert, ensuring focused and controlled unlearning. SEUF is compatible with various standard unlearning algorithms. Extensive experiments demonstrate that SEUF enhances both forget quality up to 5% and model utility by 35% on MoE LLMs across various benchmarks and LLM architectures (compared to standard unlearning algorithms), while only unlearning 0.06% of the model parameters.
Authors: Jean Kossaifi, Nikola Kovachki, Zongyi Li, David Pitt, Miguel Liu-Schiaffini, Valentin Duruisseaux, Robert Joseph George, Boris Bonev, Kamyar Azizzadenesheli, Julius Berner, Anima Anandkumar
Abstract: We present NeuralOperator, an open-source Python library for operator learning. Neural operators generalize neural networks to maps between function spaces instead of finite-dimensional Euclidean spaces. They can be trained and inferenced on input and output functions given at various discretizations, satisfying a discretization convergence properties. Built on top of PyTorch, NeuralOperator provides all the tools for training and deploying neural operator models, as well as developing new ones, in a high-quality, tested, open-source package. It combines cutting-edge models and customizability with a gentle learning curve and simple user interface for newcomers.
Authors: Eivind B{\o}hn, S{\o}lve Eidnes, Kjell Rune Jonassen
Abstract: Wastewater treatment plants are increasingly recognized as promising candidates for machine learning applications, due to their societal importance and high availability of data. However, their varied designs, operational conditions, and influent characteristics hinder straightforward automation. In this study, we use data from a pilot reactor at the Veas treatment facility in Norway to explore how machine learning can be used to optimize biological nitrate ($\mathrm{NO_3^-}$) reduction to molecular nitrogen ($\mathrm{N_2}$) in the biogeochemical process known as \textit{denitrification}. Rather than focusing solely on predictive accuracy, our approach prioritizes understanding the foundational requirements for effective data-driven modelling of wastewater treatment. Specifically, we aim to identify which process parameters are most critical, the necessary data quantity and quality, how to structure data effectively, and what properties are required by the models. We find that nonlinear models perform best on the training and validation data sets, indicating nonlinear relationships to be learned, but linear models transfer better to the unseen test data, which comes later in time. The variable measuring the water temperature has a particularly detrimental effect on the models, owing to a significant change in distributions between training and test data. We therefore conclude that multiple years of data is necessary to learn robust machine learning models. By addressing foundational elements, particularly in the context of the climatic variability faced by northern regions, this work lays the groundwork for a more structured and tailored approach to machine learning for wastewater treatment. We share publicly both the data and code used to produce the results in the paper.
Authors: Freddie Bickford Smith, Jannik Kossen, Eleanor Trollope, Mark van der Wilk, Adam Foster, Tom Rainforth
Abstract: The ideas of aleatoric and epistemic uncertainty are widely used to reason about the probabilistic predictions of machine-learning models. We identify incoherence in existing discussions of these ideas and suggest this stems from the aleatoric-epistemic view being insufficiently expressive to capture all the distinct quantities that researchers are interested in. To address this we present a decision-theoretic perspective that relates rigorous notions of uncertainty, predictive performance and statistical dispersion in data. This serves to support clearer thinking as the field moves forward. Additionally we provide insights into popular information-theoretic quantities, showing they can be poor estimators of what they are often purported to measure, while also explaining how they can still be useful in guiding data acquisition.
Authors: Muhammad Umair Danish, Mathumitha Sureshkumar, Tehara Fonseka, Umeshika Uthayakumar, Vinura Galwaduge
Abstract: This research provides an in-depth evaluation of various machine learning models for energy forecasting, focusing on the unique challenges of seasonal variations in student residential settings. The study assesses the performance of baseline models, such as LSTM and GRU, alongside state-of-the-art forecasting methods, including Autoregressive Feedforward Neural Networks, Transformers, and hybrid approaches. Special attention is given to predicting energy consumption amidst challenges like seasonal patterns, vacations, meteorological changes, and irregular human activities that cause sudden fluctuations in usage. The findings reveal that no single model consistently outperforms others across all seasons, emphasizing the need for season-specific model selection or tailored designs. Notably, the proposed Hyper Network based LSTM and MiniAutoEncXGBoost models exhibit strong adaptability to seasonal variations, effectively capturing abrupt changes in energy consumption during summer months. This study advances the energy forecasting field by emphasizing the critical role of seasonal dynamics and model-specific behavior in achieving accurate predictions.
Authors: Zhong Li, Yuhang Wang, Matthijs van Leeuwen
Abstract: Self-supervised learning (SSL) is an emerging paradigm that exploits supervisory signals generated from the data itself, and many recent studies have leveraged SSL to conduct graph anomaly detection. However, we empirically found that three important factors can substantially impact detection performance across datasets: 1) the specific SSL strategy employed; 2) the tuning of the strategy's hyperparameters; and 3) the allocation of combination weights when using multiple strategies. Most SSL-based graph anomaly detection methods circumvent these issues by arbitrarily or selectively (i.e., guided by label information) choosing SSL strategies, hyperparameter settings, and combination weights. While an arbitrary choice may lead to subpar performance, using label information in an unsupervised setting is label information leakage and leads to severe overestimation of a method's performance. Leakage has been criticized as "one of the top ten data mining mistakes", yet many recent studies on SSL-based graph anomaly detection have been using label information to select hyperparameters. To mitigate this issue, we propose to use an internal evaluation strategy (with theoretical analysis) to select hyperparameters in SSL for unsupervised anomaly detection. We perform extensive experiments using 10 recent SSL-based graph anomaly detection algorithms on various benchmark datasets, demonstrating both the prior issues with hyperparameter selection and the effectiveness of our proposed strategy.
Authors: Hao Guo, Han Wang, Di Zhu, Lun Wu, A. Stewart Fotheringham, Yu Liu
Abstract: Modeling spatial heterogeneity in the data generation process is essential for understanding and predicting geographical phenomena. Despite their prevalence in geospatial tasks, neural network models usually assume spatial stationarity, which could limit their performance in the presence of spatial process heterogeneity. By allowing model parameters to vary over space, several approaches have been proposed to incorporate spatial heterogeneity into neural networks. However, current geographically weighting approaches are ineffective on graph neural networks, yielding no significant improvement in prediction accuracy. We assume the crux lies in the over-fitting risk brought by a large number of local parameters. Accordingly, we propose to model spatial process heterogeneity at the regional level rather than at the individual level, which largely reduces the number of spatially varying parameters. We further develop a heuristic optimization procedure to learn the region partition adaptively in the process of model training. Our proposed spatial-heterogeneity-aware graph convolutional network, named RegionGCN, is applied to the spatial prediction of county-level vote share in the 2016 US presidential election based on socioeconomic attributes. Results show that RegionGCN achieves significant improvement over the basic and geographically weighted GCNs. We also offer an exploratory analysis tool for the spatial variation of non-linear relationships through ensemble learning of regional partitions from RegionGCN. Our work contributes to the practice of Geospatial Artificial Intelligence (GeoAI) in tackling spatial heterogeneity.
Authors: Pallavi Sharma, Jorge-Humberto Urrea-Quintero, Bogdan Bogdan, Adrian-Dumitru Ciotec, Laura Vasilie, Henning Wessels, Matteo Skull
Abstract: This work proposes variational autoencoders (VAEs) to predict a vehicle's jerk signals from torque demand in the context of limited real-world drivetrain datasets. We implement both unconditional and conditional VAEs, trained on experimental data from two variants of a fully electric SUV with differing torque and drivetrain configurations. The VAEs synthesize jerk signals that capture characteristics from multiple drivetrain scenarios by leveraging the learned latent space. A performance comparison with baseline physics-based and hybrid models confirms the effectiveness of the VAEs, without requiring detailed system parametrization. Unconditional VAEs generate realistic jerk signals without prior system knowledge, while conditional VAEs enable the generation of signals tailored to specific torque inputs. This approach reduces the dependence on costly and time-intensive real-world experiments and extensive manual modeling. The results support the integration of generative models such as VAEs into drivetrain simulation pipelines, both for data augmentation and for efficient exploration of complex operational scenarios, with the potential to streamline validation and accelerate vehicle development.
Authors: Mingkuan Feng, Jinyang Wu, Shuai Zhang, Pengpeng Shao, Ruihan Jin, Zhengqi Wen, Jianhua Tao, Feihu Che
Abstract: Large language models (LLMs) have achieved significant progress across various domains, but their increasing scale results in high computational and memory costs. Recent studies have revealed that LLMs exhibit sparsity, providing the potential to reduce model size through pruning techniques. However, existing pruning methods typically follow a prune-then-finetune paradigm. Since the pruned components still contain valuable information, their direct removal often leads to irreversible performance degradation, imposing a substantial computational burden to recover performance during finetuning. In this paper, we propose a novel paradigm that first applies regularization, then prunes, and finally finetunes. Based on this paradigm, we introduce DReSS, a simple and effective Data-driven Regularized Structured Streamlining method for LLMs. By leveraging a small amount of data to regularize the components to be pruned, DReSS explicitly transfers the important information to the remaining parts of the model in advance. Compared to direct pruning, this can reduce the information loss caused by parameter removal, thereby enhancing its language modeling capabilities. Experimental results demonstrate that DReSS significantly outperforms existing pruning methods even under extreme pruning ratios, significantly reducing latency and increasing throughput.
Authors: Kanata Oowada, Hideaki Iiduka
Abstract: We have theoretically analyzed the use of Riemannian stochastic gradient descent (RSGD) and found that using an increasing batch size leads to faster RSGD convergence rate than using a constant batch size not only with a constant learning rate but also with a decaying learning rate, such as cosine annealing decay and polynomial decay. The convergence rate of RSGD improves from $O(\sqrt{T^{-1}+\text{const.}})$ with a constant batch size to $O(T^{-\frac{1}{2}})$ with an increasing batch size, where $T$ denotes the number of iterations. Using principal component analysis and low-rank matrix completion tasks, we investigated, both theoretically and numerically, how increasing batch size affects computational time as measured by stochastic first-order oracle (SFO) complexity. Increasing batch size reduces the SFO complexity of RSGD. Furthermore, our numerical results demonstrated that increasing batch size offers the advantages of both small and large constant batch sizes.
Authors: Haitong Ma, Tianyi Chen, Kai Wang, Na Li, Bo Dai
Abstract: Diffusion policies have achieved superior performance in imitation learning and offline reinforcement learning (RL) due to their rich expressiveness. However, the conventional diffusion training procedure requires samples from target distribution, which is impossible in online RL since we cannot sample from the optimal policy. Backpropagating policy gradient through the diffusion process incurs huge computational costs and instability, thus being expensive and not scalable. To enable efficient training of diffusion policies in online RL, we generalize the conventional denoising score matching by reweighting the loss function. The resulting Reweighted Score Matching (RSM) preserves the optimal solution and low computational cost of denoising score matching, while eliminating the need to sample from the target distribution and allowing learning to optimize value functions. We introduce two tractable reweighted loss functions to solve two commonly used policy optimization problems, policy mirror descent and max-entropy policy, resulting in two practical algorithms named Diffusion Policy Mirror Descent (DPMD) and Soft Diffusion Actor-Critic (SDAC). We conducted comprehensive comparisons on MuJoCo benchmarks. The empirical results show that the proposed algorithms outperform recent diffusion-policy online RLs on most tasks, and the DPMD improves more than 120% over soft actor-critic on Humanoid and Ant.
Authors: Xianglong Yan, Tianao Zhang, Zhiteng Li, Yulun Zhang
Abstract: Large language models (LLMs) have achieved remarkable progress in natural language processing, but their high computational and memory costs hinder deployment on resource-constrained devices. Binarization, which reduces model weights to 1 bit, is a promising solution for efficient inference. However, binarized LLMs still exhibit redundancy that can be further compressed. Semi-structured pruning offers a favorable trade-off between model performance and hardware efficiency, but naively combining it with binarization often leads to severe performance degradation. To address this, we propose Progressive Binarization with Semi-Structured Pruning (PBS$^2$P), a novel post-training compression framework. We propose Stepwise semi-structured Pruning with Binarization Optimization (SPBO) to jointly reduce pruning and binarization error. Additionally, we develop a Coarse-to-Fine Search (CFS) strategy to more effectively select pruning elements. Extensive experiments across multiple LLM families show that PBS$^2$P consistently outperforms state-of-the-art binary post-training quantization methods in both perplexity and downstream accuracy. The code and models will be available at: https://github.com/XIANGLONGYAN/PBS2P.
Authors: Martin Eppert, Satyaki Mukherjee, Debarghya Ghoshdastidar
Abstract: Projection Pursuit is a classic exploratory technique for finding interesting projections of a dataset. We propose a method for recovering projections containing either Imbalanced Clusters or a Bernoulli-Rademacher distribution using a gradient-based technique to optimize the projection index. As sample complexity is a major limiting factor in Projection Pursuit, we analyze our algorithm's sample complexity within a Planted Vector setting where we can observe that Imbalanced Clusters can be recovered more easily than balanced ones. Additionally, we give a generalized result that works for a variety of data distributions and projection indices. We compare these results to computational lower bounds in the Low-Degree-Polynomial Framework. Finally, we experimentally evaluate our method's applicability to real-world data using FashionMNIST and the Human Activity Recognition Dataset, where our algorithm outperforms others when only a few samples are available.
Authors: Yikai Wu, Haoyu Zhao, Sanjeev Arora
Abstract: This position paper argues that the machine learning community should fundamentally rethink how AI-inspired methods are developed and evaluated for combinatorial optimization (CO). We present comprehensive empirical benchmarks comparing various recent AI-inspired GPU-based methods with several classical CPU-based solvers on the Maximum Independent Set (MIS) problem. Strikingly, even on in-distribution random graphs, leading AI-inspired methods are consistently outperformed by the state-of-the-art classical solver KaMIS, and some AI-inspired methods frequently fail to surpass even the simplest degree-based greedy heuristic. To better understand the source of these failures, we introduce a novel analysis, serialization, which reveals that non-backtracking AI methods, such as LTFT (based on GFlowNets), end up reasoning similarly to the simplest degree-based greedy heuristic, and thus worse than KaMIS. Our findings reveal three core issues: (1) Limited benchmarks and evaluation - AI-inspired methods are often tested only on small instances with very limited inference time, which covers up issues with scalability and resource usage; (2) Intrinsic hardness and learning limits - even under ideal, in-distribution conditions, learning-based approaches lag behind classical heuristics, highlighting inherent barriers that receive little attention; and (3) Insufficient use and understanding of classical heuristics - current learning frameworks often neglect to incorporate effective classical techniques. Although we use MIS as a testbed, similar gaps and challenges have been reported in other combinatorial optimization problems, suggesting broader relevance for our recommendations. We propose that future research must address these issues by rigorous benchmarking, deepening understanding of learning limitations, and integrating classical heuristics into AI-inspired methods.
Authors: Jingfeng Wu, Peter Bartlett, Matus Telgarsky, Bin Yu
Abstract: In overparameterized logistic regression, gradient descent (GD) iterates diverge in norm while converging in direction to the maximum $\ell_2$-margin solution -- a phenomenon known as the implicit bias of GD. This work investigates additional regularization effects induced by early stopping in well-specified high-dimensional logistic regression. We first demonstrate that the excess logistic risk vanishes for early-stopped GD but diverges to infinity for GD iterates at convergence. This suggests that early-stopped GD is well-calibrated, whereas asymptotic GD is statistically inconsistent. Second, we show that to attain a small excess zero-one risk, polynomially many samples are sufficient for early-stopped GD, while exponentially many samples are necessary for any interpolating estimator, including asymptotic GD. This separation underscores the statistical benefits of early stopping in the overparameterized regime. Finally, we establish nonasymptotic bounds on the norm and angular differences between early-stopped GD and $\ell_2$-regularized empirical risk minimizer, thereby connecting the implicit regularization of GD with explicit $\ell_2$-regularization.
Authors: Jixiu Zhai, Tianchi Lu, Haitian Zhong, Ziyang Xu, Yuhuan Liu, Shengrui Xu, Jingwan Wang, Dan Huang
Abstract: Accurate identification of bioactive peptides (BPs) and protein post-translational modifications (PTMs) is essential for understanding protein function and advancing therapeutic discovery. However, most computational methods remain limited in their generalizability across diverse peptide functions. Here, we present PDeepPP, a unified deep learning framework that integrates pretrained protein language models with a hybrid transformer-convolutional architecture, enabling robust identification across diverse peptide classes and PTM sites. We curated comprehensive benchmark datasets and implemented strategies to address data imbalance, allowing PDeepPP to systematically extract both global and local sequence features. Through extensive analyses-including dimensionality reduction and comparison studies-PDeepPP demonstrates strong, interpretable peptide representations and achieves state-of-the-art performance in 25 of the 33 biological identification tasks. Notably, PDeepPP attains high accuracy in antimicrobial (0.9726) and phosphorylation site (0.9984) identification, with 99.5% specificity in glycosylation site prediction and substantial reduction in false negatives in antimalarial tasks. By enabling large-scale, accurate peptide analysis, PDeepPP supports biomedical research and the discovery of novel therapeutic targets for disease treatment. All code, datasets, and pretrained models are publicly available via GitHub:https://github.com/fondress/PDeepPP and Hugging Face:https://huggingface.co/fondress/PDeppPP.
URLs: https://github.com/fondress/PDeepPP, https://huggingface.co/fondress/PDeppPP.
Authors: Xinyi Wan, Penghui Qi, Guangxing Huang, Min Lin, Jialin Li
Abstract: Pipeline parallelism (PP) is widely used for training large language models (LLMs), yet its scalability is often constrained by high activation memory consumption as the number of in-flight microbatches grows with the degree of PP. In this paper, we focus on addressing this challenge by leveraging the under-explored memory offload strategy in PP. With empirical study, we discover that in the majority of standard configurations, at least half, and potentially all, of the activations can be offloaded with negligible overhead. In the cases where full overload is not possible, we introduce a novel selective offload strategy that decreases peak activation memory in a better-than-linear manner. Furthermore, we integrate memory offload with other techniques to jointly consider overall throughput and memory limitation. Our experiments proves that the per-device activation memory effectively reduces with the total number of stages, making PP a stronger alternative than TP, offering up to a 19\% acceleration with even lower memory consumption. The implementation is open-sourced at \href{https://github.com/sail-sg/zero-bubble-pipeline-parallelism}{this url}.
URLs: https://github.com/sail-sg/zero-bubble-pipeline-parallelism
Authors: Wei Junhao, Yu Zhe, Sakuma Jun
Abstract: Model merging is a technique that combines multiple finetuned models into a single model without additional training, allowing a free-rider to cheaply inherit specialized capabilities. This study investigates methodologies to suppress unwanted model merging by free-riders. Existing methods such as model watermarking or fingerprinting can only detect merging in hindsight. In contrast, we propose a first proactive defense against model merging. Specifically, our defense method modifies the model parameters so that the model is disrupted if the model is merged with any other model, while its functionality is kept unchanged if not merged with others. Our approach consists of two modules, rearranging MLP parameters and scaling attention heads, which push the model out of the shared basin in parameter space, causing the merging performance with other models to degrade significantly. We conduct extensive experiments on image classification, image generation, and text classification to demonstrate that our defense severely disrupts merging while retaining the functionality of the post-protect model. Moreover, we analyze potential adaptive attacks and further propose a dropout-based pruning to improve our proposal's robustness.
Authors: Venmugil Elango
Abstract: Transformer-based models have emerged as a leading architecture for natural language processing, natural language generation, and image generation tasks. A fundamental element of the transformer architecture is self-attention, which allows the model to capture intricate dependencies within the data. However, the self-attention mechanism also incurs significant computational and memory costs, particularly for long sequences. In this paper, we introduce ATTENTION2D, a novel approach that exploits parallelism along two dimensions - query and key/value - of the self-attention operation. This method enables efficient distribution and parallelization of computations across multiple devices. Our approach facilitates asymptotically faster training and inference phases compared to previous methods, without relying on approximations or incurring additional computational or memory overheads. Furthermore, unlike existing techniques that struggle to scale with an increasing number of processing units, our approach effectively scales with additional processing units. Our experimental results confirm the effectiveness of our method in improving communication efficiency and scalability. Compared to Ring Attention, our approach demonstrated up to a 5x performance boost on a GPT-3-like model using 64 NVIDIA A100 GPUs across 16 nodes, and up to a 9.4x performance boost on 64 NVIDIA H100 GPUs across 64 nodes.
Authors: Sree Bhargavi Balija
Abstract: As artificial intelligence systems increasingly operate in Real-world environments, the integration of multi-modal data sources such as vision, language, and audio presents both unprecedented opportunities and critical challenges for achieving trustworthy intelligence. In this paper, we propose a novel framework that unifies federated learning with explainable multi-modal reasoning to ensure trustworthiness in decentralized, dynamic settings. Our approach, called FedMM-X (Federated Multi-Modal Explainable Intelligence), leverages cross-modal consistency checks, client-level interpretability mechanisms, and dynamic trust calibration to address challenges posed by data heterogeneity, modality imbalance, and out-of-distribution generalization. Through rigorous evaluation across federated multi-modal benchmarks involving vision-language tasks, we demonstrate improved performance in both accuracy and interpretability while reducing vulnerabilities to adversarial and spurious correlations. Further, we introduce a novel trust score aggregation method to quantify global model reliability under dynamic client participation. Our findings pave the way toward developing robust, interpretable, and socially responsible AI systems in Real-world environments.
Authors: Gerardo Flores, Abigail Schiff, Alyssa H. Smith, Julia A Fukuyama, Ashia C. Wilson
Abstract: ML-supported decisions, such as ordering tests or determining preventive custody, often involve binary classification based on probabilistic forecasts. Evaluation frameworks for such forecasts typically consider whether to prioritize independent-decision metrics (e.g., Accuracy) or top-K metrics (e.g., Precision@K), and whether to focus on fixed thresholds or threshold-agnostic measures like AUC-ROC. We highlight that a consequentialist perspective, long advocated by decision theorists, should naturally favor evaluations that support independent decisions using a mixture of thresholds given their prevalence, such as Brier scores and Log loss. However, our empirical analysis reveals a strong preference for top-K metrics or fixed thresholds in evaluations at major conferences like ICML, FAccT, and CHIL. To address this gap, we use this decision-theoretic framework to map evaluation metrics to their optimal use cases, along with a Python package, briertools, to promote the broader adoption of Brier scores. In doing so, we also uncover new theoretical connections, including a reconciliation between the Brier Score and Decision Curve Analysis, which clarifies and responds to a longstanding critique by (Assel, et al. 2017) regarding the clinical utility of proper scoring rules.
Authors: Siqing Song, Chuang Wang, Ruiqi Wang, Yi Yang, Xu-Yao Zhang
Abstract: Quantizing large language models (LLMs) to 1-bit precision significantly reduces computational costs, but existing quantization techniques suffer from noticeable performance degradation when using weight and activation precisions below 4 bits (W4A4). In this paper, we propose a post-training quantization framework with W(1+1)A(1*4) configuration, where weights are quantized to 1 bit with an additional 1 bit for fine-grain grouping and activations are quantized to 1 bit with a 4-fold increase in the number of channels. For weight quantization, we propose utilizing Hessian-aware fine-grained grouping along with an EM-based quantization scheme. For activation quantization, we decompose INT4-quantized activations into a 4 * INT1 format equivalently and simultaneously smooth the scaling factors based on quantization errors, which further reduces the quantization errors in activations. Our method surpasses state-of-the-art (SOTA) LLM quantization baselines on W2A4 across multiple tasks, pushing the boundaries of existing LLM quantization methods toward fully binarized models. Code is available at https://github.com/JimmyCrave/LLM-PTQ-binarization.
Authors: MaryLena Bleile
Abstract: From clinical dosing algorithms to autonomous robots, sequential decision-making systems routinely operate with missing or incomplete data. Classical reinforcement learning theory, which is commonly used to solve sequential decision problems, assumes Markovian observability, which may not hold under partial observability. Causal inference paradigms formalise ignorability of missingness. We show these views can be unified and generalized in order to guarantee Q-learning convergence even when the Markov property fails. To do so, we introduce the concept of \emph{relative ignorability}. Relative ignorability is a graphical-causal criterion which refines the requirements for accurate decision-making based on incomplete data. Theoretical results and simulations both reveal that non-markovian stochastic processes whose missingness is relatively ignorable with respect to causal estimands can still be optimized using standard Reinforcement Learning algorithms. These results expand the theoretical foundations of safe, data-efficient AI to real-world environments where complete information is unattainable.
Authors: Nguyen Kim Hai Bui, Nguyen Duy Chien, P\'eter Kov\'acs, Gerg\H{o} Bogn\'ar
Abstract: Financial prediction is a complex and challenging task of time series analysis and signal processing, expected to model both short-term fluctuations and long-term temporal dependencies. Transformers have remarkable success mostly in natural language processing using attention mechanism, which also influenced the time series community. The ability to capture both short and long-range dependencies helps to understand the financial market and to recognize price patterns, leading to successful applications of Transformers in stock prediction. Although, the previous research predominantly focuses on individual features and singular predictions, that limits the model's ability to understand broader market trends. In reality, within sectors such as finance and technology, companies belonging to the same industry often exhibit correlated stock price movements. In this paper, we develop a novel neural network architecture by integrating Time2Vec with the Encoder of the Transformer model. Based on the study of different markets, we propose a novel correlation feature selection method. Through a comprehensive fine-tuning of multiple hyperparameters, we conduct a comparative analysis of our results against benchmark models. We conclude that our method outperforms other state-of-the-art encoding methods such as positional encoding, and we also conclude that selecting correlation features enhance the accuracy of predicting multiple stock prices.
Authors: Andrea N\'ovoa, Luca Magri
Abstract: We propose an online learning framework for forecasting nonlinear spatio-temporal signals (fields). The method integrates (i) dimensionality reduction, here, a simple proper orthogonal decomposition (POD) projection; (ii) a generalized autoregressive model to forecast reduced dynamics, here, a reservoir computer; (iii) online adaptation to update the reservoir computer (the model), here, ensemble sequential data assimilation. We demonstrate the framework on a wake past a cylinder governed by the Navier-Stokes equations, exploring the assimilation of full flow fields (projected onto POD modes) and sparse sensors. Three scenarios are examined: a na\"ive physical state estimation; a two-fold estimation of physical and reservoir states; and a three-fold estimation that also adjusts the model parameters. The two-fold strategy significantly improves ensemble convergence and reduces reconstruction error compared to the na\"ive approach. The three-fold approach enables robust online training of partially-trained reservoir computers, overcoming limitations of a priori training. By unifying data-driven reduced order modelling with Bayesian data assimilation, this work opens new opportunities for scalable online model learning for nonlinear time series forecasting.
Authors: Maksym Shamrai
Abstract: Concatenating matrices is a common technique for uncovering shared structures in data through singular value decomposition (SVD) and low-rank approximations. The fundamental question arises: How does the singular value spectrum of the concatenated matrix relate to the spectra of its individual components? In the present work, we develop a perturbation technique that extends classical results such as Weyl's inequality to concatenated matrices. We setup analytical bounds that quantify stability of singular values under small perturbations in submatrices. The results demonstrate that if submatrices are close in a norm, dominant singular values of the concatenated matrix remain stable enabling controlled trade-offs between accuracy and compression. These provide a theoretical basis for improved matrix clustering and compression strategies with applications in the numerical linear algebra, signal processing, and data-driven modeling.
Authors: Yaoqi Chen, Jinkai Zhang, Baotong Lu, Qianxi Zhang, Chengruidong Zhang, Jingjia Luo, Di Liu, Huiqiang Jiang, Qi Chen, Jing Liu, Bailu Ding, Xiao Yan, Jiawei Jiang, Chen Chen, Mingxing Zhang, Yuqing Yang, Fan Yang, Mao Yang
Abstract: The growing context lengths of large language models (LLMs) pose significant challenges for efficient inference, primarily due to GPU memory and bandwidth constraints. We present RetroInfer, a novel system that reconceptualizes the key-value (KV) cache as a vector storage system which exploits the inherent attention sparsity to accelerate long-context LLM inference. At its core is the wave index, an Attention-aWare VEctor index that enables efficient and accurate retrieval of critical tokens through techniques such as tripartite attention approximation, accuracy-bounded attention estimation, and segmented clustering. Complementing this is the wave buffer, which coordinates KV cache placement and overlaps computation and data transfer across GPU and CPU to sustain high throughput. Unlike prior sparsity-based methods that struggle with token selection and hardware coordination, RetroInfer delivers robust performance without compromising model accuracy. Experiments on long-context benchmarks show up to 4.5X speedup over full attention within GPU memory limits and up to 10.5X over sparse attention baselines when KV cache is extended to CPU memory, all while preserving full-attention-level accuracy.
Authors: Jingnan Wang, Jie Chao, Shangshang Yang, Kaijun Ren, Kefeng Deng, Xi Chen, Yaxin Liu, Hanqiuzi Wen, Ziniu Xiao, Lifeng Zhang, Xiaodong Wang, Jiping Guan, Baoxiang Pan
Abstract: The planning and operation of renewable energy, especially wind power, depend crucially on accurate, timely, and high-resolution weather information. Coarse-grid global numerical weather forecasts are typically downscaled to meet these requirements, introducing challenges of scale inconsistency, process representation error, computation cost, and entanglement of distinct uncertainty sources from chaoticity, model bias, and large-scale forcing. We address these challenges by learning the climatological distribution of a target wind farm using its high-resolution numerical weather simulations. An optimal combination of this learned high-resolution climatological prior with coarse-grid large scale forecasts yields highly accurate, fine-grained, full-variable, large ensemble of weather pattern forecasts. Using observed meteorological records and wind turbine power outputs as references, the proposed methodology verifies advantageously compared to existing numerical/statistical forecasting-downscaling pipelines, regarding either deterministic/probabilistic skills or economic gains. Moreover, a 100-member, 10-day forecast with spatial resolution of 1 km and output frequency of 15 min takes < 1 hour on a moderate-end GPU, as contrast to $\mathcal{O}(10^3)$ CPU hours for conventional numerical simulation. By drastically reducing computational costs while maintaining accuracy, our method paves the way for more efficient and reliable renewable energy planning and operation.
Authors: Yuanze Hu, Zhaoxin Fan, Xinyu Wang, Gen Li, Ye Qiu, Zhichao Yang, Wenjun Wu, Kejian Wu, Yifan Sun, Xiaotie Deng, Jin Dong
Abstract: Lightweight Vision-Language Models (VLMs) are indispensable for resource-constrained applications. The prevailing approach to aligning vision and language models involves freezing both the vision encoder and the language model while training small connector modules. However, this strategy heavily depends on the intrinsic capabilities of the language model, which can be suboptimal for lightweight models with limited representational capacity. In this work, we investigate this alignment bottleneck through the lens of mutual information, demonstrating that the constrained capacity of the language model inherently limits the Effective Mutual Information (EMI) between multimodal inputs and outputs, thereby compromising alignment quality. To address this challenge, we propose TinyAlign, a novel framework inspired by Retrieval-Augmented Generation, which strategically retrieves relevant context from a memory bank to enrich multimodal inputs and enhance their alignment. Extensive empirical evaluations reveal that TinyAlign significantly reduces training loss, accelerates convergence, and enhances task performance. Remarkably, it allows models to achieve baseline-level performance with only 40\% of the fine-tuning data, highlighting exceptional data efficiency. Our work thus offers a practical pathway for developing more capable lightweight VLMs while introducing a fresh theoretical lens to better understand and address alignment bottlenecks in constrained multimodal systems.
Authors: Ruiquan Huang, Donghao Li, Chengshuai Shi, Cong Shen, Jing Yang
Abstract: This paper investigates a hybrid learning framework for reinforcement learning (RL) in which the agent can leverage both an offline dataset and online interactions to learn the optimal policy. We present a unified algorithm and analysis and show that augmenting confidence-based online RL algorithms with the offline dataset outperforms any pure online or offline algorithm alone and achieves state-of-the-art results under two learning metrics, i.e., sub-optimality gap and online learning regret. Specifically, we show that our algorithm achieves a sub-optimality gap $\tilde{O}(\sqrt{1/(N_0/\mathtt{C}(\pi^*|\rho)+N_1}) )$, where $\mathtt{C}(\pi^*|\rho)$ is a new concentrability coefficient, $N_0$ and $N_1$ are the numbers of offline and online samples, respectively. For regret minimization, we show that it achieves a constant $\tilde{O}( \sqrt{N_1/(N_0/\mathtt{C}(\pi^{-}|\rho)+N_1)} )$ speed-up compared to pure online learning, where $\mathtt{C}(\pi^-|\rho)$ is the concentrability coefficient over all sub-optimal policies. Our results also reveal an interesting separation on the desired coverage properties of the offline dataset for sub-optimality gap minimization and regret minimization. We further validate our theoretical findings in several experiments in special RL models such as linear contextual bandits and Markov decision processes (MDPs).
Authors: Myung Jun Kim, F\'elix Lefebvre, Ga\"etan Brison, Alexandre Perez-Lebel, Ga\"el Varoquaux
Abstract: Table foundation models bring high hopes to data science: pre-trained on tabular data to embark knowledge or priors, they should facilitate downstream tasks on tables. One specific challenge is that of data semantics: numerical entries take their meaning from context, e.g., column name. Pre-trained neural networks that jointly model column names and table entries have recently boosted prediction accuracy. While these models outline the promises of world knowledge to interpret table values, they lack the convenience of popular foundation models in text or vision. Indeed, they must be fine-tuned to bring benefits, come with sizeable computation costs, and cannot easily be reused or combined with other architectures. Here we introduce TARTE, a foundation model that transforms tables to knowledge-enhanced vector representations using the string to capture semantics. Pre-trained on large relational data, TARTE yields representations that facilitate subsequent learning with little additional cost. These representations can be fine-tuned or combined with other learners, giving models that push the state-of-the-art prediction performance and improve the prediction/computation performance trade-off. Specialized to a task or a domain, TARTE gives domain-specific representations that facilitate further learning. Our study demonstrates an effective approach to knowledge pre-training for tabular learning.
Authors: Tianheng Ling, Chao Qian, Lukas Johannes Ha{\ss}ler, Gregor Schiele
Abstract: Transformer-based models have shown strong performance across diverse time-series tasks, but their deployment on resource-constrained devices remains challenging due to high memory and computational demand. While prior work targeting Microcontroller Units (MCUs) has explored hardware-specific optimizations, such approaches are often task-specific and limited to 8-bit fixed-point precision. Field-Programmable Gate Arrays (FPGAs) offer greater flexibility, enabling fine-grained control over data precision and architecture. However, existing FPGA-based deployments of Transformers for time-series analysis typically focus on high-density platforms with manual configuration. This paper presents a unified and fully automated deployment framework for Tiny Transformers on embedded FPGAs. Our framework supports a compact encoder-only Transformer architecture across three representative time-series tasks (forecasting, classification, and anomaly detection). It combines quantization-aware training (down to 4 bits), hardware-aware hyperparameter search using Optuna, and automatic VHDL generation for seamless deployment. We evaluate our framework on six public datasets across two embedded FPGA platforms. Results show that our framework produces integer-only, task-specific Transformer accelerators achieving as low as 0.033 mJ per inference with millisecond latency on AMD Spartan-7, while also providing insights into deployment feasibility on Lattice iCE40. All source code will be released in the GitHub repository (https://github.com/Edwina1030/TinyTransformer4TS).
Authors: C\'edric Goemaere, Gaspard Oliviers, Rafal Bogacz, Thomas Demeester
Abstract: Predictive Coding (PC) offers a biologically plausible alternative to backpropagation for neural network training, yet struggles with deeper architectures. This paper identifies the root cause: an inherent signal decay problem where gradients attenuate exponentially with depth, becoming computationally negligible due to numerical precision constraints. To address this fundamental limitation, we introduce Error Optimization (EO), a novel reparameterization that preserves PC's theoretical properties while eliminating signal decay. By optimizing over prediction errors rather than states, EO enables signals to reach all layers simultaneously and without attenuation, converging orders of magnitude faster than standard PC. Experiments across multiple architectures and datasets demonstrate that EO matches backpropagation's performance even for deeper models where conventional PC struggles. Besides practical improvements, our work provides theoretical insight into PC dynamics and establishes a foundation for scaling biologically-inspired learning to deeper architectures on digital hardware and beyond.
Authors: Yuan Gao, Ruiqi Shu, Hao Wu, Fan Xu, Yanfei Xiang, Ruijian Gou, Qingsong Wen, Xian Wu, Xiaomeng Huang
Abstract: Accurate Subseasonal-to-Seasonal (S2S) ocean simulation is critically important for marine research, yet remains challenging due to its substantial thermal inertia and extended time delay. Machine learning (ML)-based models have demonstrated significant advancements in simulation accuracy and computational efficiency compared to traditional numerical methods. Nevertheless, a significant limitation of current ML models for S2S ocean simulation is their inadequate incorporation of physical consistency and the slow-changing properties of the ocean system. In this work, we propose a neural ocean model (NeuralOM) for S2S ocean simulation with a multi-scale interactive graph neural network to emulate diverse physical phenomena associated with ocean systems effectively. Specifically, we propose a multi-stage framework tailored to model the ocean's slowly changing nature. Additionally, we introduce a multi-scale interactive messaging module to capture complex dynamical behaviors, such as gradient changes and multiplicative coupling relationships inherent in ocean dynamics. Extensive experimental evaluations confirm that our proposed NeuralOM outperforms state-of-the-art models in S2S and extreme event simulation. The codes are available at https://github.com/YuanGao-YG/NeuralOM.
Authors: Tommy Xu, Zhitian Zhang, Xiangyu Sun, Lauren Kelly Zung, Hossein Hajimirsadeghi, Greg Mori
Abstract: Predictive modeling on tabular data is the cornerstone of many real-world applications. Although gradient boosting machines and some recent deep models achieve strong performance on tabular data, they often lack interpretability. On the other hand, large language models (LLMs) have demonstrated powerful capabilities to generate human-like reasoning and explanations, but remain under-performed for tabular data prediction. In this paper, we propose a new approach that leverages reasoning-based LLMs, trained using reinforcement learning, to perform more accurate and explainable predictions on tabular data. Our method introduces custom reward functions that guide the model not only toward better prediction accuracy but also toward human-understandable reasons for its predictions. The proposed method is evaluated on financial benchmark datasets and compared against established LLMs.
Authors: Mert Onur Cakiroglu, Idil Bilge Altun, Mehmet Dalkilic, Elham Buxton, Hasan Kurban
Abstract: Time series forecasting remains a challenging task for foundation models due to temporal heterogeneity, high dimensionality, and the lack of inherent symbolic structure. In this work, we propose DRAGON (Discrete Representation and Augmented Graph encoding Over de BruijN Graphs), a novel encoder that introduces Multivariate de Bruijn Graphs (MdBGs) to bridge the gap between symbolic representations and neural modeling. DRAGON discretizes continuous input sequences and maps them onto a fixed graph structure, enabling dynamic context recovery via graph-based attention. Integrated as an auxiliary module within a dual-branch architecture, DRAGON augments conventional CNN-based encoders with symbolic, structure-aware representations. All code developed for this study is available at: https://github.com/KurbanIntelligenceLab/MultdBG-Time-Series-Library
URLs: https://github.com/KurbanIntelligenceLab/MultdBG-Time-Series-Library
Authors: C. Evans Hedges
Abstract: We provide evidence that orthogonalizing gradients during training improves model calibration without sacrificing accuracy. On CIFAR-10 with 10\% labeled data, $\perp$Grad matches SGD in accuracy but yields consistently improved calibration metrics such as lower test loss, reduced softmax overconfidence, and higher predictive entropy. These benefits persist under input corruption (CIFAR-10C) and extended training, where $\perp$Grad models degrade more gracefully than SGD-trained counterparts. $\perp$Grad is optimizer-agnostic, incurs minimal overhead, and works well with post-hoc calibration techniques like temperature scaling. Theoretically, we prove convergence of a simplified version of $\perp$Grad under mild assumptions and characterize its stationary points in positive homogeneous networks: $\perp$Grad converges to solutions where further loss reduction requires confidence scaling rather than decision boundary improvement. Code for this paper can be found at: https://github.com/evanshedges2/orthograd\_improves\_calibration.
URLs: https://github.com/evanshedges2/orthograd\_improves\_calibration.
Authors: Xiaohua Lu, Liubov Tupikina, Mehwish Alam
Abstract: Real-world knowledge can take various forms, including structured, semi-structured, and unstructured data. Among these, knowledge graphs are a form of structured human knowledge that integrate heterogeneous data sources into structured representations but typically reduce complex n-ary relations to simple triples, thereby losing higher-order relational details. In contrast, hypergraphs naturally represent n-ary relations with hyperedges, which directly connect multiple entities together. Yet hypergraph representation learning often overlooks entity roles in hyperedges, limiting the finegrained semantic modelling. To address these issues, knowledge hypergraphs and hyper-relational knowledge graphs combine the advantages of knowledge graphs and hypergraphs to better capture the complex structures and role-specific semantics of real world knowledge. This survey provides a comprehensive review of methods handling n-ary relational data, covering both knowledge hypergraphs and hyper-relational knowledge graphs literatures. We propose a two-dimensional taxonomy: the first dimension categorises models based on their methodology, i.e., translation-based models, tensor factorisation-based models, deep neural network-based models, logic rules-based models, and hyperedge expansion-based models. The second dimension classifies models according to their awareness of entity roles and positions in n-ary relations, dividing them into aware-less, position-aware, and role-aware approaches. Finally, we discuss existing datasets, training settings and strategies, and outline open challenges to inspire future research.
Authors: Chi-Sheng Chen
Abstract: Recent advancements in quantum machine learning have shown promise in enhancing classical neural network architectures, particularly in domains involving complex, high-dimensional data. Building upon prior work in temporal sequence modeling, this paper introduces Vision-QRWKV, a hybrid quantum-classical extension of the Receptance Weighted Key Value (RWKV) architecture, applied for the first time to image classification tasks. By integrating a variational quantum circuit (VQC) into the channel mixing component of RWKV, our model aims to improve nonlinear feature transformation and enhance the expressive capacity of visual representations. We evaluate both classical and quantum RWKV models on a diverse collection of 14 medical and standard image classification benchmarks, including MedMNIST datasets, MNIST, and FashionMNIST. Our results demonstrate that the quantum-enhanced model outperforms its classical counterpart on a majority of datasets, particularly those with subtle or noisy class distinctions (e.g., ChestMNIST, RetinaMNIST, BloodMNIST). This study represents the first systematic application of quantum-enhanced RWKV in the visual domain, offering insights into the architectural trade-offs and future potential of quantum models for lightweight and efficient vision tasks.
Authors: Yonatan Ashlag, Uri Koren, Mirco Mutti, Esther Derman, Pierre-Luc Bacon, Shie Mannor
Abstract: State entropy regularization has empirically shown better exploration and sample complexity in reinforcement learning (RL). However, its theoretical guarantees have not been studied. In this paper, we show that state entropy regularization improves robustness to structured and spatially correlated perturbations. These types of variation are common in transfer learning but often overlooked by standard robust RL methods, which typically focus on small, uncorrelated changes. We provide a comprehensive characterization of these robustness properties, including formal guarantees under reward and transition uncertainty, as well as settings where the method performs poorly. Much of our analysis contrasts state entropy with the widely used policy entropy regularization, highlighting their different benefits. Finally, from a practical standpoint, we illustrate that compared with policy entropy, the robustness advantages of state entropy are more sensitive to the number of rollouts used for policy evaluation.
Authors: Seokbin Yoon, Keumjin Lee
Abstract: Aircraft trajectory modeling plays a crucial role in Air Traffic Management (ATM) and is important for various downstream tasks, including conflict detection and landing time prediction. Dataset augmentation through the addition of synthetically generated trajectory data is necessary to develop a more robust aircraft trajectory model and ensure that the trajectory dataset is sufficient and balanced. In this work, we propose a novel framework called ATRADA for aircraft trajectory dataset augmentation. In the proposed framework, a Transformer encoder learns the underlying patterns in the original trajectory dataset and converts each data point into a context vector in the learned latent space. The converted dataset in the latent space is projected into reduced dimensions using principal component analysis (PCA), and a Gaussian mixture model (GMM) is applied to fit the probability distribution of the data points in the reduced-dimensional space. Finally, new samples are drawn from the fitted GMM, the dimension of the samples is reverted to the original dimension, and they are decoded with a Multi-Layer Perceptron (MLP). Several experiments demonstrate that the framework effectively generates new, high-quality synthetic aircraft trajectory data, which were compared to the results of several baselines.
Authors: Dongkyu Cho, Rumi Chunara
Abstract: Data augmentation is a promising tool for enhancing out-of-distribution generalization, where the key is to produce diverse, challenging variations of the source domain via costly targeted augmentations that maximize its generalization effect. Conversely, random augmentation is inexpensive but is deemed suboptimal due to its limited effect. In this paper, we revisit random augmentation and explore methods to address its shortcomings. We show that the stochastic nature of random augmentation can produce a set of colliding augmentations that distorts the learned features, similar to catastrophic forgetting. We propose a simple solution that improves the generalization effect of random augmentation by addressing forgetting, which displays strong generalization performance across various single source domain generalization (sDG) benchmarks.
Authors: Sizhe Dang, Yangyang Guo, Yanjun Zhao, Haishan Ye, Xiaodong Zheng, Guang Dai, Ivor Tsang
Abstract: Fine-tuning large language models (LLMs) often faces GPU memory bottlenecks: the backward pass of first-order optimizers like Adam increases memory usage to more than 10 times the inference level (e.g., 633 GB for OPT-30B). Zeroth-order (ZO) optimizers avoid this cost by estimating gradients only from forward passes, yet existing methods like MeZO usually require many more steps to converge. Can this trade-off between speed and memory in ZO be fundamentally improved? Normalized-SGD demonstrates strong empirical performance with greater memory efficiency than Adam. In light of this, we introduce FZOO, a Fast Zeroth-Order Optimizer toward Adam-Scale Speed. FZOO reduces the total forward passes needed for convergence by employing batched one-sided estimates that adapt step sizes based on the standard deviation of batch losses. It also accelerates per-batch computation through the use of Rademacher random vector perturbations coupled with CUDA's parallel processing. Extensive experiments on diverse models, including RoBERTa-large, OPT (350M-66B), Phi-2, and Llama3, across 11 tasks validate FZOO's effectiveness. On average, FZOO outperforms MeZO by 3 percent in accuracy while requiring 3 times fewer forward passes. For RoBERTa-large, FZOO achieves average improvements of 5.6 percent in accuracy and an 18 times reduction in forward passes compared to MeZO, achieving convergence speeds comparable to Adam. We also provide theoretical analysis proving FZOO's formal equivalence to a normalized-SGD update rule and its convergence guarantees. FZOO integrates smoothly into PEFT techniques, enabling even larger memory savings. Overall, our results make single-GPU, high-speed, full-parameter fine-tuning practical and point toward future work on memory-efficient pre-training.
Authors: Yuhui Ding, Thomas Hofmann
Abstract: Equivariant diffusion models have achieved impressive performance in 3D molecule generation. These models incorporate Euclidean symmetries of 3D molecules by utilizing an SE(3)-equivariant denoising network. However, specialized equivariant architectures limit the scalability and efficiency of diffusion models. In this paper, we propose an approach that relaxes such equivariance constraints. Specifically, our approach learns a sample-dependent SO(3) transformation for each molecule to construct an aligned latent space. A non-equivariant diffusion model is then trained over the aligned representations. Experimental results demonstrate that our approach performs significantly better than previously reported non-equivariant models. It yields sample quality comparable to state-of-the-art equivariant diffusion models and offers improved training and sampling efficiency. Our code is available at https://github.com/skeletondyh/RADM
Authors: Luc Raszewski, Christine De Kock
Abstract: Malicious sockpuppet detection on Wikipedia is critical to preserving access to reliable information on the internet and preventing the spread of disinformation. Prior machine learning approaches rely on stylistic and meta-data features, but do not prioritise adaptability to author-specific behaviours. As a result, they struggle to effectively model the behaviour of specific sockpuppet-groups, especially when text data is limited. To address this, we propose the application of meta-learning, a machine learning technique designed to improve performance in data-scarce settings by training models across multiple tasks. Meta-learning optimises a model for rapid adaptation to the writing style of a new sockpuppet-group. Our results show that meta-learning significantly enhances the precision of predictions compared to pre-trained models, marking an advancement in combating sockpuppetry on open editing platforms. We release a new dataset of sockpuppet investigations to foster future research in both sockpuppetry and meta-learning fields.
Authors: Filip Sondej, Yushi Yang, Miko{\l}aj Kniejski, Marcel Windys
Abstract: Language models can retain dangerous knowledge and skills even after extensive safety fine-tuning, posing both misuse and misalignment risks. Recent studies show that even specialized unlearning methods can be easily reversed. To address this, we systematically evaluate many existing and novel components of unlearning methods and identify ones crucial for irreversible unlearning. We introduce Disruption Masking, a technique in which we only allow updating weights, where the signs of the unlearning gradient and the retaining gradient are the same. This ensures all updates are non-disruptive. Additionally, we identify the need for normalizing the unlearning gradients, and also confirm the usefulness of meta-learning. We combine these insights into MUDMAN (Meta-Unlearning with Disruption Masking and Normalization) and validate its effectiveness at preventing the recovery of dangerous capabilities. MUDMAN outperforms the prior TAR method by 40%, setting a new state-of-the-art for robust unlearning.
Authors: Bilal Faye, Hanane Azzag, Mustapha Lebbah
Abstract: Single-trajectory reinforcement learning (RL) methods aim to optimize policies from datasets consisting of (prompt, response, reward) triplets, where scalar rewards are directly available. This supervision format is highly practical, as it mirrors real-world human feedback, such as thumbs-up/down signals, and avoids the need for structured preference annotations. In contrast, pairwise preference-based methods like Direct Preference Optimization (DPO) rely on datasets with both preferred and dispreferred responses, which are harder to construct and less natural to collect. Among single-trajectory approaches, Direct Reward Optimization (DRO) has shown strong empirical performance due to its simplicity and stability. However, DRO requires approximating a value function, which introduces several limitations: high off-policy variance, coupling between policy and value learning, and a lack of absolute supervision on the policy itself. We introduce Reward Partitioning Optimization (RPO), a new method that resolves these limitations by removing the need to model the value function. Instead, RPO normalizes observed rewards using a partitioning approach estimated directly from data. This leads to a straightforward supervised learning objective on the policy, with no auxiliary models and no joint optimization. RPO provides direct and stable supervision on the policy, making it robust and easy to implement in practice. We validate RPO on scalar-feedback language modeling tasks using Flan-T5 encoder-decoder models. Our results demonstrate that RPO outperforms existing single-trajectory baselines such as DRO and Kahneman-Tversky Optimization (KTO). These findings confirm that RPO is a simple, effective, and theoretically grounded method for single-trajectory policy optimization.
Authors: Ben Finkelshtein, \.Ismail \.Ilkan Ceylan, Michael Bronstein, Ron Levie
Abstract: Graph machine learning architectures are typically tailored to specific tasks on specific datasets, which hinders their broader applicability. This has led to a new quest in graph machine learning: how to build graph foundation models capable of generalizing across arbitrary graphs and features? In this work, we present a recipe for designing graph foundation models for node-level tasks from first principles. The key ingredient underpinning our study is a systematic investigation of the symmetries that a graph foundation model must respect. In a nutshell, we argue that label permutation-equivariance alongside feature permutation-invariance are necessary in addition to the common node permutation-equivariance on each local neighborhood of the graph. To this end, we first characterize the space of linear transformations that are equivariant to permutations of nodes and labels, and invariant to permutations of features. We then prove that the resulting network is a universal approximator on multisets that respect the aforementioned symmetries. Our recipe uses such layers on the multiset of features induced by the local neighborhood of the graph to obtain a class of graph foundation models for node property prediction. We validate our approach through extensive experiments on 29 real-world node classification datasets, demonstrating both strong zero-shot empirical performance and consistent improvement as the number of training graphs increases.
Authors: Shen Yuan, Yin Zheng, Taifeng Wang, Binbin Liu, Hongteng Xu
Abstract: Adapting large-scale foundation models in multi-task scenarios often suffers from task conflict and oblivion. To mitigate such issues, we propose a novel ''model MoE-ization'' strategy that leads to a conflict- and oblivion-resistant multi-task adaptation method. Given a weight matrix of a pre-trained model, our method applies SVD to it and introduces a learnable router to adjust its singular values based on tasks and samples. Accordingly, the weight matrix becomes a Mixture of Orthogonal Rank-one Experts (MoORE), in which each expert corresponds to the outer product of a left singular vector and the corresponding right one. We can improve the model capacity by imposing a learnable orthogonal transform on the right singular vectors. Unlike low-rank adaptation (LoRA) and its MoE-driven variants, MoORE guarantees the experts' orthogonality and maintains the column space of the original weight matrix. These two properties make the adapted model resistant to the conflicts among the new tasks and the oblivion of its original tasks, respectively. Experiments on various datasets demonstrate that MoORE outperforms existing multi-task adaptation methods consistently, showing its superiority in terms of conflict- and oblivion-resistance. The code of the experiments is available at https://github.com/DaShenZi721/MoORE.
Authors: Gerardo A. Flores, Alyssa H. Smith, Julia A. Fukuyama, Ashia C. Wilson
Abstract: Machine learning-based decision support systems are increasingly deployed in clinical settings, where probabilistic scoring functions are used to inform and prioritize patient management decisions. However, widely used scoring rules, such as accuracy and AUC-ROC, fail to adequately reflect key clinical priorities, including calibration, robustness to distributional shifts, and sensitivity to asymmetric error costs. In this work, we propose a principled yet practical evaluation framework for selecting calibrated thresholded classifiers that explicitly accounts for the uncertainty in class prevalences and domain-specific cost asymmetries often found in clinical settings. Building on the theory of proper scoring rules, particularly the Schervish representation, we derive an adjusted variant of cross-entropy (log score) that averages cost-weighted performance over clinically relevant ranges of class balance. The resulting evaluation is simple to apply, sensitive to clinical deployment conditions, and designed to prioritize models that are both calibrated and robust to real-world variations.
Authors: Yikai Hou, Peng Tang
Abstract: Modern intelligent transportation systems rely on accurate spatiotemporal traffic analysis to optimize urban mobility and infrastructure resilience. However, pervasive missing data caused by sensor failures and heterogeneous sensing gaps fundamentally hinders reliable traffic modeling. This paper proposes a Neural Canonical Polyadic Factorization (NCPF) model that synergizes low-rank tensor algebra with deep representation learning for robust traffic data imputation. The model innovatively embeds CP decomposition into neural architecture through learnable embedding projections, where sparse traffic tensors are encoded into dense latent factors across road segments, time intervals, and mobility metrics. A hierarchical feature fusion mechanism employs Hadamard products to explicitly model multilinear interactions, while stacked multilayer perceptron layers nonlinearly refine these representations to capture complex spatiotemporal couplings. Extensive evaluations on six urban traffic datasets demonstrate NCPF's superiority over six state-of-the-art baselines. By unifying CP decomposition's interpretable factor analysis with neural network's nonlinear expressive power, NCPF provides a principled yet flexible approaches for high-dimensional traffic data imputation, offering critical support for next-generation transportation digital twins and adaptive traffic control systems.
Authors: Eric V. Strobl
Abstract: Causal inference in longitudinal biomedical data remains a central challenge, especially in psychiatry, where symptom heterogeneity and latent confounding frequently undermine classical estimators. Most existing methods for treatment effect estimation presuppose a fixed outcome variable and address confounding through observed covariate adjustment. However, the assumption of unconfoundedness may not hold for a fixed outcome in practice. To address this foundational limitation, we directly optimize the outcome definition to maximize causal identifiability. Our DEBIAS (Durable Effects with Backdoor-Invariant Aggregated Symptoms) algorithm learns non-negative, clinically interpretable weights for outcome aggregation, maximizing durable treatment effects and empirically minimizing both observed and latent confounding by leveraging the time-limited direct effects of prior treatments in psychiatric longitudinal data. The algorithm also furnishes an empirically verifiable test for outcome unconfoundedness. DEBIAS consistently outperforms state-of-the-art methods in recovering causal effects for clinically interpretable composite outcomes across comprehensive experiments in depression and schizophrenia.
Authors: Zhuo He, Shuang Li, Wenze Song, Longhui Yuan, Jian Liang, Han Li, Kun Gai
Abstract: Endowing deep models with the ability to generalize in dynamic scenarios is of vital significance for real-world deployment, given the continuous and complex changes in data distribution. Recently, evolving domain generalization (EDG) has emerged to address distribution shifts over time, aiming to capture evolving patterns for improved model generalization. However, existing EDG methods may suffer from spurious correlations by modeling only the dependence between data and targets across domains, creating a shortcut between task-irrelevant factors and the target, which hinders generalization. To this end, we design a time-aware structural causal model (SCM) that incorporates dynamic causal factors and the causal mechanism drifts, and propose \textbf{S}tatic-D\textbf{YN}amic \textbf{C}ausal Representation Learning (\textbf{SYNC}), an approach that effectively learns time-aware causal representations. Specifically, it integrates specially designed information-theoretic objectives into a sequential VAE framework which captures evolving patterns, and produces the desired representations by preserving intra-class compactness of causal factors both across and within domains. Moreover, we theoretically show that our method can yield the optimal causal predictor for each time domain. Results on both synthetic and real-world datasets exhibit that SYNC can achieve superior temporal generalization performance.
Authors: Sree Bhargavi Balija, Amitash Nanda, Debashis Sahoo
Abstract: Federated learning has significantly advanced distributed training of machine learning models across decentralized data sources. However, existing frameworks often lack comprehensive solutions that combine uncertainty quantification, interpretability, and robustness. To address this, we propose FedNAM+, a federated learning framework that integrates Neural Additive Models (NAMs) with a novel conformal prediction method to enable interpretable and reliable uncertainty estimation. Our method introduces a dynamic level adjustment technique that utilizes gradient-based sensitivity maps to identify key input features influencing predictions. This facilitates both interpretability and pixel-wise uncertainty estimates. Unlike traditional interpretability methods such as LIME and SHAP, which do not provide confidence intervals, FedNAM+ offers visual insights into prediction reliability. We validate our approach through experiments on CT scan, MNIST, and CIFAR datasets, demonstrating high prediction accuracy with minimal loss (e.g., only 0.1% on MNIST), along with transparent uncertainty measures. Visual analysis highlights variable uncertainty intervals, revealing low-confidence regions where model performance can be improved with additional data. Compared to Monte Carlo Dropout, FedNAM+ delivers efficient and global uncertainty estimates with reduced computational overhead, making it particularly suitable for federated learning scenarios. Overall, FedNAM+ provides a robust, interpretable, and computationally efficient framework that enhances trust and transparency in decentralized predictive modeling.
Authors: Qing Feng, Samuel Daulton, Benjamin Letham, Maximilian Balandat, Eytan Bakshy
Abstract: Online experiments in internet systems, also known as A/B tests, are used for a wide range of system tuning problems, such as optimizing recommender system ranking policies and learning adaptive streaming controllers. Decision-makers generally wish to optimize for long-term treatment effects of the system changes, which often requires running experiments for a long time as short-term measurements can be misleading due to non-stationarity in treatment effects over time. The sequential experimentation strategies--which typically involve several iterations--can be prohibitively long in such cases. We describe a novel approach that combines fast experiments (e.g., biased experiments run only for a few hours or days) and/or offline proxies (e.g., off-policy evaluation) with long-running, slow experiments to perform sequential, Bayesian optimization over large action spaces in a short amount of time.
Authors: Alan N. Amin, Andres Potapczynski, Andrew Gordon Wilson
Abstract: To understand how genetic variants in human genomes manifest in phenotypes -- traits like height or diseases like asthma -- geneticists have sequenced and measured hundreds of thousands of individuals. Geneticists use this data to build models that predict how a genetic variant impacts phenotype given genomic features of the variant, like DNA accessibility or the presence of nearby DNA-bound proteins. As more data and features become available, one might expect predictive models to improve. Unfortunately, training these models is bottlenecked by the need to solve expensive linear algebra problems because variants in the genome are correlated with nearby variants, requiring inversion of large matrices. Previous methods have therefore been restricted to fitting small models, and fitting simplified summary statistics, rather than the full likelihood of the statistical model. In this paper, we leverage modern fast linear algebra techniques to develop DeepWAS (Deep genome Wide Association Studies), a method to train large and flexible neural network predictive models to optimize likelihood. Notably, we find that larger models only improve performance when using our full likelihood approach; when trained by fitting traditional summary statistics, larger models perform no better than small ones. We find larger models trained on more features make better predictions, potentially improving disease predictions and therapeutic target identification.
Authors: Rylan Schaeffer, Joshua Kazdan, Yegor Denisov-Blanch, Brando Miranda, Matthias Gerstgrasser, Susan Zhang, Andreas Haupt, Isha Gupta, Elyas Obbad, Jesse Dodge, Jessica Zosa Forde, Koustuv Sinha, Francesco Orabona, Sanmi Koyejo, David Donoho
Abstract: Science progresses by iteratively advancing and correcting humanity's understanding of the world. In machine learning (ML) research, rapid advancements have led to an explosion of publications, but have also led to misleading, incorrect, flawed or perhaps even fraudulent studies being accepted and sometimes highlighted at ML conferences due to the fallibility of peer review. While such mistakes are understandable, ML conferences do not offer robust processes to help the field systematically correct when such errors are made. This position paper argues that ML conferences should establish a dedicated "Refutations and Critiques" (R&C) Track. This R&C Track would provide a high-profile, reputable platform to support vital research that critically challenges prior research, thereby fostering a dynamic self-correcting research ecosystem. We discuss key considerations including track design, review principles, potential pitfalls, and provide an illustrative example submission concerning a recent ICLR 2025 Oral. We conclude that ML conferences should create official, reputable mechanisms to help ML research self-correct.
Authors: Jaris K\"uken, Lennart Purucker, Frank Hutter
Abstract: Tabular foundation models have shown strong performance across various tabular learning tasks via in-context learning, offering robust generalization without any downstream finetuning. However, their inference-time costs remain high, particularly for larger datasets. To address this, we propose early-stopping the in-context learning process. We achieve this by dynamically evaluating whether to stop in-context learning after each Transformer encoder layer. Once stopped, we decode the embedding using a pre-trained layer-wise decoder. Experiments across 34 small classification tasks size show that early stopping in-context learning accelerates inference by up to x1.3 with negligible degradation in predictive performance. To assess scalability, we further evaluate our method on five larger classification tasks, achieving speedups of up to x2.2. Our results demonstrate the potential of early exiting as an effective and practical strategy for improving the efficiency of tabular in-context learning.
Authors: Chen Wang, Lai Wei, Yanzhi Zhang, Chenyang Shao, Zedong Dan, Weiran Huang, Yue Wang, Yuzhi Zhang
Abstract: Recent advances in reinforcement learning (RL) have significantly enhanced the reasoning capabilities of large language models (LLMs). Group Relative Policy Optimization (GRPO), an efficient variant of PPO that lowers RL's computational cost, still faces limited exploration, low sample efficiency and instability, constraining its performance on complex reasoning tasks. To address these limitations, we introduce EFRame, an Exploration-Filtering-Replay framework that systematically augments GRPO along three critical dimensions. EFRame performs additional rollouts to explore high-quality trajectories, applies online filtering to eliminate low-quality samples that introduce noise and variance, and leverages experience replay to repeatedly exploit rare but informative samples. EFRame establishes a complete and stable learning cycle, guiding the model through a structured transition from exploration to convergence. Our experiments across a variety of reasoning benchmarks demonstrate that EFRame not only improves the robustness and efficiency of training, but also enables access to deeper reasoning capabilities that remain unattainable under vanilla GRPO. Furthermore, EFRame enables a more fine-grained categorization of training samples, allowing for a deeper analysis of how different types of samples contribute to the learning process in RL. Our code is available at https://github.com/597358816/EFRame.
Authors: Shengling Shi, Anastasios Tsiamis, Bart De Schutter
Abstract: This work analyzes how the trade-off between the modeling error, the terminal value function error, and the prediction horizon affects the performance of a nominal receding-horizon linear quadratic (LQ) controller. By developing a novel perturbation result of the Riccati difference equation, a novel performance upper bound is obtained and suggests that for many cases, the prediction horizon can be either one or infinity to improve the control performance, depending on the relative difference between the modeling error and the terminal value function error. The result also shows that when an infinite horizon is desired, a finite prediction horizon that is larger than the controllability index can be sufficient for achieving a near-optimal performance, revealing a close relation between the prediction horizon and controllability. The obtained suboptimality performance upper bound is applied to provide novel sample complexity and regret guarantees for nominal receding-horizon LQ controllers in a learning-based setting. We show that an adaptive prediction horizon that increases as a logarithmic function of time is beneficial for regret minimization.
Authors: Kaito Fujii
Abstract: This paper investigates equilibrium computation and the price of anarchy for Bayesian games, which are the fundamental models of games with incomplete information. In normal-form games with complete information, it is known that efficiently computable no-regret dynamics converge to correlated equilibria, and the price of anarchy for correlated equilibria can be bounded for a broad class of games called smooth games. However, in Bayesian games, as surveyed by Forges (1993), several non-equivalent extensions of correlated equilibria exist, and it remains unclear whether they can be efficiently computed or whether their price of anarchy can be bounded. In this paper, we identify a natural extension of correlated equilibria that can be computed efficiently and is guaranteed to have bounds on the price of anarchy in various games. First, we propose a variant of regret called untruthful swap regret. If each player minimizes it in repeated play of Bayesian games, the empirical distribution of these dynamics is guaranteed to converge to communication equilibria, which is one of the extensions of correlated equilibria proposed by Myerson (1982). We present an efficient algorithm for minimizing untruthful swap regret with a sublinear upper bound, which we prove to be tight in terms of the number of types. As a result, by simulating the dynamics with our algorithm, we can approximately compute a communication equilibrium in polynomial time. Furthermore, we extend existing lower bounds on the price of anarchy based on the smoothness arguments from Bayes--Nash equilibria to equilibria obtained by the proposed dynamics.
Authors: Abhinav Sagar, Sai Teja Gilukara
Abstract: Sampling-based path planning algorithms play an important role in autonomous robotics. However, a common problem among the RRT-based algorithms is that the initial path generated is not optimal, and the convergence is too slow for real-world applications. In this paper, we propose a novel image-based learning algorithm using a Convolutional Block Attention Generative Adversarial Network (CBAGAN-RRT) with a combination of spatial and channel attention and a novel loss function to design the heuristics, find a better optimal path, and improve the convergence of the algorithm, both concerning time and speed. The probability distribution of the paths generated from our GAN model is used to guide the sampling process for the RRT algorithm. We demonstrate that our algorithm outperforms the previous state-of-the-art algorithms using both the image quality generation metrics, like IOU Score, Dice Score, FID score, and path planning metrics like time cost and the number of nodes. Ablation studies show the effectiveness of various components in our network architecture. The advantage of our approach is that we can avoid the complicated preprocessing in the state space, our model can be generalized to complex environments like those containing turns and narrow passages without loss of accuracy, and our model can be easily integrated with other sampling-based path planning algorithms.
Authors: Niklas Muennighoff, Alexander M. Rush, Boaz Barak, Teven Le Scao, Aleksandra Piktus, Nouamane Tazi, Sampo Pyysalo, Thomas Wolf, Colin Raffel
Abstract: The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations.
Authors: Skyler Ruiter, Seth Wolfgang, Marc Tunnell, Timothy Triche Jr., Erin Carrier, Zachary DeBruine
Abstract: Compressed Sparse Column (CSC) and Coordinate (COO) are popular compression formats for sparse matrices. However, both CSC and COO are general purpose and cannot take advantage of any of the properties of the data other than sparsity, such as data redundancy. Highly redundant sparse data is common in many machine learning applications, such as genomics, and is often too large for in-core computation using conventional sparse storage formats. In this paper, we present two extensions to CSC: (1) Value-Compressed Sparse Column (VCSC) and (2) Index- and Value-Compressed Sparse Column (IVCSC). VCSC takes advantage of high redundancy within a column to further compress data up to 3-fold over COO and 2.25-fold over CSC, without significant negative impact to performance characteristics. IVCSC extends VCSC by compressing index arrays through delta encoding and byte-packing, achieving a 10-fold decrease in memory usage over COO and 7.5-fold decrease over CSC. Our benchmarks on simulated and real data show that VCSC and IVCSC can be read in compressed form with little added computational cost. These two novel compression formats offer a broadly useful solution to encoding and reading redundant sparse data.
Authors: Jiaqi Lin, Sen Lu, Malyaban Bal, Abhronil Sengupta
Abstract: Spiking Neural Networks (SNNs), providing more realistic neuronal dynamics, have been shown to achieve performance comparable to Artificial Neural Networks (ANNs) in several machine learning tasks. Information is processed as spikes within SNNs in an event-based mechanism that significantly reduces energy consumption. However, training SNNs is challenging due to the non-differentiable nature of the spiking mechanism. Traditional approaches, such as Backpropagation Through Time (BPTT), have shown effectiveness but come with additional computational and memory costs and are biologically implausible. In contrast, recent works propose alternative learning methods with varying degrees of locality, demonstrating success in classification tasks. In this work, we show that these methods share similarities during the training process, while they present a trade-off between biological plausibility and performance. Further, given the implicitly recurrent nature of SNNs, this research investigates the influence of the addition of explicit recurrence to SNNs. We experimentally prove that the addition of explicit recurrent weights enhances the robustness of SNNs. We also investigate the performance of local learning methods under gradient and non-gradient-based adversarial attacks.
Authors: Yi Xin, Jianjiang Yang, Siqi Luo, Yuntao Du, Qi Qin, Kangrui Cen, Yangfan He, Bin Fu, Xiaokang Yang, Guangtao Zhai, Ming-Hsuan Yang, Xiaohong Liu
Abstract: Pre-trained vision models (PVMs) have demonstrated remarkable adaptability across a wide range of downstream vision tasks, showcasing exceptional performance. However, as these models scale to billions or even trillions of parameters, conventional full fine-tuning has become increasingly impractical due to its high computational and storage demands. To address these challenges, parameter-efficient fine-tuning (PEFT) has emerged as a promising alternative, aiming to achieve performance comparable to full fine-tuning while making minimal adjustments to the model parameters. This paper presents a comprehensive survey of the latest advancements in the visual PEFT field, systematically reviewing current methodologies and categorizing them into four primary categories: addition-based, partial-based, unified-based, and multi-task tuning. In addition, this paper offers an in-depth analysis of widely used visual datasets and real-world applications where PEFT methods have been successfully applied. Furthermore, this paper introduces the V-PEFT Bench, a unified benchmark designed to standardize the evaluation of PEFT methods across a diverse set of vision tasks, ensuring consistency and fairness in comparison. Finally, the paper outlines potential directions for future research to propel advances in the PEFT field. A comprehensive collection of resources is available at https://github.com/synbol/Awesome-Parameter-Efficient-Transfer-Learning.
URLs: https://github.com/synbol/Awesome-Parameter-Efficient-Transfer-Learning.
Authors: Kexin Chen, Yuyang Du, Junyou Li, Hanqun Cao, Menghao Guo, Xilin Dang, Lanqing Li, Jiezhong Qiu, Pheng Ann Heng, Guangyong Chen
Abstract: The development of AI-assisted chemical synthesis tools requires comprehensive datasets covering diverse reaction types, yet current high-throughput experimental (HTE) approaches are expensive and limited in scope. Chemical literature represents a vast, underexplored data source containing thousands of reactions published annually. However, extracting reaction information from literature faces significant challenges including varied writing styles, complex coreference relationships, and multimodal information presentation. This paper proposes ChemMiner, a novel end-to-end framework leveraging multiple agents powered by large language models (LLMs) to extract high-fidelity chemical data from literature. ChemMiner incorporates three specialized agents: a text analysis agent for coreference mapping, a multimodal agent for non-textual information extraction, and a synthesis analysis agent for data generation. Furthermore, we developed a comprehensive benchmark with expert-annotated chemical literature to evaluate both extraction efficiency and precision. Experimental results demonstrate reaction identification rates comparable to human chemists while significantly reducing processing time, with high accuracy, recall, and F1 scores. Our open-sourced benchmark facilitates future research in chemical literature data mining.
Authors: Chuyu Zhang, Hui Ren, Xuming He
Abstract: Deep clustering, which learns representation and semantic clustering without labels information, poses a great challenge for deep learning-based approaches. Despite significant progress in recent years, most existing methods focus on uniformly distributed datasets, significantly limiting the practical applicability of their methods. In this paper, we propose a more practical problem setting named deep imbalanced clustering, where the underlying classes exhibit an imbalance distribution. To address this challenge, we introduce a novel optimal transport-based pseudo-label learning framework. Our framework formulates pseudo-label generation as a Semantic-regularized Progressive Partial Optimal Transport (SP$^2$OT) problem, which progressively transports each sample to imbalanced clusters under prior and semantic relation constraints, thus generating high-quality and imbalance-aware pseudo-labels. To solve the SP$^2$OT problem, we propose a projected mirror descent algorithm, which alternates between: (1) computing the gradient of the SP$^2$OT objective, and (2) performing gradient descent with projection via an entropy-regularized progressive partial optimal transport formulation. Furthermore, we formulate the second step as an unbalanced optimal transport problem with augmented constraints and develop an efficient solution based on fast matrix scaling algorithms. Experiments on various datasets, including a human-curated long-tailed CIFAR100, challenging ImageNet-R, and large-scale subsets of fine-grained iNaturalist2018 datasets, demonstrate the superiority of our method. Code is available: https://github.com/rhfeiyang/SPPOT
Authors: Zhiwei Hu, V\'ictor Guti\'errez-Basulto, Zhiliang Xiang, Ru Li, Jeff Z. Pan
Abstract: In a hyper-relational knowledge graph (HKG), each fact is composed of a main triple associated with attribute-value qualifiers, which express additional factual knowledge. The hyper-relational knowledge graph completion (HKGC) task aims at inferring plausible missing links in a HKG. Most existing approaches to HKGC focus on enhancing the communication between qualifier pairs and main triples, while overlooking two important properties that emerge from the monotonicity of the hyper-relational graphs representation regime. Stage Reasoning allows for a two-step reasoning process, facilitating the integration of coarse-grained inference results derived solely from main triples and fine-grained inference results obtained from hyper-relational facts with qualifiers. In the initial stage, coarse-grained results provide an upper bound for correct predictions, which are subsequently refined in the fine-grained step. More generally, Qualifier Monotonicity implies that by attaching more qualifier pairs to a main triple, we may only narrow down the answer set, but never enlarge it. This paper proposes the HyperMono model for hyper-relational knowledge graph completion, which realizes stage reasoning and qualifier monotonicity. To implement qualifier monotonicity HyperMono resorts to cone embeddings. Experiments on three real-world datasets with three different scenario conditions demonstrate the strong performance of HyperMono when compared to the SoTA.
Authors: Vaisakh Shaj
Abstract: Machines that can replicate human intelligence with type 2 reasoning capabilities should be able to reason at multiple levels of spatio-temporal abstractions and scales using internal world models. Devising formalisms to develop such internal world models, which accurately reflect the causal hierarchies inherent in the dynamics of the real world, is a critical research challenge in the domains of artificial intelligence and machine learning. This thesis identifies several limitations with the prevalent use of state space models (SSMs) as internal world models and propose two new probabilistic formalisms namely Hidden-Parameter SSMs and Multi-Time Scale SSMs to address these drawbacks. The structure of graphical models in both formalisms facilitates scalable exact probabilistic inference using belief propagation, as well as end-to-end learning via backpropagation through time. This approach permits the development of scalable, adaptive hierarchical world models capable of representing nonstationary dynamics across multiple temporal abstractions and scales. Moreover, these probabilistic formalisms integrate the concept of uncertainty in world states, thus improving the system's capacity to emulate the stochastic nature of the real world and quantify the confidence in its predictions. The thesis also discuss how these formalisms are in line with related neuroscience literature on Bayesian brain hypothesis and predicitive processing. Our experiments on various real and simulated robots demonstrate that our formalisms can match and in many cases exceed the performance of contemporary transformer variants in making long-range future predictions. We conclude the thesis by reflecting on the limitations of our current models and suggesting directions for future research.
Authors: Alexander Atanasov, Jacob A. Zavatone-Veth, Cengiz Pehlevan
Abstract: From benign overfitting in overparameterized models to rich power-law scalings in performance, simple ridge regression displays surprising behaviors sometimes thought to be limited to deep neural networks. This balance of phenomenological richness with analytical tractability makes ridge regression the model system of choice in high-dimensional machine learning. In this paper, we present a unifying perspective on recent results on ridge regression using the basic tools of random matrix theory and free probability, aimed at readers with backgrounds in physics and deep learning. We highlight the fact that statistical fluctuations in empirical covariance matrices can be absorbed into a renormalization of the ridge parameter. This `deterministic equivalence' allows us to obtain analytic formulas for the training and generalization errors in a few lines of algebra by leveraging the properties of the $S$-transform of free probability. From these precise asymptotics, we can easily identify sources of power-law scaling in model performance. In all models, the $S$-transform corresponds to the train-test generalization gap, and yields an analogue of the generalized-cross-validation estimator. Using these techniques, we derive fine-grained bias-variance decompositions for a very general class of random feature models with structured covariates. This allows us to discover a scaling regime for random feature models where the variance due to the features limits performance in the overparameterized setting. We also demonstrate how anisotropic weight structure in random feature models can limit performance and lead to nontrivial exponents for finite-width corrections in the overparameterized setting. Our results extend and provide a unifying perspective on earlier models of neural scaling laws.
Authors: Maja Pavlovic, Massimo Poesio
Abstract: Large Language Models (LLMs) have emerged as powerful support tools across various natural language tasks and a range of application domains. Recent studies focus on exploring their capabilities for data annotation. This paper provides a comparative overview of twelve studies investigating the potential of LLMs in labelling data. While the models demonstrate promising cost and time-saving benefits, there exist considerable limitations, such as representativeness, bias, sensitivity to prompt variations and English language preference. Leveraging insights from these studies, our empirical analysis further examines the alignment between human and GPT-generated opinion distributions across four subjective datasets. In contrast to the studies examining representation, our methodology directly obtains the opinion distribution from GPT. Our analysis thereby supports the minority of studies that are considering diverse perspectives when evaluating data annotation tasks and highlights the need for further research in this direction.
Authors: Qing Feng, Tianyi Ma, Ruihao Zhu
Abstract: Motivated by the concept of satisficing in decision-making, we consider the problem of satisficing regret minimization in bandit optimization. In this setting, the learner aims at selecting satisficing arms (arms with mean reward exceeding a certain threshold value) as frequently as possible. The performance is measured by satisficing regret, which is the cumulative deficit of the chosen arm's mean reward compared to the threshold. We propose SELECT, a general algorithmic template for Satisficing REgret Minimization via SampLing and LowEr Confidence bound Testing, that attains constant expected satisficing regret for a wide variety of bandit optimization problems in the realizable case (i.e., a satisficing arm exists). As a complement, SELECT also enjoys the same (standard) regret guarantee as the oracle in the non-realizable case. To further ensure stability of the algorithm, we introduce SELECT-LITE that achieves a light-tailed satisficing regret distribution plus a constant expected satisficing regret in the realizable case and a sub-linear expected (standard) regret in the non-realizable case. Notably, SELECT-LITE can operate on learning oracles with heavy-tailed (standard) regret distribution. More importantly, our results reveal the surprising compatibility between constant expected satisficing regret and light-tailed satisficing regret distribution, which is in sharp contrast to the case of (standard) regret. Finally, we conduct numerical experiments to validate the performance of SELECT and SELECT-LITE on both synthetic datasets and a real-world dynamic pricing case study.
Authors: Harsh Rangwani, Aishwarya Agarwal, Kuldeep Kulkarni, R. Venkatesh Babu, Srikrishna Karanam
Abstract: Image composition and generation are processes where the artists need control over various parts of the generated images. However, the current state-of-the-art generation models, like Stable Diffusion, cannot handle fine-grained part-level attributes in the text prompts. Specifically, when additional attribute details are added to the base text prompt, these text-to-image models either generate an image vastly different from the image generated from the base prompt or ignore the attribute details. To mitigate these issues, we introduce PartComposer, a training-free method that enables image generation based on fine-grained part-level attributes specified for objects in the base text prompt. This allows more control for artists and enables novel object compositions by combining distinctive object parts. PartComposer first localizes object parts by denoising the object region from a specific diffusion process. This enables each part token to be localized to the right region. After obtaining part masks, we run a localized diffusion process in each part region based on fine-grained part attributes and combine them to produce the final image. All stages of PartComposer are based on repurposing a pre-trained diffusion model, which enables it to generalize across domains. We demonstrate the effectiveness of part-level control provided by PartComposer through qualitative visual examples and quantitative comparisons with contemporary baselines.
Authors: Tuan-Luc Huynh, Thuy-Trang Vu, Weiqing Wang, Yinwei Wei, Trung Le, Dragan Gasevic, Yuan-Fang Li, Thanh-Toan Do
Abstract: Differentiable Search Index (DSI) utilizes pre-trained language models to perform indexing and document retrieval via end-to-end learning without relying on external indexes. However, DSI requires full re-training to index new documents, causing significant computational inefficiencies. Continual learning (CL) offers a solution by enabling the model to incrementally update without full re-training. Existing CL solutions in document retrieval rely on memory buffers or generative models for rehearsal, which is infeasible when accessing previous training data is restricted due to privacy concerns. To this end, we introduce PromptDSI, a prompt-based, rehearsal-free continual learning approach for document retrieval. PromptDSI follows the Prompt-based Continual Learning (PCL) framework, using learnable prompts to efficiently index new documents without accessing previous documents or queries. To improve retrieval latency, we remove the initial forward pass of PCL, which otherwise greatly increases training and inference time, with a negligible trade-off in performance. Additionally, we introduce a novel topic-aware prompt pool that employs neural topic embeddings as fixed keys, eliminating the instability of prompt key optimization while maintaining competitive performance with existing PCL prompt pools. In a challenging rehearsal-free continual learning setup, we demonstrate that PromptDSI variants outperform rehearsal-based baselines, match the strong cache-based baseline in mitigating forgetting, and significantly improving retrieval performance on new corpora.
Authors: Roman Vashurin, Ekaterina Fadeeva, Artem Vazhentsev, Lyudmila Rvanova, Akim Tsvigun, Daniil Vasilev, Rui Xing, Abdelrahman Boda Sadallah, Kirill Grishchenkov, Sergey Petrakov, Alexander Panchenko, Timothy Baldwin, Preslav Nakov, Maxim Panov, Artem Shelmanov
Abstract: The rapid proliferation of large language models (LLMs) has stimulated researchers to seek effective and efficient approaches to deal with LLM hallucinations and low-quality outputs. Uncertainty quantification (UQ) is a key element of machine learning applications in dealing with such challenges. However, research to date on UQ for LLMs has been fragmented in terms of techniques and evaluation methodologies. In this work, we address this issue by introducing a novel benchmark that implements a collection of state-of-the-art UQ baselines and offers an environment for controllable and consistent evaluation of novel UQ techniques over various text generation tasks. Our benchmark also supports the assessment of confidence normalization methods in terms of their ability to provide interpretable scores. Using our benchmark, we conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches. Code: https://github.com/IINemo/lm-polygraph Benchmark: https://huggingface.co/LM-Polygraph
URLs: https://github.com/IINemo/lm-polygraph, https://huggingface.co/LM-Polygraph
Authors: Zhiqiang Cai, Anastassia Doktorova, Robert D. Falgout, C\'esar Herrera
Abstract: This paper studies the shallow Ritz method for solving one-dimensional diffusion-reaction problems. The method is capable of improving the order of approximation for non-smooth problems. By following a similar approach to the one presented in [9], we present a damped block Newton (dBN) method to achieve nearly optimal order of approximation. The dBN method optimizes the Ritz functional by alternating between the linear and non-linear parameters of the shallow ReLU neural network (NN). For diffusion-reaction problems, new difficulties arise: (1) for the linear parameters, the mass matrix is dense and even more ill-conditioned than the stiffness matrix, and (2) for the non-linear parameters, the Hessian matrix is dense and may be singular. This paper addresses these challenges, resulting in a dBN method with computational cost of ${\cal O}(n)$. The ideas presented for diffusion-reaction problems can also be applied to least-squares approximation problems. For both applications, starting with the non-linear parameters as a uniform partition, numerical experiments show that the dBN method moves the mesh points to nearly optimal locations.
Authors: Yunpeng Jiang, Yutong Ban, Paul Weng
Abstract: Data augmentation is widely applied and has shown its benefits in different machine learning tasks. However, as recently observed, it may have an unfair effect in multi-class classification. While data augmentation generally improves the overall performance (and therefore is beneficial for many classes), it can actually be detrimental for other classes, which can be problematic in some application domains. In this paper, to counteract this phenomenon, we propose CLAM, a CLAss-dependent Multiplicative-weights method. To derive it, we first formulate the training of a classifier as a non-linear optimization problem that aims at simultaneously maximizing the individual class performances and balancing them. By rewriting this optimization problem as an adversarial two-player game, we propose a novel multiplicative weight algorithm, for which we prove the convergence. Interestingly, our formulation also reveals that the class-dependent effects of data augmentation is not due to data augmentation only, but is in fact a general phenomenon. Our empirical results over six datasets demonstrate that the performance of learned classifiers is indeed more fairly distributed over classes, with only limited impact on the average accuracy.
Authors: Benjamin Reichman, Adar Avsian, Kartik Talamadupula, Toshish Jawale, Larry Heck
Abstract: Queries to large language models (LLMs) can be divided into two parts: the instruction/question and the accompanying context. The context for retrieval-augmented generation (RAG) systems in most benchmarks comes from Wikipedia-like texts written in a neutral and factual tone. However, real-world RAG applications often retrieve internet-based text with diverse tones and linguistic styles, posing challenges for downstream tasks. This paper introduces (a) a dataset that transforms RAG-retrieved passages into emotionally inflected and sarcastic text, (b) an emotion translation model for adapting text to different tones, and (c) a prompt-based method to improve LLMs' pragmatic interpretation of retrieved text.
Authors: Joseph M. Cavanagh, Kunyang Sun, Andrew Gritsevskiy, Dorian Bagni, Yingze Wang, Thomas D. Bannister, Teresa Head-Gordon
Abstract: Here we show that a general-purpose large language model (LLM) chatbot, Llama-3.1-8B-Instruct, can be transformed via supervised fine-tuning of engineered prompts into a chemical language model (CLM), SmileyLlama, for molecule generation. We benchmark SmileyLlama by comparing it to CLMs trained from scratch on large amounts of ChEMBL data for their ability to generate valid and novel drug-like molecules. We also use direct preference optimization to both improve SmileyLlama's adherence to a prompt and to generate molecules within the iMiner reinforcement learning framework to predict new drug molecules with optimized 3D conformations and high binding affinity to drug targets, illustrated with the SARS-Cov-2 Main Protease. This overall framework allows a LLM to speak directly as a CLM which can generate molecules with user-specified properties, rather than acting only as a chatbot with knowledge of chemistry or as a helpful virtual assistant. While our dataset and analyses are geared toward drug discovery, this general procedure can be extended to other chemical applications such as chemical synthesis.
Authors: August Y. Chen, Karthik Sridharan
Abstract: In this paper, we prove that optimizability of any function F using Gradient Flow from all initializations implies a Poincar\'e Inequality for Gibbs measures mu_{beta} = e^{-beta F}/Z at low temperature. In particular, under mild regularity assumptions on the convergence rate of Gradient Flow, we establish that mu_{beta} satisfies a Poincar\'e Inequality with constant O(C'+1/beta) for beta >= Omega(d), where C' is the Poincar\'e constant of mu_{beta} restricted to a neighborhood of the global minimizers of F. Under an additional mild condition on F, we show that mu_{beta} satisfies a Log-Sobolev Inequality with constant O(beta max(S, 1) max(C', 1)) where S denotes the second moment of mu_{beta}. Here asymptotic notation hides F-dependent parameters. At a high level, this establishes that optimizability via Gradient Flow from every initialization implies a Poincar\'e and Log-Sobolev Inequality for the low-temperature Gibbs measure, which in turn imply sampling from all initializations. Analogously, we establish that under the same assumptions, if F can be initialized from everywhere except some set S, then mu_{beta} satisfies a Weak Poincar\'e Inequality with parameters (O(C'+1/beta), O(mu_{beta}(S))) for \beta = Omega(d). At a high level, this shows while optimizability from 'most' initializations implies a Weak Poincar\'e Inequality, which in turn implies sampling from suitable warm starts. Our regularity assumptions are mild and as a consequence, we show we can efficiently sample from several new natural and interesting classes of non-log-concave densities, an important setting with relatively few examples. As another corollary, we obtain efficient discrete-time sampling results for log-concave measures satisfying milder regularity conditions than smoothness, similar to Lehec (2023).
Authors: Qingni Wang, Tiantian Geng, Zhiyuan Wang, Teng Wang, Bo Fu, Feng Zheng
Abstract: Multimodal Large Language Models (MLLMs) exhibit promising advancements across various tasks, yet they still encounter significant trustworthiness issues. Prior studies apply Split Conformal Prediction (SCP) in language modeling to construct prediction sets with statistical guarantees. However, these methods typically rely on internal model logits or are restricted to multiple-choice settings, which hampers their generalizability and adaptability in dynamic, open-ended environments. In this paper, we introduce TRON, a two-step framework for risk control and assessment, applicable to any MLLM that supports sampling in both open-ended and closed-ended scenarios. TRON comprises two main components: (1) a novel conformal score to sample response sets of minimum size, and (2) a nonconformity score to identify high-quality responses based on self-consistency theory, controlling the error rates by two specific risk levels. Furthermore, we investigate semantic redundancy in prediction sets within open-ended contexts for the first time, leading to a promising evaluation metric for MLLMs based on average set size. Our comprehensive experiments across four Video Question-Answering (VideoQA) datasets utilizing eight MLLMs show that TRON achieves desired error rates bounded by two user-specified risk levels. Additionally, deduplicated prediction sets maintain adaptiveness while being more efficient and stable for risk assessment under different risk levels.
Authors: Yixin Ji, Yang Xiang, Juntao Li, Qingrong Xia, Ping Li, Xinyu Duan, Zhefeng Wang, Min Zhang
Abstract: As large language models (LLMs) are widely applied across various fields, model compression has become increasingly crucial for reducing costs and improving inference efficiency. Post-training pruning is a promising method that does not require resource-intensive iterative training and only needs a small amount of calibration data to assess the importance of parameters. Recent research has enhanced post-training pruning from different aspects but few of them systematically explore the effects of calibration data, and it is unclear if there exist better calibration data construction strategies. We fill this blank and surprisingly observe that calibration data is also crucial to post-training pruning, especially for high sparsity. Through controlled experiments on important influence factors of calibration data, including the pruning settings, the amount of data, and its similarity with pre-training data, we observe that a small size of data is adequate, and more similar data to its pre-training stage can yield better performance. As pre-training data is usually inaccessible for advanced LLMs, we further provide a self-generating calibration data synthesis strategy to construct feasible calibration data. Experimental results on recent strong open-source LLMs (e.g., DCLM, and LLaMA-3) show that the proposed strategy can enhance the performance of strong pruning methods (e.g., Wanda, DSnoT, OWL) by a large margin (up to $2.68\%$). Code is available at https://github.com/Dereck0602/calibration_data.
Authors: Yijie Zhang, Luzhe Huang, Nir Pillar, Yuzhu Li, Hanlong Chen, Aydogan Ozcan
Abstract: Virtual staining of tissue offers a powerful tool for transforming label-free microscopy images of unstained tissue into equivalents of histochemically stained samples. This study presents a diffusion model-based super-resolution virtual staining approach utilizing a Brownian bridge process to enhance both the spatial resolution and fidelity of label-free virtual tissue staining, addressing the limitations of traditional deep learning-based methods. Our approach integrates novel sampling techniques into a diffusion model-based image inference process to significantly reduce the variance in the generated virtually stained images, resulting in more stable and accurate outputs. Blindly applied to lower-resolution auto-fluorescence images of label-free human lung tissue samples, the diffusion-based super-resolution virtual staining model consistently outperformed conventional approaches in resolution, structural similarity and perceptual accuracy, successfully achieving a super-resolution factor of 4-5x, increasing the output space-bandwidth product by 16-25-fold compared to the input label-free microscopy images. Diffusion-based super-resolved virtual tissue staining not only improves resolution and image quality but also enhances the reliability of virtual staining without traditional chemical staining, offering significant potential for clinical diagnostics.
Authors: Sohail Bahmani
Abstract: We derive a fundamental trade-off between standard and adversarial risk in a rather general situation that formalizes the following simple intuition: "If no (nearly) optimal predictor is smooth, adversarial robustness comes at the cost of accuracy." As a concrete example, we evaluate the derived trade-off in regression with polynomial ridge functions under mild regularity conditions. Generalizing our analysis of this example, we formulate a necessary condition under which adversarial robustness can be achieved without significant degradation of the accuracy. This necessary condition is expressed in terms of a quantity that resembles the Poincar\'{e} constant of the data distribution.
Authors: Qiao Qiao, Yuepei Li, Qing Wang, Kang Zhou, Qi Li
Abstract: Knowledge graph completion (KGC) is a task of inferring missing triples based on existing Knowledge Graphs (KGs). Both structural and semantic information are vital for successful KGC. However, existing methods only use either the structural knowledge from the KG embeddings or the semantic information from pre-trained language models (PLMs), leading to suboptimal model performance. Moreover, since PLMs are not trained on KGs, directly using PLMs to encode triples may be inappropriate. To overcome these limitations, we propose a novel framework called Bridge, which jointly encodes structural and semantic information of KGs. Specifically, we strategically encode entities and relations separately by PLMs to better utilize the semantic knowledge of PLMs and enable structured representation learning via a structural learning principle. Furthermore, to bridge the gap between KGs and PLMs, we employ a self-supervised representation learning method called BYOL to fine-tune PLMs with two different views of a triple. Unlike BYOL, which uses augmentation methods to create two semantically similar views of the same image, potentially altering the semantic information. We strategically separate the triple into two parts to create different views, thus avoiding semantic alteration. Experiments demonstrate that Bridge outperforms the SOTA models on three benchmark datasets.
Authors: David Robinson, Marius Miron, Masato Hagiwara, Benno Weck, Sara Keen, Milad Alizadeh, Gagan Narula, Matthieu Geist, Olivier Pietquin
Abstract: Large language models (LLMs) prompted with text and audio have achieved state-of-the-art performance across various auditory tasks, including speech, music, and general audio, showing emergent abilities on unseen tasks. However, their potential has yet to be fully demonstrated in bioacoustics tasks, such as detecting animal vocalizations in large recordings, classifying rare and endangered species, and labeling context and behavior -- tasks that are crucial for conservation, biodiversity monitoring, and animal behavior studies. In this work, we present NatureLM-audio, the first audio-language foundation model specifically designed for bioacoustics. Our training dataset consists of carefully curated text-audio pairs spanning bioacoustics, speech, and music, designed to address the field's limited availability of annotated data. We demonstrate successful transfer of learned representations from music and speech to bioacoustics, and our model shows promising generalization to unseen taxa and tasks. We evaluate NatureLM-audio on a novel benchmark (BEANS-Zero) and it sets a new state of the art on several bioacoustics tasks, including zero-shot classification of unseen species. To advance bioacoustics research, we release our model weights, benchmark data, and open-source the code for training and benchmark data generation and model training.
Authors: Anton Kuznietsov, Dirk Schweickard, Steven Peters
Abstract: In automated driving, object detection is crucial for perceiving the environment. Although deep learning-based detectors offer high performance, their black-box nature complicates safety assurance. We propose a novel methodology to analyze how object- and environment-related factors affect LiDAR- and camera-based 3D object detectors. A statistical univariate analysis relates each factor to pedestrian detection errors. Additionally, a Random Forest (RF) model predicts errors from meta-information, with Shapley Values interpreting feature importance. By capturing feature dependencies, the RF enables a nuanced analysis of detection errors. Understanding these factors reveals detector performance gaps and supports safer object detection system development.
Authors: Daniel P. Jeong, Pranav Mani, Saurabh Garg, Zachary C. Lipton, Michael Oberst
Abstract: Several recent works seek to adapt general-purpose large language models (LLMs) and vision-language models (VLMs) for medical applications through continued pretraining on publicly available biomedical corpora. These works typically claim that such domain-adaptive pretraining improves performance on various downstream medical tasks, such as answering medical exam questions. In this paper, we compare ten "medical" LLMs and two VLMs against their corresponding base models, arriving at a different conclusion: all medical VLMs and nearly all medical LLMs fail to consistently improve over their base models in the zero-/few-shot prompting and supervised fine-tuning regimes for medical question answering (QA). For instance, on clinical-note-based QA tasks in the 3-shot setting, medical LLMs outperform their base models in only 26.7% of cases, reach a (statistical) tie in 16.7% of cases, and perform significantly worse in the remaining 56.7% of cases. Our conclusions are based on (i) comparing each medical model directly against its base model; (ii) optimizing the prompts for each model separately in zero-/few-shot prompting; and (iii) accounting for statistical uncertainty in comparisons. Our findings suggest that state-of-the-art general-domain models may already exhibit strong medical knowledge and reasoning capabilities, and offer recommendations to strengthen the conclusions of future studies.
Authors: Shijie Zhou, Huaisheng Zhu, Rohan Sharma, Jiayi Chen, Ruiyi Zhang, Kaiyi Ji, Changyou Chen
Abstract: Diffusion models have emerged as a powerful foundation model for visual generations. With an appropriate sampling process, it can effectively serve as a generative prior for solving general inverse problems. Current posterior sampling-based methods take the measurement (i.e., degraded image sample) into the posterior sampling to infer the distribution of the target data (i.e., clean image sample). However, in this manner, we show that high-frequency information can be prematurely introduced during the early stages, which could induce larger posterior estimate errors during restoration sampling. To address this observation, we first reveal that forming the log-posterior gradient with the noisy measurement ( i.e., noisy measurement from a diffusion forward process) instead of the clean one can benefit the early posterior sampling. Consequently, we propose a novel diffusion posterior sampling method DPS-CM, which incorporates a Crafted Measurement (i.e., noisy measurement crafted by a reverse denoising process, rather than constructed from the diffusion forward process) to form the posterior estimate. This integration aims to mitigate the misalignment with the diffusion prior caused by cumulative posterior estimate errors. Experimental results demonstrate that our approach significantly improves the overall capacity to solve general and noisy inverse problems, such as Gaussian deblurring, super-resolution, inpainting, nonlinear deblurring, and tasks with Poisson noise, relative to existing approaches. Code is available at: https://github.com/sjz5202/DPS-CM.
Authors: Sanjay Das, Swastik Bhattacharya, Souvik Kundu, Shamik Kundu, Anand Menon, Arnab Raha, Kanad Basu
Abstract: Large Language Models (LLMs) have revolutionized natural language processing (NLP), excelling in tasks like text generation and summarization. However, their increasing adoption in mission-critical applications raises concerns about hardware-based threats, particularly bit-flip attacks (BFAs). BFAs, enabled by fault injection methods such as Rowhammer, target model parameters in memory, compromising both integrity and performance. Identifying critical parameters for BFAs in the vast parameter space of LLMs poses significant challenges. While prior research suggests transformer-based architectures are inherently more robust to BFAs compared to traditional deep neural networks, we challenge this assumption. For the first time, we demonstrate that as few as three bit-flips can cause catastrophic performance degradation in an LLM with billions of parameters. Current BFA techniques are inadequate for exploiting this vulnerability due to the difficulty of efficiently identifying critical parameters within the immense parameter space. To address this, we propose AttentionBreaker, a novel framework tailored for LLMs that enables efficient traversal of the parameter space to identify critical parameters. Additionally, we introduce GenBFA, an evolutionary optimization strategy designed to refine the search further, isolating the most critical bits for an efficient and effective attack. Empirical results reveal the profound vulnerability of LLMs to AttentionBreaker. For example, merely three bit-flips (4.129 x 10^-9% of total parameters) in the LLaMA3-8B-Instruct 8-bit quantized (W8) model result in a complete performance collapse: accuracy on MMLU tasks drops from 67.3% to 0%, and Wikitext perplexity skyrockets from 12.6 to 4.72 x 10^5. These findings underscore the effectiveness of AttentionBreaker in uncovering and exploiting critical vulnerabilities within LLM architectures.
Authors: Yang Liu, Siao Liu, Xiaoguang Zhu, Jielin Li, Hao Yang, Liangyu Teng, Juncen Guo, Yan Wang, Dingkang Yang, Jing Liu
Abstract: Video Anomaly Detection (VAD) aims to automatically analyze spatiotemporal patterns in surveillance videos collected from open spaces to detect anomalous events that may cause harm, such as fighting, stealing, and car accidents. However, vision-based surveillance systems such as closed-circuit television often capture personally identifiable information. The lack of transparency and interpretability in video transmission and usage raises public concerns about privacy and ethics, limiting the real-world application of VAD. Recently, researchers have focused on privacy concerns in VAD by conducting systematic studies from various perspectives including data, features, and systems, making Privacy-Preserving Video Anomaly Detection (P2VAD) a hotspot in the AI community. However, current research in P2VAD is fragmented, and prior reviews have mostly focused on methods using RGB sequences, overlooking privacy leakage and appearance bias considerations. To address this gap, this article is the first to systematically reviews the progress of P2VAD, defining its scope and providing an intuitive taxonomy. We outline the basic assumptions, learning frameworks, and optimization objectives of various approaches, analyzing their strengths, weaknesses, and potential correlations. Additionally, we provide open access to research resources such as benchmark datasets and available code. Finally, we discuss key challenges and future opportunities from the perspectives of AI development and P2VAD deployment, aiming to guide future work in the field.
Authors: Dietmar Jannach, Alan Said, Marko Tkal\v{c}i\v{c}, Markus Zanker
Abstract: In the area of recommender systems, the vast majority of research efforts is spent on developing increasingly sophisticated recommendation models, also using increasingly more computational resources. Unfortunately, most of these research efforts target a very small set of application domains, mostly e-commerce and media recommendation. Furthermore, many of these models are never evaluated with users, let alone put into practice. The scientific, economic and societal value of much of these efforts by scholars therefore remains largely unclear. To achieve a stronger positive impact resulting from these efforts, we posit that we as a research community should more often address use cases where recommender systems contribute to societal good (RS4Good). In this opinion piece, we first discuss a number of examples where the use of recommender systems for problems of societal concern has been successfully explored in the literature. We then proceed by outlining a paradigmatic shift that is needed to conduct successful RS4Good research, where the key ingredients are interdisciplinary collaborations and longitudinal evaluation approaches with humans in the loop.
Authors: Ziyao Zeng, Jingcheng Ni, Daniel Wang, Patrick Rim, Younjoon Chung, Fengyu Yang, Byung-Woo Hong, Alex Wong
Abstract: Traditional monocular depth estimation suffers from inherent ambiguity and visual nuisance. We argue that language prior can enhance monocular depth estimation by leveraging the inductive bias learned during the text-to-image pre-training of diffusion models. The ability of these models to generate images that align with text indicates that they have learned the spatial relationships, size, and shape of specified objects, which can be applied to improve depth estimation. Thus, we propose PriorDiffusion, using a pre-trained text-to-image diffusion model that takes both images and corresponding text descriptions to infer affine-invariant depth through a denoising process. We also show that language prior enhances the model's perception of specific regions of images that users care about and describe. Simultaneously, language prior acts as a constraint to accelerate the convergence of both training and the inference diffusion trajectory. By training on HyperSim and Virtual KITTI, we achieve faster training convergence, fewer inference diffusion steps, and state-of-the-art zero-shot performance across NYUv2, KITTI, ETH3D, and ScanNet. Code will be released upon acceptance.
Authors: Taekyung Ki, Dongchan Min, Gyeongsu Chae
Abstract: With the rapid advancement of diffusion-based generative models, portrait image animation has achieved remarkable results. However, it still faces challenges in temporally consistent video generation and fast sampling due to its iterative sampling nature. This paper presents FLOAT, an audio-driven talking portrait video generation method based on flow matching generative model. Instead of a pixel-based latent space, we take advantage of a learned orthogonal motion latent space, enabling efficient generation and editing of temporally consistent motion. To achieve this, we introduce a transformer-based vector field predictor with an effective frame-wise conditioning mechanism. Additionally, our method supports speech-driven emotion enhancement, enabling a natural incorporation of expressive motions. Extensive experiments demonstrate that our method outperforms state-of-the-art audio-driven talking portrait methods in terms of visual quality, motion fidelity, and efficiency.
Authors: Rongkun Xue, Jinouwen Zhang, Yazhe Niu, Dazhong Shen, Bingqi Ma, Yu Liu, Jing Yang
Abstract: Recent generative models based on score matching and flow matching have significantly advanced generation tasks, but their potential in discriminative tasks remains underexplored. Previous approaches, such as generative classifiers, have not fully leveraged the capabilities of these models for discriminative tasks due to their intricate designs. We propose Pretrained Reversible Generation (PRG), which extracts unsupervised representations by reversing the generative process of a pretrained continuous generation model. PRG effectively reuses unsupervised generative models, leveraging their high capacity to serve as robust and generalizable feature extractors for downstream tasks. This framework enables the flexible selection of feature hierarchies tailored to specific downstream tasks. Our method consistently outperforms prior approaches across multiple benchmarks, achieving state-of-the-art performance among generative model based methods, including 78% top-1 accuracy on ImageNet at a resolution of 64*64. Extensive ablation studies, including out-of-distribution evaluations, further validate the effectiveness of our approach.PRG is available at https://github.com/opendilab/PRG.
Authors: Guy Kornowski, Ohad Shamir
Abstract: We study the problem of solving matrix games of the form $\max_{\mathbf{w}\in\mathcal{W}}\min_{\mathbf{p}\in\Delta}\mathbf{p}^{\top}A\mathbf{w}$, where $A$ is some matrix and $\Delta$ is the probability simplex. This problem encapsulates canonical tasks such as finding a linear separator and computing Nash equilibria in zero-sum games. However, perhaps surprisingly, its inherent complexity (as formalized in the standard framework of oracle complexity [Nemirovski and Yudin, 1983]) is not well-understood. In this work, we first identify different oracle models which are implicitly used by prior algorithms, amounting to multiplying the matrix $A$ by a vector from either one or both sides. We then prove complexity lower bounds for algorithms under both access models, which in particular imply a separation between them. Specifically, we start by showing that algorithms for linear separability based on one-sided multiplications must require $\Omega(\gamma_A^{-2})$ iterations, where $\gamma_A$ is the margin, as matched by the Perceptron algorithm. We then prove that accelerated algorithms for this task, which utilize multiplications from both sides, must require $\tilde{\Omega}(\gamma_{A}^{-2/3})$ iterations, establishing the first oracle complexity barrier for such algorithms. Finally, by adapting our lower bound to $\ell_1$ geometry, we prove that computing an $\epsilon$-approximate Nash equilibrium requires $\tilde{\Omega}(\epsilon^{-2/5})$ iterations, which is an exponential improvement over the previously best-known lower bound due to Hadiji et al. [2024].
Authors: Runtao Liu, I Chieh Chen, Jindong Gu, Jipeng Zhang, Renjie Pi, Qifeng Chen, Philip Torr, Ashkan Khakzar, Fabio Pizzati
Abstract: Text-to-image (T2I) models are widespread, but their limited safety guardrails expose end users to harmful content and potentially allow for model misuse. Current safety measures are typically limited to text-based filtering or concept removal strategies, able to remove just a few concepts from the model's generative capabilities. In this work, we introduce AlignGuard, a method for safety alignment of T2I models. We enable the application of Direct Preference Optimization (DPO) for safety purposes in T2I models by synthetically generating a dataset of harmful and safe image-text pairs, which we call CoProV2. Using a custom DPO strategy and this dataset, we train safety experts, in the form of low-rank adaptation (LoRA) matrices, able to guide the generation process away from specific safety-related concepts. Then, we merge the experts into a single LoRA using a novel merging strategy for optimal scaling performance. This expert-based approach enables scalability, allowing us to remove 7x more harmful concepts from T2I models compared to baselines. AlignGuard consistently outperforms the state-of-the-art on many benchmarks and establishes new practices for safety alignment in T2I networks. Code and data will be shared at https://safetydpo.github.io/.
Authors: Giang Nguyen, Ivan Brugere, Shubham Sharma, Sanjay Kariyappa, Anh Totti Nguyen, Freddy Lecue
Abstract: Interpretability in Table Question Answering (Table QA) is critical, especially in high-stakes domains like finance and healthcare. While recent Table QA approaches based on Large Language Models (LLMs) achieve high accuracy, they often produce ambiguous explanations of how answers are derived. We propose Plan-of-SQLs (POS), a new Table QA method that makes the model's decision-making process interpretable. POS decomposes a question into a sequence of atomic steps, each directly translated into an executable SQL command on the table, thereby ensuring that every intermediate result is transparent. Through extensive experiments, we show that: First, POS generates the highest-quality explanations among compared methods, which markedly improves the users' ability to simulate and verify the model's decisions. Second, when evaluated on standard Table QA benchmarks (TabFact, WikiTQ, and FeTaQA), POS achieves QA accuracy that is competitive to existing methods, while also offering greater efficiency-requiring significantly fewer LLM calls and table database queries (up to 25x fewer)-and more robust performance on large-sized tables. Finally, we observe high agreement (up to 90.59% in forward simulation) between LLMs and human users when making decisions based on the same explanations, suggesting that LLMs could serve as an effective proxy for humans in evaluating Table QA explanations.
Authors: Yang Cai, Siddharth Mitra, Xiuyuan Wang, Andre Wibisono
Abstract: We study zero-sum games in the space of probability distributions over the Euclidean space $\mathbb{R}^d$ with entropy regularization, in the setting when the interaction function between the players is smooth and strongly convex-strongly concave. We prove an exponential convergence guarantee for the mean-field min-max Langevin dynamics to compute the equilibrium distribution of the zero-sum game. We also study the finite-particle approximation of the mean-field min-max Langevin dynamics, both in continuous and discrete times. We prove biased convergence guarantees for the continuous-time finite-particle min-max Langevin dynamics to the stationary mean-field equilibrium distribution with an explicit bias term which does not scale with the number of particles. We also prove biased convergence guarantees for the discrete-time finite-particle min-max Langevin algorithm to the stationary mean-field equilibrium distribution with an additional bias term which scales with the step size and the number of particles. This provides an explicit iteration complexity for the average particle along the finite-particle algorithm to approximately compute the equilibrium distribution of the zero-sum game.
Authors: Tien Dang, Viet Thanh Duy Nguyen, Minh Tuan Le, Truong-Son Hy
Abstract: Biomedical Knowledge Graphs (BKGs) integrate diverse datasets to elucidate complex relationships within the biomedical field. Effective link prediction on these graphs can uncover valuable connections, such as potential novel drug-disease relations. We introduce a novel multimodal approach that unifies embeddings from specialized Language Models (LMs) with Graph Contrastive Learning (GCL) to enhance intra-entity relationships while employing a Knowledge Graph Embedding (KGE) model to capture inter-entity relationships for effective link prediction. To address limitations in existing BKGs, we present PrimeKG++, an enriched knowledge graph incorporating multimodal data, including biological sequences and textual descriptions for each entity type. By combining semantic and relational information in a unified representation, our approach demonstrates strong generalizability, enabling accurate link predictions even for unseen nodes. Experimental results on PrimeKG++ and the DrugBank drug-target interaction dataset demonstrate the effectiveness and robustness of our method across diverse biomedical datasets. Our source code, pre-trained models, and data are publicly available at https://github.com/HySonLab/BioMedKG
Authors: Yixin Ji, Juntao Li, Yang Xiang, Hai Ye, Kaixin Wu, Kai Yao, Jia Xu, Linjian Mo, Min Zhang
Abstract: The remarkable performance of the o1 model in complex reasoning demonstrates that test-time compute scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time compute scaling. We trace the concept of test-time compute back to System-1 models. In System-1 models, test-time compute addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time compute in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out advanced topics and future directions.
Authors: Mingyang Song, Zhaochen Su, Xiaoye Qu, Jiawei Zhou, Yu Cheng
Abstract: Process-level Reward Models (PRMs) are crucial for complex reasoning and decision-making tasks, where each intermediate step plays an important role in the reasoning process. Since language models are prone to various types of errors during the reasoning process, PRMs are required to possess nuanced capabilities for detecting various implicit error types in real-world scenarios. However, current benchmarks primarily focus on step correctness, failing to evaluate PRMs' performance systematically. To address this gap, we introduce PRMBench, a process-level benchmark specifically designed to assess the fine-grained error detection capabilities of PRMs. PRMBench comprises 6,216 carefully designed problems and 83,456 step-level labels, evaluating models across multiple dimensions, including simplicity, soundness, and sensitivity. In our experiments on 15 models, spanning both open-source PRMs and closed-source large language models prompted as critic models, we uncover significant weaknesses in current PRMs. These findings underscore the challenges inherent in process-level evaluation and highlight key directions for future research. We hope PRMBench can be a robust bench for advancing research on PRM evaluation and development.
Authors: Joris Bekkers
Abstract: Pressing is a fundamental defensive strategy in football, characterized by applying pressure on the ball owning team to regain possession. Despite its significance, existing metrics for measuring pressing often lack precision or comprehensive consideration of positional data, player movement and speed. This research introduces an innovative framework for quantifying pressing intensity, leveraging advancements in positional tracking data and components from Spearman's Pitch Control model. Our method integrates player velocities, movement directions, and reaction times to compute the time required for a defender to intercept an attacker or the ball. This time-to-intercept measure is then transformed into probabilistic values using a logistic function, enabling dynamic and intuitive analysis of pressing situations at the individual frame level. the model captures how every player's movement influences pressure on the field, offering actionable insights for coaches, analysts, and decision-makers. By providing a robust and intepretable metric, our approach facilitates the identification of pressing strategies, advanced situational analyses, and the derivation of metrics, advancing the analytical capabilities for modern football.
Authors: Gursimran Singh, Xinglu Wang, Yifan Hu, Timothy Yu, Linzi Xing, Wei Jiang, Zhefeng Wang, Xiaolong Bai, Yi Li, Ying Xiong, Yong Zhang, Zhenan Fan
Abstract: Large Multimodal Models (LMMs) extend Large Language Models (LLMs) by handling diverse inputs such as images, audio, and video, but at the cost of adding a multimodal encoding stage that increases both computational and memory overhead. This step negatively affects key Service Level Objectives (SLOs), such as time to first token (TTFT) and time per output token (TPOT). We introduce Encode-Prefill-Decode (EPD) Disaggregation, a novel framework that separates the encoding, prefill, and decode stages onto dedicated resources. Unlike current systems, which bundle encoding and prefill together, our approach decouples these steps, unlocking new opportunities and optimizations. These include a mechanism to cache multimedia tokens for efficient transfer, a novel way to parallelize the encoding load within a request, a module for optimal resource allocation for disaggregated serving, and a novel role-switching method to handle changing workload characteristics. Experimental evaluations with popular LMMs show substantial gains in memory efficiency (up to 15x lower peak memory utilization), batch sizes (up to 22x larger), 10x more images per request, and 2.2x larger KV caches. Furthermore, it leads to significant improvements in SLO attainment (up to 90-100% improvement) and TTFT (up to 71% reduction), compared to systems that do not disaggregate. The code is available at https://github.com/vbdi/epdserve.
Authors: Lars van der Laan, David Hubbard, Allen Tran, Nathan Kallus, Aur\'elien Bibaut
Abstract: Long-term causal effects often must be estimated from short-term data due to limited follow-up in healthcare, economics, and online platforms. Markov Decision Processes (MDPs) provide a natural framework for capturing such long-term dynamics through sequences of states, actions, and rewards. Double Reinforcement Learning (DRL) enables efficient inference on policy values in MDPs, but nonparametric implementations require strong intertemporal overlap assumptions and often exhibit high variance and instability. We propose a semiparametric extension of DRL for efficient inference on linear functionals of the Q-function--such as policy values--in infinite-horizon, time-homogeneous MDPs. By imposing structural restrictions on the Q-function, our approach relaxes the strong overlap conditions required by nonparametric methods and improves statistical efficiency. Under model misspecification, our estimators target the functional of the best-approximating Q-function, with only second-order bias. We provide conditions for valid inference using sieve methods and data-driven model selection. A central challenge in DRL is the estimation of nuisance functions, such as density ratios, which often involve difficult minimax optimization. To address this, we introduce a novel plug-in estimator based on isotonic Bellman calibration, which combines fitted Q-iteration with an isotonic regression adjustment. The estimator is debiased without requiring estimation of additional nuisance functions and reduces high-dimensional overlap assumptions to a one-dimensional condition. Bellman calibration extends isotonic calibration--widely used in prediction and classification--to the MDP setting and may be of independent interest.
Authors: Fei Wu, Jia Hu, Geyong Min, Shiqiang Wang
Abstract: Pre-trained Language Models (PLMs) have demonstrated their superiority and versatility in modern Natural Language Processing (NLP), effectively adapting to various downstream tasks through further fine-tuning. Federated Parameter-Efficient Fine-Tuning (FedPEFT) has emerged as a promising solution to address privacy and efficiency challenges in distributed training for PLMs on resource-constrained local devices. However, our measurements reveal two key limitations of FedPEFT: heterogeneous data across devices exacerbates performance degradation of low-rank adaptation, and a fixed parameter configuration results in communication inefficiency. To overcome these limitations, we propose FedARA, a novel Adaptive Rank Allocation framework for federated parameter-efficient fine-tuning of language models. Specifically, FedARA employs truncated Singular Value Decomposition (SVD) adaptation to enhance similar feature representation across clients, significantly mitigating the adverse effects of data heterogeneity. Subsequently, it utilizes dynamic rank allocation to progressively identify critical ranks, effectively improving communication efficiency. Lastly, it leverages rank-based module pruning to automatically remove inactive modules, steadily reducing local computational cost and memory usage in each federated learning round. Extensive experiments show that FedARA consistently outperforms baselines by an average of 6.95% to 8.49% across various datasets and models under heterogeneous data while significantly improving communication efficiency by 2.40$ \times$. Moreover, experiments on various edge devices demonstrate substantial decreases in total training time and energy consumption by up to 48.90% and 46.95%, respectively.
Authors: Kenza Tazi, Sun Woo P. Kim, Marc Girona-Mata, Richard E. Turner
Abstract: High Mountain Asia (HMA) holds the highest concentration of frozen water outside the polar regions, serving as a crucial water source for more than 1.9 billion people. Precipitation represents the largest source of uncertainty for future hydrological modelling in this area. In this study, we propose a probabilistic machine learning framework to combine monthly precipitation from 13 regional climate models developed under the Coordinated Regional Downscaling Experiment (CORDEX) over HMA via a mixture of experts (MoE). This approach accounts for seasonal and spatial biases within the models, enabling the prediction of more faithful precipitation distributions. The MoE is trained and validated against gridded historical precipitation data, yielding 32% improvement over an equally-weighted average and 254% improvement over choosing any single ensemble member. This approach is then used to generate precipitation projections for the near future (2036-2065) and far future (2066-2095) under RCP4.5 and RCP8.5 scenarios. Compared to previous estimates, the MoE projects wetter summers but drier winters over the western Himalayas and Karakoram and wetter winters over the Tibetan Plateau, Hengduan Shan, and South East Tibet.
Authors: Ying Chen, Paul Griffin, Paolo Recchia, Lei Zhou, Hongrui Zhang
Abstract: Recovery rate prediction plays a pivotal role in bond investment strategies by enhancing risk assessment, optimizing portfolio allocation, improving pricing accuracy, and supporting effective credit risk management. However, accurate forecasting remains challenging due to complex nonlinear dependencies, high-dimensional feature spaces, and limited sample sizes-conditions under which classical machine learning models are prone to overfitting. We propose a hybrid Quantum Machine Learning (QML) model with Amplitude Encoding, leveraging the unitarity constraint of Parametrized Quantum Circuits (PQC) and the exponential data compression capability of qubits. We evaluate the model on a global recovery rate dataset comprising 1,725 observations and 256 features from 1996 to 2023. Our hybrid method significantly outperforms both classical neural networks and QML models using Angle Encoding, achieving a lower Root Mean Squared Error (RMSE) of 0.228, compared to 0.246 and 0.242, respectively. It also performs competitively with ensemble tree methods such as XGBoost. While practical implementation challenges remain for Noisy Intermediate-Scale Quantum (NISQ) hardware, our quantum simulation and preliminary results on noisy simulators demonstrate the promise of hybrid quantum-classical architectures in enhancing the accuracy and robustness of recovery rate forecasting. These findings illustrate the potential of quantum machine learning in shaping the future of credit risk prediction.
Authors: Ali Naseh, Yuefeng Peng, Anshuman Suri, Harsh Chaudhari, Alina Oprea, Amir Houmansadr
Abstract: Retrieval-Augmented Generation (RAG) enables Large Language Models (LLMs) to generate grounded responses by leveraging external knowledge databases without altering model parameters. Although the absence of weight tuning prevents leakage via model parameters, it introduces the risk of inference adversaries exploiting retrieved documents in the model's context. Existing methods for membership inference and data extraction often rely on jailbreaking or carefully crafted unnatural queries, which can be easily detected or thwarted with query rewriting techniques common in RAG systems. In this work, we present Interrogation Attack (IA), a membership inference technique targeting documents in the RAG datastore. By crafting natural-text queries that are answerable only with the target document's presence, our approach demonstrates successful inference with just 30 queries while remaining stealthy; straightforward detectors identify adversarial prompts from existing methods up to ~76x more frequently than those generated by our attack. We observe a 2x improvement in TPR@1%FPR over prior inference attacks across diverse RAG configurations, all while costing less than $0.02 per document inference.
Authors: Jinhao Liang, Jacob K Christopher, Sven Koenig, Ferdinando Fioretto
Abstract: Recent advances in diffusion models hold significant potential in robotics, enabling the generation of diverse and smooth trajectories directly from raw representations of the environment. Despite this promise, applying diffusion models to motion planning remains challenging due to their difficulty in enforcing critical constraints, such as collision avoidance and kinematic feasibility. These limitations become even more pronounced in Multi-Robot Motion Planning (MRMP), where multiple robots must coordinate in shared spaces. To address these challenges, this work proposes Simultaneous MRMP Diffusion (SMD), a novel approach integrating constrained optimization into the diffusion sampling process to produce collision-free, kinematically feasible trajectories. Additionally, the paper introduces a comprehensive MRMP benchmark to evaluate trajectory planning algorithms across scenarios with varying robot densities, obstacle complexities, and motion constraints. Experimental results show SMD consistently outperforms classical and other learning-based motion planners, achieving higher success rates and efficiency in complex multi-robot environments.
Authors: Xiaorui Su, Shvat Messica, Yepeng Huang, Ruth Johnson, Lukas Fesser, Shanghua Gao, Faryad Sahneh, Marinka Zitnik
Abstract: Foundation models trained on patient electronic health records (EHRs) require tokenizing medical data into sequences of discrete vocabulary items. Existing tokenizers treat medical codes from EHRs as isolated textual tokens. However, each medical code is defined by its textual description, its position in ontological hierarchies, and its relationships to other codes, such as disease co-occurrences and drug-treatment associations. Medical vocabularies contain more than 600,000 codes with critical information for clinical reasoning. We introduce MedTok, a multimodal medical code tokenizer that uses the text descriptions and relational context of codes. MedTok processes text using a language model encoder and encodes the relational structure with a graph encoder. It then quantizes both modalities into a unified token space, preserving modality-specific and cross-modality information. We integrate MedTok into five EHR models and evaluate it on operational and clinical tasks across in-patient and out-patient datasets, including outcome prediction, diagnosis classification, drug recommendation, and risk stratification. Swapping standard EHR tokenizers with MedTok improves AUPRC across all EHR models, by 4.10% on MIMIC-III, 4.78% on MIMIC-IV, and 11.32% on EHRShot, with the largest gains in drug recommendation. Beyond EHR modeling, we demonstrate using MedTok tokenizer with medical QA systems. Our results demonstrate the potential of MedTok as a unified tokenizer for medical codes, improving tokenization for medical foundation models.
Authors: Andre Wibisono
Abstract: We study the mixing time guarantee for sampling in relative Fisher information via the Proximal Sampler algorithm, which is an approximate proximal discretization of the Langevin dynamics. We show that when the target probability distribution is strongly log-concave, the relative Fisher information converges exponentially fast along the Proximal Sampler; this matches the exponential convergence rate of the relative Fisher information along the continuous-time Langevin dynamics for strongly log-concave target. When combined with a standard implementation of the Proximal Sampler via rejection sampling, this exponential convergence rate provides a high-accuracy iteration complexity guarantee for the Proximal Sampler in relative Fisher information when the target distribution is strongly log-concave and log-smooth. Our proof proceeds by establishing a strong data processing inequality for relative Fisher information along the Gaussian channel under strong log-concavity, and a data processing inequality along the reverse Gaussian channel for a special distribution. The forward and reverse Gaussian channels compose to form the Proximal Sampler, and these data processing inequalities imply the exponential convergence rate of the relative Fisher information along the Proximal Sampler.
Authors: Lars van der Laan, Ahmed Alaa
Abstract: Ensuring model calibration is critical for reliable prediction, yet popular distribution-free methods such as histogram binning and isotonic regression offer only asymptotic guarantees. We introduce a unified framework for Venn and Venn-Abers calibration that extends Vovk's approach beyond binary classification to a broad class of prediction problems defined by generic loss functions. Our method transforms any perfectly in-sample calibrated predictor into a set-valued predictor that, in finite samples, outputs at least one marginally calibrated point prediction. These set predictions shrink asymptotically and converge to a conditionally calibrated prediction, capturing epistemic uncertainty. We further propose Venn multicalibration, a new approach for achieving finite-sample calibration across subpopulations. For quantile loss, our framework recovers group-conditional and multicalibrated conformal prediction as special cases and yields novel prediction intervals with quantile-conditional coverage.
Authors: Zijiang Yan, Jianhua Pei, Hongda Wu, Hina Tabassum, Ping Wang
Abstract: This paper proposes a novel Semantic Communication (SemCom) framework for real-time adaptive-bitrate video streaming by integrating Latent Diffusion Models (LDMs) within the FFmpeg techniques. This solution addresses the challenges of high bandwidth usage, storage inefficiencies, and quality of experience (QoE) degradation associated with traditional Constant Bitrate Streaming (CBS) and Adaptive Bitrate Streaming (ABS). The proposed approach leverages LDMs to compress I-frames into a latent space, offering significant storage and semantic transmission savings without sacrificing high visual quality. While retaining B-frames and P-frames as adjustment metadata to support efficient refinement of video reconstruction at the user side, the proposed framework further incorporates state-of-the-art denoising and Video Frame Interpolation (VFI) techniques. These techniques mitigate semantic ambiguity and restore temporal coherence between frames, even in noisy wireless communication environments. Experimental results demonstrate the proposed method achieves high-quality video streaming with optimized bandwidth usage, outperforming state-of-the-art solutions in terms of QoE and resource efficiency. This work opens new possibilities for scalable real-time video streaming in 5G and future post-5G networks.
Authors: Yizhe Zhang, Richard Bai, Zijin Gu, Ruixiang Zhang, Jiatao Gu, Emmanuel Abbe, Samy Bengio, Navdeep Jaitly
Abstract: Language models usually use left-to-right (L2R) autoregressive factorization. However, L2R factorization may not always be the best inductive bias. Therefore, we investigate whether alternative factorizations of the text distribution could be beneficial in some tasks. We investigate right-to-left (R2L) training as a compelling alternative, focusing on multiple-choice questions (MCQs) as a test bed for knowledge extraction and reasoning. Through extensive experiments across various model sizes (2B-8B parameters) and training datasets, we find that R2L models can significantly outperform L2R models on several MCQ benchmarks, including logical reasoning, commonsense understanding, and truthfulness assessment tasks. Our analysis reveals that this performance difference may be fundamentally linked to multiple factors including calibration, computability, and directional conditional entropy. We analyze the impact of these factors through controlled simulation studies using arithmetic tasks, where the impacting factors can be better disentangled. Our work demonstrates that exploring alternative factorizations of the text distribution can lead to improvements in LLM capabilities and provides theoretical insights into optimal factorization towards approximating human language distribution, and when each reasoning order might be more advantageous. Our code and checkpoints are released at https://github.com/apple/ml-reversal-blessing.
Authors: Joshua Kazdan, Abhay Puri, Rylan Schaeffer, Lisa Yu, Chris Cundy, Jason Stanley, Sanmi Koyejo, Krishnamurthy Dvijotham
Abstract: Leading language model (LM) providers like OpenAI and Anthropic allow customers to fine-tune frontier LMs for specific use cases. To prevent abuse, these providers apply filters to block fine-tuning on overtly harmful data. In this setting, we make three contributions: First, while past work has shown that safety alignment is "shallow", we correspondingly demonstrate that existing fine-tuning attacks are shallow -- attacks target only the first several tokens of the model response, and consequently can be blocked by generating the first several response tokens with an aligned model. Second, we conceptually illustrate how to make attacks deeper by introducing a new fine-tuning attack that trains models to first refuse harmful requests before answering them; this "refuse-then-comply" strategy bypasses shallow defenses and produces harmful responses that evade output filters. Third, we demonstrate the potency of our new fine-tuning attack by jailbreaking both open-source models equipped with defenses and production models, achieving attack success rates of 57% and 72% against GPT-4o and Claude Haiku, respectively. Our attack received a $2000 bug bounty from OpenAI and was acknowledged as a vulnerability by Anthropic. Our work undermines the notion that models are safe because they initially refuse harmful requests and broadens awareness of the scope of attacks that face production fine-tuning APIs.
Authors: Hamed Khosravi, Mohammad Reza Shafie, Ahmed Shoyeb Raihan, Srinjoy Das, Imtiaz Ahmed
Abstract: Existing contextual multi-armed bandit (MAB) algorithms fail to effectively capture both long-term trends and local patterns across all arms, leading to suboptimal performance in environments with rapidly changing reward structures. They also rely on static exploration rates, which do not dynamically adjust to changing conditions. To overcome these limitations, we propose LNUCB-TA, a hybrid bandit model integrating a novel nonlinear component (adaptive k-Nearest Neighbors (k-NN)) for reducing time complexity, alongside a global-and-local attention-based exploration mechanism. Our approach uniquely combines linear and nonlinear estimation techniques, with the nonlinear module dynamically adjusting k based on reward variance to enhance spatiotemporal pattern recognition. This reduces the likelihood of selecting suboptimal arms while improving reward estimation accuracy and computational efficiency. The attention-based mechanism ranks arms by past performance and selection frequency, dynamically adjusting exploration and exploitation in real time without requiring manual tuning of exploration rates. By integrating global attention (assessing all arms collectively) and local attention (focusing on individual arms), LNUCB-TA efficiently adapts to temporal and spatial complexities. Empirical results show LNUCB-TA significantly outperforms state-of-the-art linear, nonlinear, and hybrid bandits in cumulative and mean reward, convergence, and robustness across different exploration rates. Theoretical analysis further confirms its reliability with a sub-linear regret bound.
Authors: Yaxuan Kong, Yiyuan Yang, Yoontae Hwang, Wenjie Du, Stefan Zohren, Zhangyang Wang, Ming Jin, Qingsong Wen
Abstract: Time series data are foundational in finance, healthcare, and energy domains. However, most existing methods and datasets remain focused on a narrow spectrum of tasks, such as forecasting or anomaly detection. To bridge this gap, we introduce Time Series Multi-Task Question Answering (Time-MQA), a unified framework that enables natural language queries across multiple time series tasks - numerical analytical tasks and open-ended question answering with reasoning. Central to Time-MQA is the TSQA dataset, a large-scale dataset containing $\sim$200k question-answer pairs derived from diverse time series spanning environment, traffic, etc. This comprehensive resource covers various time series lengths and promotes robust model development. We further demonstrate how continually pre-training large language models (Mistral 7B, Llama-3 8B, and Qwen-2.5 7B) on the TSQA dataset enhanced time series reasoning capabilities, moving beyond mere numeric tasks and enabling more advanced and intuitive interactions with temporal data. The complete TSQA dataset, models, user study questionnaires for evaluation, and other related materials have been open-sourced.
Authors: Ritwik Gupta, Rodolfo Corona, Jiaxin Ge, Eric Wang, Dan Klein, Trevor Darrell, David M. Chan
Abstract: Large language models (LLMs) exhibit the ability to generalize given few-shot examples in their input prompt, an emergent capability known as in-context learning (ICL). We investigate whether LLMs use ICL to perform structured reasoning in ways that are consistent with a Bayesian framework or rely on pattern matching. Using a controlled setting of biased coin flips, we find that: (1) LLMs often possess biased priors, causing initial divergence in zero-shot settings, (2) in-context evidence outweighs explicit bias instructions, (3) LLMs broadly follow Bayesian posterior updates, with deviations primarily due to miscalibrated priors rather than flawed updates, and (4) attention magnitude has negligible effect on Bayesian inference. With sufficient demonstrations of biased coin flips via ICL, LLMs update their priors in a Bayesian manner.
Authors: Jinze Li, Yixing Xu, Haiduo Huang, Xuanwu Yin, Dong Li, Edith C. H. Ngai, Emad Barsoum
Abstract: Speculative decoding (SPD) aims to accelerate the auto-regressive token generation process of a target Large Language Model (LLM). Some approaches employ a draft model with multiple heads to predict a sequence of future tokens, where each head handles a token in the sequence. The target LLM verifies the predicted sequence and accepts aligned tokens, enabling efficient multi-token generation. However, existing methods assume that all tokens within a sequence are equally important, employing identical head structures and relying on a single-generation paradigm, either serial or parallel. To this end, we theoretically demonstrate that initial tokens in the draft sequence are more important than later ones. Building on this insight, we propose Gumiho, a hybrid model combining serial and parallel heads. Specifically, given the critical importance of early tokens, we employ a sophisticated Transformer architecture for the early draft heads in a serial configuration to improve accuracy. For later tokens, we utilize multiple lightweight MLP heads operating in parallel to enhance efficiency. By allocating more advanced model structures and longer running times to the early heads, Gumiho achieves improved overall performance. The experimental results demonstrate that our method outperforms existing approaches, fully validating its effectiveness.
Authors: Guy Barzilai, Ohad Shamir, Moslem Zamani
Abstract: In this paper, we study when we might expect the optimization curve induced by gradient descent to be \emph{convex} -- precluding, for example, an initial plateau followed by a sharp decrease, making it difficult to decide when optimization should stop. Although such undesirable behavior can certainly occur when optimizing general functions, might it also occur in the benign and well-studied case of smooth convex functions? As far as we know, this question has not been tackled in previous work. We show, perhaps surprisingly, that the answer crucially depends on the choice of the step size. In particular, for the range of step sizes which are known to result in monotonic convergence to an optimal value, we characterize a regime where the optimization curve will be provably convex, and a regime where the curve can be non-convex. We also extend our results to gradient flow, and to the closely-related but different question of whether the gradient norm decreases monotonically.
Authors: Victor Rielly, Kamel Lahouel, Chau Nguyen, Anthony Kolshorn, Nicholas Fisher, Bruno Jedynak
Abstract: We present a Representer Theorem result for a large class of weak formulation problems. We provide examples of applications of our formulation both in traditional machine learning and numerical methods as well as in new and emerging techniques. Finally we apply our formulation to generalize the multivariate occupation kernel (MOCK) method for learning dynamical systems from data proposing the more general Riesz Occupation Kernel (ROCK) method. Our generalized methods are both more computationally efficient and performant on most of the benchmarks we test against.
Authors: Guido Carnevale, Nicola Bastianello
Abstract: In this paper, we design a novel distributed learning algorithm using stochastic compressed communications. In detail, we pursue a modular approach, merging ADMM and a gradient-based approach, benefiting from the robustness of the former and the computational efficiency of the latter. Additionally, we integrate a stochastic integral action (error feedback) enabling almost sure rejection of the compression error. We analyze the resulting method in nonconvex scenarios and guarantee almost sure asymptotic convergence to the set of stationary points of the problem. This result is obtained using system-theoretic tools based on stochastic timescale separation. We corroborate our findings with numerical simulations in nonconvex classification.
Authors: George Panagopoulos, Johannes F. Lutzeyer, Sofiane Ennadir, Michalis Vazirgiannis, Jun Pang
Abstract: Genomic studies, including CRISPR-based PerturbSeq analyses, face a vast hypothesis space, while gene perturbations remain costly and time-consuming. Gene expression models based on graph neural networks are trained to predict the outcomes of gene perturbations to facilitate such experiments. Active learning methods are often employed to train these models due to the cost of the genomic experiments required to build the training set. However, poor model initialization in active learning can result in suboptimal early selections, wasting time and valuable resources. While typical active learning mitigates this issue over many iterations, the limited number of experimental cycles in genomic studies exacerbates the risk. To this end, we propose graph-based data filtering as an alternative. Unlike active learning, data filtering selects the gene perturbations before training, meaning it is free of bias due to random initialization and initial random selection. Moreover, reducing the iterations between the wet lab and the model provides several operational advantages resulting in significant acceleration. The proposed methods are motivated by theoretical studies of graph neural network generalization. The criteria are defined over the input graph and are optimized with submodular maximization. We compare them empirically to baselines and active learning methods that are state-of-the-art. The results demonstrate that graph-based data filtering achieves comparable accuracy while alleviating the aforementioned risks.
Authors: Kunal Mukherjee, Zachary Harrison, Saeid Balaneshin
Abstract: Transparency and interpretability are crucial for enhancing customer confidence and user engagement, especially when dealing with black-box Machine Learning (ML)-based recommendation systems. Modern recommendation systems leverage Graph Neural Network (GNN) due to their ability to produce high-quality recommendations in terms of both relevance and diversity. Therefore, the explainability of GNN is especially important for Link Prediction (LP) tasks since recommending relevant items can be viewed as predicting links between users and items. GNN explainability has been a well-studied field, but existing methods primarily focus on node or graph-level tasks, leaving a gap in LP explanation techniques. This work introduces Z-REx, a GNN explanation framework designed explicitly for heterogeneous link prediction tasks. Z-REx utilizes structural and attribute perturbation to identify critical substructures and important features while reducing the search space by leveraging domain-specific knowledge. In our experimentation, we show the efficacy of Z-REx in generating contextually relevant and human-interpretable explanations for ZiGNN, a GNN-based recommendation engine, using a real-world real-estate dataset from Zillow Group, Inc. We compare against State-of-The-Art (SOTA) GNN explainers to show Z-REx outperforms them by 61% in the Fidelity metric by producing superior human-interpretable explanations.
Authors: Youngjun Song, Youngsik Hwang, Jonghun Lee, Heechang Lee, Dong-Young Lim
Abstract: Domain generalization (DG) aims to learn models that perform well on unseen target domains by training on multiple source domains. Sharpness-Aware Minimization (SAM), known for finding flat minima that improve generalization, has therefore been widely adopted in DG. However, our analysis reveals that SAM in DG may converge to \textit{fake flat minima}, where the total loss surface appears flat in terms of global sharpness but remains sharp with respect to individual source domains. To understand this phenomenon more precisely, we formalize the average worst-case domain risk as the maximum loss under domain distribution shifts within a bounded divergence, and derive a generalization bound that reveals the limitations of global sharpness-aware minimization. In contrast, we show that individual sharpness provides a valid upper bound on this risk, making it a more suitable proxy for robust domain generalization. Motivated by these insights, we shift the DG paradigm toward minimizing individual sharpness across source domains. We propose \textit{Decreased-overhead Gradual SAM (DGSAM)}, which applies gradual domain-wise perturbations in a computationally efficient manner to consistently reduce individual sharpness. Extensive experiments demonstrate that DGSAM not only improves average accuracy but also reduces performance variance across domains, while incurring less computational overhead than SAM.
Authors: Fernando Koch, Aladin Djuhera, Alecio Binotto
Abstract: Large Foundation Models (LFMs), including multi-modal and generative models, promise to unlock new capabilities for next-generation Edge AI applications. However, performing inference with LFMs in resource-constrained and heterogeneous edge environments, such as Multi-access Edge Computing (MEC), presents significant challenges for workload orchestration due to time-varying network, compute, and storage conditions. In particular, current split inference strategies, which partition LFM layers across nodes, are not designed to adapt to fluctuating workloads, dynamic bandwidth conditions, or evolving privacy constraints in high-utilization MEC environments. In this work, we propose a novel adaptive split inference orchestration framework that elevates both the placement and partitioning of LFM layers to runtime-tunable variables. Specifically, our framework enables real-time, quality-of-service (QoS)-aware management of inference workloads by extending conventional orchestrators with three key services: (1) Capacity-aware workload distribution, which continuously profiles node resources and selects an optimal subset of MEC nodes; (2) Dynamic partition migration, which transparently relocates pre-cut LFM segments in response to changes in utilization or network conditions; (3) Real-time reconfiguration, which dynamically re-splits LFM layers to balance latency, throughput, and privacy. We formalize the joint placement-partitioning problem, outline a reference architecture and algorithmic workflow, and discuss applicability in representative smart city, V2X, and industrial edge scenarios.
Authors: Gen Li, Yang Xiao, Jie Ji, Kaiyuan Deng, Bo Hui, Linke Guo, Xiaolong Ma
Abstract: Text-to-image (T2I) diffusion models have achieved remarkable success in generating high-quality images from textual prompts. However, their ability to store vast amounts of knowledge raises concerns in scenarios where selective forgetting is necessary, such as removing copyrighted content, reducing biases, or eliminating harmful concepts. While existing unlearning methods can remove certain concepts, they struggle with multi-concept forgetting due to instability, residual knowledge persistence, and generation quality degradation. To address these challenges, we propose \textbf{Dynamic Mask coupled with Concept-Aware Loss}, a novel unlearning framework designed for multi-concept forgetting in diffusion models. Our \textbf{Dynamic Mask} mechanism adaptively updates gradient masks based on current optimization states, allowing selective weight modifications that prevent interference with unrelated knowledge. Additionally, our \textbf{Concept-Aware Loss} explicitly guides the unlearning process by enforcing semantic consistency through superclass alignment, while a regularization loss based on knowledge distillation ensures that previously unlearned concepts remain forgotten during sequential unlearning. We conduct extensive experiments to evaluate our approach. Results demonstrate that our method outperforms existing unlearning techniques in forgetting effectiveness, output fidelity, and semantic coherence, particularly in multi-concept scenarios. Our work provides a principled and flexible framework for stable and high-fidelity unlearning in generative models. The code will be released publicly.
Authors: Haris Riaz, Sourav Bhabesh, Vinayak Arannil, Miguel Ballesteros, Graham Horwood
Abstract: Recent smaller language models such Phi-3.5 and Phi-4 rely on synthetic data generated using larger Language models. Questions remain about leveraging synthetic data for other use cases, such as adapting LLMs to specific domains. A key limitation of synthetic data is low diversity, which negatively impacts its downstream applicability for improving other models. To address this, we propose MetaSynth, a method for generating synthetic data that enhances diversity through meta-prompting, where a language model orchestrates multiple "expert" LLM agents to collaboratively generate data. Using only 25 million tokens of synthetic data generated with MetaSynth, we successfully adapt a well-trained LLM (Mistral-7B-v0.3) to two specialized domains-Finance and Biomedicine-without compromising the capabilities of the resulting model in general tasks. In addition, we evaluate the diversity of our synthetic data using seven automated metrics, and find that it approaches the diversity of LLM pre-training corpora. Continually pre-training Mistral-7B-v0.3 with MetaSynth notably outperforms the base LLM, showing improvements of up to 4.08% in Finance and 13.75% in Biomedicine. The same model shows degraded performance when trained on data generated using a template prompt, even when the template includes prior generations and varying In-Context exemplars of real data. Our findings suggest that a few million tokens of diverse synthetic data without mixing any real data, is sufficient for effective domain adaptation when using MetaSynth.
Authors: Zhiyuan Wang, Qingni Wang, Yue Zhang, Tianlong Chen, Xiaofeng Zhu, Xiaoshuang Shi, Kaidi Xu
Abstract: As large language models are increasingly utilized in real-world applications, guarantees of task-specific metrics are essential for their reliable deployment. Previous studies have introduced various criteria of conformal uncertainty grounded in split conformal prediction, which offer user-specified correctness coverage. However, existing frameworks often fail to identify uncertainty data outliers that violate the exchangeability assumption, leading to unbounded miscoverage rates and unactionable prediction sets. In this paper, we propose a novel approach termed Selective Conformal Uncertainty (SConU), which, for the first time, implements significance tests, by developing two conformal p-values that are instrumental in determining whether a given sample deviates from the uncertainty distribution of the calibration set at a specific manageable risk level. Our approach not only facilitates rigorous management of miscoverage rates across both single-domain and interdisciplinary contexts, but also enhances the efficiency of predictions. Furthermore, we comprehensively analyze the components of the conformal procedures, aiming to approximate conditional coverage, particularly in high-stakes question-answering tasks.
Authors: Yuxin Zuo, Kaiyan Zhang, Li Sheng, Shang Qu, Ganqu Cui, Xuekai Zhu, Haozhan Li, Yuchen Zhang, Xinwei Long, Ermo Hua, Biqing Qi, Youbang Sun, Zhiyuan Ma, Lifan Yuan, Ning Ding, Bowen Zhou
Abstract: This paper investigates Reinforcement Learning (RL) on data without explicit labels for reasoning tasks in Large Language Models (LLMs). The core challenge of the problem is reward estimation during inference while not having access to ground-truth information. While this setting appears elusive, we find that common practices in Test-Time Scaling (TTS), such as majority voting, yield surprisingly effective rewards suitable for driving RL training. In this work, we introduce Test-Time Reinforcement Learning (TTRL), a novel method for training LLMs using RL on unlabeled data. TTRL enables self-evolution of LLMs by utilizing the priors in the pre-trained models. Our experiments demonstrate that TTRL consistently improves performance across a variety of tasks and models. Notably, TTRL boosts the pass@1 performance of Qwen-2.5-Math-7B by approximately 211% on the AIME 2024 with only unlabeled test data. Furthermore, although TTRL is only supervised by the maj@n metric, TTRL has demonstrated performance to consistently surpass the upper limit of the initial model maj@n, and approach the performance of models trained directly on test data with ground-truth labels. Our experimental findings validate the general effectiveness of TTRL across various tasks and highlight TTRL's potential for broader tasks and domains. GitHub: https://github.com/PRIME-RL/TTRL
Authors: Chuangtao Chen, Qinglin Zhao, MengChu Zhou, Dusit Niyato, Zhimin He, Haozhen Situ
Abstract: Discrete diffusion models represent a significant advance in generative modeling, demonstrating remarkable success in synthesizing complex, high-quality discrete data. However, to avoid exponential computational costs, they typically rely on calculating per-dimension transition probabilities when learning high-dimensional distributions. In this study, we rigorously prove that this approach leads to a worst-case linear scaling of Kullback-Leibler (KL) divergence with data dimension. To address this, we propose a Quantum Discrete Denoising Diffusion Probabilistic Model (QD3PM), which enables joint probability learning through diffusion and denoising in exponentially large Hilbert spaces, offering a theoretical pathway to faithfully capture the true joint distribution. By deriving posterior states through quantum Bayes' theorem, similar to the crucial role of posterior probabilities in classical diffusion models, and by learning the joint probability, we establish a solid theoretical foundation for quantum-enhanced diffusion models. For denoising, we design a quantum circuit that utilizes temporal information for parameter sharing and incorporates learnable classical-data-controlled rotations for encoding. Exploiting joint distribution learning, our approach enables single-step sampling from pure noise, eliminating iterative requirements of existing models. Simulations demonstrate the proposed model's superior accuracy in modeling complex distributions compared to factorization methods. Hence, this paper establishes a new theoretical paradigm in generative models by leveraging the quantum advantage in joint distribution learning.
Authors: Xiangqi Wang, Yue Huang, Yanbo Wang, Xiaonan Luo, Kehan Guo, Yujun Zhou, Xiangliang Zhang
Abstract: LLMs often need effective configurations, like temperature and reasoning steps, to handle tasks requiring sophisticated reasoning and problem-solving, ranging from joke generation to mathematical reasoning. Existing prompting approaches usually adopt general-purpose, fixed configurations that work 'well enough' across tasks but seldom achieve task-specific optimality. To address this gap, we introduce AdaReasoner, an LLM-agnostic plugin designed for any LLM to automate adaptive reasoning configurations for tasks requiring different types of thinking. AdaReasoner is trained using a reinforcement learning (RL) framework, combining a factorized action space with a targeted exploration strategy, along with a pretrained reward model to optimize the policy model for reasoning configurations with only a few-shot guide. AdaReasoner is backed by theoretical guarantees and experiments of fast convergence and a sublinear policy gap. Across six different LLMs and a variety of reasoning tasks, it consistently outperforms standard baselines, preserves out-of-distribution robustness, and yield gains on knowledge-intensive tasks through tailored prompts.
Authors: Chun-Yi Kuan, Hung-yi Lee
Abstract: Audio-aware large language models (ALLMs) have recently made great strides in understanding and processing audio inputs. These models are typically adapted from text-based large language models (LLMs) through additional training on audio-related tasks. However, this adaptation process presents two major limitations. First, ALLMs often suffer from catastrophic forgetting, where crucial textual capabilities like instruction-following are lost after training on audio data. In some cases, models may even hallucinate sounds that are not present in the input audio, raising concerns about reliability. Second, achieving cross-modal alignment between audio and language typically relies on large collections of task-specific question-answer pairs for instruction tuning, making it resource-intensive. To address these issues, previous works have leveraged the backbone LLMs to synthesize general-purpose, caption-style alignment data. In this paper, we propose a data generation framework that produces contrastive-like training data, designed to enhance ALLMs' ability to differentiate between present and absent sounds. We further extend our approach to multi-audio scenarios, enabling the model to either explain differences between audio inputs or produce unified captions that describe all inputs, thereby enhancing audio-language alignment. We refer to the entire ALLM training framework as bootstrapping audio-language alignment via synthetic data generation from backbone LLMs (BALSa). Experimental results indicate that our method effectively mitigates audio hallucinations while reliably maintaining strong performance on audio understanding and reasoning benchmarks, as well as instruction-following skills. Moreover, incorporating multi-audio training further enhances the model's comprehension and reasoning capabilities. Overall, BALSa offers an efficient and scalable approach to developing ALLMs.
Authors: Chenghao Qian, Wenjing Li, Yuhu Guo, Gustav Markkula
Abstract: In this work, we present WeatherEdit, a novel weather editing pipeline for generating realistic weather effects with controllable types and severity in 3D scenes. Our approach is structured into two key components: weather background editing and weather particle construction. For weather background editing, we introduce an all-in-one adapter that integrates multiple weather styles into a single pretrained diffusion model, enabling the generation of diverse weather effects in 2D image backgrounds. During inference, we design a Temporal-View (TV-) attention mechanism that follows a specific order to aggregate temporal and spatial information, ensuring consistent editing across multi-frame and multi-view images. To construct the weather particles, we first reconstruct a 3D scene using the edited images and then introduce a dynamic 4D Gaussian field to generate snowflakes, raindrops and fog in the scene. The attributes and dynamics of these particles are precisely controlled through physical-based modelling and simulation, ensuring realistic weather representation and flexible severity adjustments. Finally, we integrate the 4D Gaussian field with the 3D scene to render consistent and highly realistic weather effects. Experiments on multiple driving datasets demonstrate that WeatherEdit can generate diverse weather effects with controllable condition severity, highlighting its potential for autonomous driving simulation in adverse weather. See project page: https://jumponthemoon.github.io/w-edit
Authors: Matteo Gallici, Haitz S\'aez de Oc\'ariz Borde
Abstract: Fine-tuning pre-trained generative models with Reinforcement Learning (RL) has emerged as an effective approach for aligning outputs more closely with nuanced human preferences. In this paper, we investigate the application of Group Relative Policy Optimization (GRPO) to fine-tune next-scale visual autoregressive (VAR) models. Our empirical results demonstrate that this approach enables alignment to intricate reward signals derived from aesthetic predictors and CLIP embeddings, significantly enhancing image quality and enabling precise control over the generation style. Interestingly, by leveraging CLIP, our method can help VAR models generalize beyond their initial ImageNet distribution: through RL-driven exploration, these models can generate images aligned with prompts referencing image styles that were absent during pre-training. In summary, we show that RL-based fine-tuning is both efficient and effective for VAR models, benefiting particularly from their fast inference speeds, which are advantageous for online sampling, an aspect that poses significant challenges for diffusion-based alternatives.
Authors: Benjamin Arnav, Pablo Bernabeu-P\'erez, Nathan Helm-Burger, Tim Kostolansky, Hannes Whittingham, Mary Phuong
Abstract: As AI models are deployed with increasing autonomy, it is important to ensure they do not take harmful actions unnoticed. As a potential mitigation, we investigate Chain-of-Thought (CoT) monitoring, wherein a weaker trusted monitor model continuously oversees the intermediate reasoning steps of a more powerful but untrusted model. We compare CoT monitoring to action-only monitoring, where only final outputs are reviewed, in a red-teaming setup where the untrusted model is instructed to pursue harmful side tasks while completing a coding problem. We find that CoT monitoring improves detection by up to 27 percentage points in scenarios where action-only monitoring fails to reliably identify sabotage. However, CoT traces can also contain misleading rationalizations that deceive the monitor, reducing performance in more obvious sabotage cases. To address this, we introduce a hybrid protocol that independently scores both reasoning and final outputs and combines them using a weighted average. This hybrid monitor consistently outperforms both CoT and action-only monitors across all tested models and tasks, with detection rates over four times higher than action-only monitoring for subtle deception scenarios.
Authors: Giovanni Acampora, Andris Ambainis, Natalia Ares, Leonardo Banchi, Pallavi Bhardwaj, Daniele Binosi, G. Andrew D. Briggs, Tommaso Calarco, Vedran Dunjko, Jens Eisert, Olivier Ezratty, Paul Erker, Federico Fedele, Elies Gil-Fuster, Martin G\"arttner, Mats Granath, Markus Heyl, Iordanis Kerenidis, Matthias Klusch, Anton Frisk Kockum, Richard Kueng, Mario Krenn, J\"org L\"assig, Antonio Macaluso, Sabrina Maniscalco, Florian Marquardt, Kristel Michielsen, Gorka Mu\~noz-Gil, Daniel M\"ussig, Hendrik Poulsen Nautrup, Sophie A. Neubauer, Evert van Nieuwenburg, Roman Orus, J\"org Schmiedmayer, Markus Schmitt, Philipp Slusallek, Filippo Vicentini, Christof Weitenberg, Frank K. Wilhelm
Abstract: This white paper discusses and explores the various points of intersection between quantum computing and artificial intelligence (AI). It describes how quantum computing could support the development of innovative AI solutions. It also examines use cases of classical AI that can empower research and development in quantum technologies, with a focus on quantum computing and quantum sensing. The purpose of this white paper is to provide a long-term research agenda aimed at addressing foundational questions about how AI and quantum computing interact and benefit one another. It concludes with a set of recommendations and challenges, including how to orchestrate the proposed theoretical work, align quantum AI developments with quantum hardware roadmaps, estimate both classical and quantum resources - especially with the goal of mitigating and optimizing energy consumption - advance this emerging hybrid software engineering discipline, and enhance European industrial competitiveness while considering societal implications.
Authors: M. S\"uzen
Abstract: A proposition that connects randomness and compression is put forward via Gibbs entropy over set of measurement vectors associated with a compression process. The proposition states that a lossy compression process is equivalent to {\it directed randomness} that preserves information content. The proposition originated from the observed behaviour in newly proposed {\it Dual Tomographic Compression} (DTC) compress-train framework. This is akin to tomographic reconstruction of layer weight matrices via building compressed sensed projections, via so-called {\it weight rays}. This tomographic approach is applied to previous and next layers in a dual fashion, that triggers neuronal-level pruning. This novel model compress-train scheme appears in iterative fashion and acts as a smart neural architecture search, The experiments demonstrated the utility of this dual-tomography producing state-of-the-art performance with efficient compression during training, accelerating and supporting lottery ticket hypothesis. However, random compress-train iterations having similar performance demonstrated the connection between randomness and compression from statistical physics perspective, we formulated the so-called {\it Gibbs randomness-compression proposition}, signifying randomness-compression relationship via Gibbs entropy. Practically, the DTC framework provides a promising approach for massively energy- and resource-efficient deep learning training.
Authors: Mahmud Ashraf Shamim, Eric A F Reinhardt, Talal Ahmed Chowdhury, Sergei Gleyzer, Paulo T Araujo
Abstract: Neural Quantum States (NQS) are a class of variational wave functions parametrized by neural networks (NNs) to study quantum many-body systems. In this work, we propose \texttt{SineKAN}, a NQS \textit{ansatz} based on Kolmogorov-Arnold Networks (KANs), to represent quantum mechanical wave functions as nested univariate functions. We show that \texttt{SineKAN} wavefunction with learnable sinusoidal activation functions can capture the ground state energies, fidelities and various correlation functions of the one dimensional Transverse-Field Ising model, Anisotropic Heisenberg model, and Antiferromagnetic $J_{1}-J_{2}$ model with different chain lengths. In our study of the $J_1-J_2$ model with $L=100$ sites, we find that the \texttt{SineKAN} model outperforms several previously explored neural quantum state \textit{ans\"atze}, including Restricted Boltzmann Machines (RBMs), Long Short-Term Memory models (LSTMs), and Multi-layer Perceptrons (MLP) \textit{a.k.a.} Feed Forward Neural Networks, when compared to the results obtained from the Density Matrix Renormalization Group (DMRG) algorithm. We find that \texttt{SineKAN} models can be trained to high precisions and accuracies with minimal computational costs.
Authors: Junghyun Lee, Kyoungseok Jang, Kwang-Sung Jun, Milan Vojnovi\'c, Se-Young Yun
Abstract: We present `GL-LowPopArt`, a novel Catoni-style estimator for generalized low-rank trace regression. Building on `LowPopArt` (Jang et al., 2024), it employs a two-stage approach: nuclear norm regularization followed by matrix Catoni estimation. We establish state-of-the-art estimation error bounds, surpassing existing guarantees (Fan et al., 2019; Kang et al., 2022), and reveal a novel experimental design objective, $\mathrm{GL}(\pi)$. The key technical challenge is controlling bias from the nonlinear inverse link function, which we address by our two-stage approach. We prove a *local* minimax lower bound, showing that our `GL-LowPopArt` enjoys instance-wise optimality up to the condition number of the ground-truth Hessian. Applications include generalized linear matrix completion, where `GL-LowPopArt` achieves a state-of-the-art Frobenius error guarantee, and **bilinear dueling bandits**, a novel setting inspired by general preference learning (Zhang et al., 2024). Our analysis of a `GL-LowPopArt`-based explore-then-commit algorithm reveals a new, potentially interesting problem-dependent quantity, along with improved Borda regret bound than vectorization (Wu et al., 2024).
Authors: Yang Cai, Alkis Kalavasis, Katerina Mamali, Anay Mehrotra, Manolis Zampetakis
Abstract: Most of the widely used estimators of the average treatment effect (ATE) in causal inference rely on the assumptions of unconfoundedness and overlap. Unconfoundedness requires that the observed covariates account for all correlations between the outcome and treatment. Overlap requires the existence of randomness in treatment decisions for all individuals. Nevertheless, many types of studies frequently violate unconfoundedness or overlap, for instance, observational studies with deterministic treatment decisions - popularly known as Regression Discontinuity designs - violate overlap. In this paper, we initiate the study of general conditions that enable the identification of the average treatment effect, extending beyond unconfoundedness and overlap. In particular, following the paradigm of statistical learning theory, we provide an interpretable condition that is sufficient and necessary for the identification of ATE. Moreover, this condition also characterizes the identification of the average treatment effect on the treated (ATT) and can be used to characterize other treatment effects as well. To illustrate the utility of our condition, we present several well-studied scenarios where our condition is satisfied and, hence, we prove that ATE can be identified in regimes that prior works could not capture. For example, under mild assumptions on the data distributions, this holds for the models proposed by Tan (2006) and Rosenbaum (2002), and the Regression Discontinuity design model introduced by Thistlethwaite and Campbell (1960). For each of these scenarios, we also show that, under natural additional assumptions, ATE can be estimated from finite samples. We believe these findings open new avenues for bridging learning-theoretic insights and causal inference methodologies, particularly in observational studies with complex treatment mechanisms.
Authors: Xiangchen Li, Dimitrios Spatharakis, Saeid Ghafouri, Jiakun Fan, Hans Vandierendonck, Deepu John, Bo Ji, Dimitrios Nikolopoulos
Abstract: The growing gap between the increasing complexity of large language models (LLMs) and the limited computational budgets of edge devices poses a key challenge for efficient on-device inference, despite gradual improvements in hardware capabilities. Existing strategies, such as aggressive quantization, pruning, or remote inference, trade accuracy for efficiency or lead to substantial cost burdens. This position paper introduces a new framework that leverages speculative decoding, previously viewed primarily as a decoding acceleration technique for autoregressive generation of LLMs, as a promising approach specifically adapted for edge computing by orchestrating computation across heterogeneous devices. We propose \acronym, a framework that allows lightweight edge devices to draft multiple candidate tokens locally using diverse draft models, while a single, shared edge server verifies the tokens utilizing a more precise target model. To further increase the efficiency of verification, the edge server batch the diverse verification requests from devices. This approach supports device heterogeneity and reduces server-side memory footprint by sharing the same upstream target model across multiple devices. Our initial experiments with Jetson Orin Nano, Raspberry Pi 4B/5, and an edge server equipped with 4 Nvidia A100 GPUs indicate substantial benefits: 2.2 more system throughput, 2.8 more system capacity, and better cost efficiency, all without sacrificing model accuracy.
Authors: Jiaqi Wu, Jiahong Ouyang, Farshad Moradi, Mohammad Mehdi Khalighi, Greg Zaharchuk
Abstract: Fluorodeoxyglucose (FDG) PET to evaluate patients with epilepsy is one of the most common applications for simultaneous PET/MRI, given the need to image both brain structure and metabolism, but is suboptimal due to the radiation dose in this young population. Little work has been done synthesizing diagnostic quality PET images from MRI data or MRI data with ultralow-dose PET using advanced generative AI methods, such as diffusion models, with attention to clinical evaluations tailored for the epilepsy population. Here we compared the performance of diffusion- and non-diffusion-based deep learning models for the MRI-to-PET image translation task for epilepsy imaging using simultaneous PET/MRI in 52 subjects (40 train/2 validate/10 hold-out test). We tested three different models: 2 score-based generative diffusion models (SGM-Karras Diffusion [SGM-KD] and SGM-variance preserving [SGM-VP]) and a Transformer-Unet. We report results on standard image processing metrics as well as clinically relevant metrics, including congruency measures (Congruence Index and Congruency Mean Absolute Error) that assess hemispheric metabolic asymmetry, which is a key part of the clinical analysis of these images. The SGM-KD produced the best qualitative and quantitative results when synthesizing PET purely from T1w and T2 FLAIR images with the least mean absolute error in whole-brain specific uptake value ratio (SUVR) and highest intraclass correlation coefficient. When 1% low-dose PET images are included in the inputs, all models improve significantly and are interchangeable for quantitative performance and visual quality. In summary, SGMs hold great potential for pure MRI-to-PET translation, while all 3 model types can synthesize full-dose FDG-PET accurately using MRI and ultralow-dose PET.
Authors: Yinghao Ma, Siyou Li, Juntao Yu, Emmanouil Benetos, Akira Maezawa
Abstract: Recent advances in audio-text large language models (LLMs) have opened new possibilities for music understanding and generation. However, existing benchmarks are limited in scope, often relying on simplified tasks or multi-choice evaluations that fail to reflect the complexity of real-world music analysis. We reinterpret a broad range of traditional MIR annotations as instruction-following formats and introduce CMI-Bench, a comprehensive music instruction following benchmark designed to evaluate audio-text LLMs on a diverse set of music information retrieval (MIR) tasks. These include genre classification, emotion regression, emotion tagging, instrument classification, pitch estimation, key detection, lyrics transcription, melody extraction, vocal technique recognition, instrument performance technique detection, music tagging, music captioning, and (down)beat tracking: reflecting core challenges in MIR research. Unlike previous benchmarks, CMI-Bench adopts standardized evaluation metrics consistent with previous state-of-the-art MIR models, ensuring direct comparability with supervised approaches. We provide an evaluation toolkit supporting all open-source audio-textual LLMs, including LTU, Qwen-audio, SALMONN, MusiLingo, etc. Experiment results reveal significant performance gaps between LLMs and supervised models, along with their culture, chronological and gender bias, highlighting the potential and limitations of current models in addressing MIR tasks. CMI-Bench establishes a unified foundation for evaluating music instruction following, driving progress in music-aware LLMs.
Authors: Kasidit Srimahajariyapong, Supanut Thanasilp, Thiparat Chotibut
Abstract: Variational quantum algorithms (VQAs) promise near-term quantum advantage, yet parametrized quantum states commonly built from the digital gate-based approach often suffer from scalability issues such as barren plateaus, where the loss landscape becomes flat. We study an analog VQA ans\"atze composed of $M$ quenches of a disordered Ising chain, whose dynamics is native to several quantum simulation platforms. By tuning the disorder strength we place each quench in either a thermalized phase or a many-body-localized (MBL) phase and analyse (i) the ans\"atze's expressivity and (ii) the scaling of loss variance. Numerics shows that both phases reach maximal expressivity at large $M$, but barren plateaus emerge at far smaller $M$ in the thermalized phase than in the MBL phase. Exploiting this gap, we propose an MBL initialisation strategy: initialise the ans\"atze in the MBL regime at intermediate quench $M$, enabling an initial trainability while retaining sufficient expressivity for subsequent optimization. The results link quantum phases of matter and VQA trainability, and provide practical guidelines for scaling analog-hardware VQAs.
Authors: Meenatchi Sundaram Muthu Selva Annamalai, Borja Balle, Jamie Hayes, Georgios Kaissis, Emiliano De Cristofaro
Abstract: This paper systematizes research on auditing Differential Privacy (DP) techniques, aiming to identify key insights into the current state of the art and open challenges. First, we introduce a comprehensive framework for reviewing work in the field and establish three cross-contextual desiderata that DP audits should target--namely, efficiency, end-to-end-ness, and tightness. Then, we systematize the modes of operation of state-of-the-art DP auditing techniques, including threat models, attacks, and evaluation functions. This allows us to highlight key details overlooked by prior work, analyze the limiting factors to achieving the three desiderata, and identify open research problems. Overall, our work provides a reusable and systematic methodology geared to assess progress in the field and identify friction points and future directions for our community to focus on.
Authors: Lincan Li, Eren Erman Ozguven, Yue Zhao, Guang Wang, Yiqun Xie, Yushun Dong
Abstract: Accurate typhoon track forecasting is crucial for early system warning and disaster response. While Transformer-based models have demonstrated strong performance in modeling the temporal dynamics of dense trajectories of humans and vehicles in smart cities, they usually lack access to broader contextual knowledge that enhances the forecasting reliability of sparse meteorological trajectories, such as typhoon tracks. To address this challenge, we propose TyphoFormer, a novel framework that incorporates natural language descriptions as auxiliary prompts to improve typhoon trajectory forecasting. For each time step, we use Large Language Model (LLM) to generate concise textual descriptions based on the numerical attributes recorded in the North Atlantic hurricane database. The language descriptions capture high-level meteorological semantics and are embedded as auxiliary special tokens prepended to the numerical time series input. By integrating both textual and sequential information within a unified Transformer encoder, TyphoFormer enables the model to leverage contextual cues that are otherwise inaccessible through numerical features alone. Extensive experiments are conducted on HURDAT2 benchmark, results show that TyphoFormer consistently outperforms other state-of-the-art baseline methods, particularly under challenging scenarios involving nonlinear path shifts and limited historical observations.
Authors: Siyang Song, Yupeng Huo, Shiqing Tang, Jiaee Cheong, Rui Gao, Michel Valstar, Hatice Gunes
Abstract: Depression is a common mental illness across current human society. Traditional depression assessment relying on inventories and interviews with psychologists frequently suffer from subjective diagnosis results, slow and expensive diagnosis process as well as lack of human resources. Since there is a solid evidence that depression is reflected by various human internal brain activities and external expressive behaviours, early traditional machine learning (ML) and advanced deep learning (DL) models have been widely explored for human behaviour-based automatic depression assessment (ADA) since 2012. However, recent ADA surveys typically only focus on a limited number of human behaviour modalities. Despite being used as a theoretical basis for developing ADA approaches, existing ADA surveys lack a comprehensive review and summary of multi-modal depression-related human behaviours. To bridge this gap, this paper specifically summarises depression-related human behaviours across a range of modalities (e.g. the human brain, verbal language and non-verbal audio/facial/body behaviours). We focus on conducting an up-to-date and comprehensive survey of ML-based ADA approaches for learning depression cues from these behaviours as well as discussing and comparing their distinctive features and limitations. In addition, we also review existing ADA competitions and datasets, identify and discuss the main challenges and opportunities to provide further research directions for future ADA researchers.
Authors: Jiahui Hu, Wenjun Dong
Abstract: We present Compressible Atmospheric Model-Network (CAM-NET), an AI model designed to predict neutral atmospheric variables from the Earth's surface to the ionosphere with high accuracy and computational efficiency. Accurate modeling of the entire atmosphere is critical for understanding the upward propagation of gravity waves, which influence upper-atmospheric dynamics and coupling across atmospheric layers. CAM-NET leverages the Spherical Fourier Neural Operator (SFNO) to capture global-scale atmospheric dynamics while preserving the Earth's spherical structure. Trained on a decade of datasets from the Whole Atmosphere Community Climate Model with thermosphere and ionosphere eXtension (WACCM-X), CAM-NET demonstrates accuracy comparable to WACCM-X while achieving a speedup of over 1000x in inference time, can provide one year simulation within a few minutes once trained. The model effectively predicts key atmospheric parameters, including zonal and meridional winds, temperature, and time rate of pressure. Inspired by traditional modeling approaches that use external couplers to simulate tracer transport, CAM-NET introduces a modular architecture that explicitly separates tracer prediction from core dynamics. The core backbone of CAM-NET focuses on forecasting primary physical variables (e.g., temperature, wind velocity), while tracer variables are predicted through a lightweight, fine-tuned model. This design allows for efficient adaptation to specific tracer scenarios with minimal computational cost, avoiding the need to retrain the entire model. We have validated this approach on the $O^2$ tracer, demonstrating strong performance and generalization capabilities.
Authors: Gilad Lerman, Kang Li, Tyler Maunu, Teng Zhang
Abstract: Robust subspace estimation is fundamental to many machine learning and data analysis tasks. Iteratively Reweighted Least Squares (IRLS) is an elegant and empirically effective approach to this problem, yet its theoretical properties remain poorly understood. This paper establishes that, under deterministic conditions, a variant of IRLS with dynamic smoothing regularization converges linearly to the underlying subspace from any initialization. We extend these guarantees to affine subspace estimation, a setting that lacks prior recovery theory. Additionally, we illustrate the practical benefits of IRLS through an application to low-dimensional neural network training. Our results provide the first global convergence guarantees for IRLS in robust subspace recovery and, more broadly, for nonconvex IRLS on a Riemannian manifold.
Authors: Xinzhuo Li, Adheesh Juvekar, Xingyou Liu, Muntasir Wahed, Kiet A. Nguyen, Ismini Lourentzou
Abstract: Recent progress in vision-language segmentation has significantly advanced grounded visual understanding. However, these models often exhibit hallucinations by producing segmentation masks for objects not grounded in the image content or by incorrectly labeling irrelevant regions. Existing evaluation protocols for segmentation hallucination primarily focus on label or textual hallucinations without manipulating the visual context, limiting their capacity to diagnose critical failures. In response, we introduce HalluSegBench, the first benchmark specifically designed to evaluate hallucinations in visual grounding through the lens of counterfactual visual reasoning. Our benchmark consists of a novel dataset of 1340 counterfactual instance pairs spanning 281 unique object classes, and a set of newly introduced metrics that quantify hallucination sensitivity under visually coherent scene edits. Experiments on HalluSegBench with state-of-the-art vision-language segmentation models reveal that vision-driven hallucinations are significantly more prevalent than label-driven ones, with models often persisting in false segmentation, highlighting the need for counterfactual reasoning to diagnose grounding fidelity.
Authors: Peibo Li, Shuang Ao, Hao Xue, Yang Song, Maarten de Rijke, Johan Barth\'elemy, Tomasz Bednarz, Flora D. Salim
Abstract: Large language models (LLMs) have been adopted for next point-of-interest (POI) recommendation tasks. Typical LLM-based recommenders fall into two categories: prompt-based and supervised fine-tuning (SFT)-based models. Prompt-based models generally offer greater output flexibility but deliver lower accuracy, whereas SFT-based models achieve higher performance yet face a fundamental mismatch: next POI recommendation data does not naturally suit supervised fine-tuning. In SFT, the model is trained to reproduce the exact ground truth, but each training example provides only a single target POI, so there is no ground truth for producing a top-k list. To address this, we propose Refine-POI, a reinforcement fine-tuning framework for next POI recommendation. We introduce recommendation-driven rewards that enable LLMs to learn to generate top-k recommendation lists using only one ground-truth POI per example. Experiments on real-world datasets demonstrate that Refine-POI achieves state-of-the-art top-k recommendation performance.