Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Machine Learning

cs.LG Watch Machine Learning (sign in to use)

Sunday 25 January

Machine Learning

$\infty$-MoE: Generalizing Mixture of Experts to Infinite Experts

The Mixture of Experts (MoE) selects a few feed-forward networks (FFNs) per token, achieving an effective trade-off between computational cost and performance. In conventional MoE, each expert is treated as entirely independent, and experts are combined in a discrete space. As a result, when the number of experts increases, it becomes difficult to train each expert effectively. To stabilize training while increasing the number of experts, we propose $\infty$-MoE that selects a portion of the parameters of large FFNs based on continuous values sampled for each token. By considering experts in a continuous space, this approach allows for an infinite number of experts while maintaining computational efficiency. Experiments show that a GPT-2 Small-based $\infty$-MoE model, with 129M active and 186M total parameters, achieves comparable performance to a dense GPT-2 Medium with 350M parameters. Adjusting the number of sampled experts at inference time allows for a flexible trade-off between accuracy and speed, with an improvement of up to 2.5% in accuracy over conventional MoE.
Read the full abstract

All authors Shota Takashiro, Takeshi Kojima, Shohei Taniguchi, Yusuke Iwasawa, Yutaka Matsuo

Machine Learning · Cross-listed from cs.SD

BanglaRobustNet: A Hybrid Denoising-Attention Architecture for Robust Bangla Speech Recognition

Bangla, one of the most widely spoken languages, remains underrepresented in state-of-the-art automatic speech recognition (ASR) research, particularly under noisy and speaker-diverse conditions. This paper presents BanglaRobustNet, a hybrid denoising-attention framework built on Wav2Vec-BERT, designed to address these challenges. The architecture integrates a diffusion-based denoising module to suppress environmental noise while preserving Bangla-specific phonetic cues, and a contextual cross-attention module that conditions recognition on speaker embeddings for robustness across gender, age, and dialects. Trained end-to-end with a composite objective combining CTC loss, phonetic consistency, and speaker alignment, BanglaRobustNet achieves substantial reductions in word error rate (WER) and character error rate (CER) compared to Wav2Vec-BERT and Whisper baselines. Evaluations on Mozilla Common Voice Bangla and augmented noisy speech confirm the effectiveness of our approach, establishing BanglaRobustNet as a robust ASR system tailored to low-resource, noise-prone linguistic settings.
Read the full abstract

All authors Md Sazzadul Islam Ridoy, Mubaswira Ibnat Zidney, Sumi Akter, Md. Aminur Rahman

Machine Learning

Fast KVzip: Efficient and Accurate LLM Inference with Gated KV Eviction

Efficient key-value (KV) cache management is crucial for the practical deployment of large language models (LLMs), yet existing compression techniques often incur a trade-off between performance degradation and computational overhead. We propose a novel gating-based KV cache eviction method for frozen-weight LLMs that achieves high compression ratios with negligible computational cost. Our approach introduces lightweight sink-attention gating modules to identify and retain critical KV pairs, and integrates seamlessly into both the prefill and decoding stages. The proposed gate training algorithm relies on forward passes of an LLM, avoiding expensive backpropagation, while achieving strong task generalization through a task-agnostic reconstruction objective. Extensive experiments across the Qwen2.5-1M, Qwen3, and Gemma3 families show that our method maintains near-lossless performance while evicting up to 70% of the KV cache. The results are consistent across a wide range of tasks, including long-context understanding, code comprehension, and mathematical reasoning, demonstrating the generality of our approach.
Read the full abstract

Machine Learning

Entropic Risk-Aware Monte Carlo Tree Search

We propose a provably correct Monte Carlo tree search (MCTS) algorithm for solving risk-aware Markov decision processes (MDPs) with entropic risk measure (ERM) objectives. We provide a non-asymptotic analysis of our proposed algorithm, showing that the algorithm: (i) is correct in the sense that the empirical ERM obtained at the root node converges to the optimal ERM; and (ii) enjoys polynomial regret concentration. Our algorithm successfully exploits the dynamic programming formulations for solving risk-aware MDPs with ERM objectives introduced by previous works in the context of an upper confidence bound-based tree search algorithm. Finally, we provide a set of illustrative experiments to evaluate our risk-aware MCTS method, showing it successfully trades off between risk-neutral and risk-averse behavior.
Read the full abstract

All authors Pedro P. Santos, Jacopo Silvestrin, Manuel Lopes, Alberto Sardinha, Francisco S. Melo

Machine Learning

Kareus: Joint Reduction of Dynamic and Static Energy in Large Model Training

The computing demand of AI is growing at an unprecedented rate, but energy supply is not keeping pace. As a result, energy has become an expensive and contended resource that requires explicit management and optimization. Although recent works have made significant progress in large model training optimization, they focus on optimizing either dynamic or static energy consumption. We find that fine-grained kernel scheduling and frequency scaling jointly and interdependently impact both dynamic and static energy consumption. Based on this finding, we design Kareus, a training system that pushes the time-energy tradeoff frontier by optimizing both aspects. Kareus decomposes the intractable joint optimization problem into local, partition-based subproblems. It then uses a multi-pass multi-objective optimization algorithm to find execution schedules that push the time-energy tradeoff frontier. Compared to the state of the art, Kareus reduces training energy by up to 28.3% at the same training time, or reduces training time by up to 27.5% at the same energy consumption.
Read the full abstract

Machine Learning

Knowledge-Guided Time-Varying Causal Inference for Arctic Sea Ice Dynamics

Quantifying the causal relationship between sea ice thickness and sea surface height (SSH) is essential for understanding the mechanisms driving polar climate dynamics. Conventional deep learning models often struggle with treatment effect estimation in climate settings due to time-varying confounding and the lack of physical constraints. To address these challenges, we propose the Knowledge-Guided Causal Model Variational Autoencoder (KGCM-VAE) to quantify the effect of SSH on sea ice thickness. The framework leverages established physical relationships between SSH and surface velocity to generate physically grounded, time-varying continuous treatments, where each treatment value can change at every time step within a sequence. The model also incorporates Maximum Mean Discrepancy (MMD) to balance treated and control distributions in the latent space, mitigating observed confounding bias. Using synthetic data, we evaluated the model's ability to predict sea ice thickness responses under hypothetical SSH forcing scenarios, demonstrating that KGCM-VAE achieves lower PEHE compared to state-of-the-art baselines. Ablation studies further confirm that MMD consistently enhances treatment effect estimation over the base model. Additionally, we conducted a real-world case study to examine the sensitivity of sea ice thickness to SSH forcing and validate our findings against existing physical modeling results.
Read the full abstract

Machine Learning

A Mosco sufficient condition for intrinsic stability of non-unique convex Empirical Risk Minimization

Empirical risk minimization (ERM) stability is usually studied via single-valued outputs, while convex non-strict losses yield set-valued minimizers. We identify Painlevé-Kuratowski upper semicontinuity (PK-u.s.c.) as the intrinsic stability notion for the ERM solution correspondence (set-level Hadamard well-posedness) and a prerequisite to interpret stability of selections. We then characterize a minimal non-degenerate qualitative regime: Mosco-consistent perturbations and locally bounded minimizers imply PK-u.s.c., minimal-value continuity, and consistency of vanishing-gap near-minimizers. Quadratic growth yields explicit quantitative deviation bounds.
Read the full abstract

Machine Learning

RPNT: Robust Pre-trained Neural Transformer – A Pathway for Generalized Motor Decoding

Brain motor decoding aims to interpret and translate neural activity into behaviors. Decoding models should generalize across variations, such as recordings from different brain sites, experimental sessions, behavior types, and subjects, will be critical for real-world applications. Current decoding models only partially address these challenges. In this work, we develop a pretrained neural transformer model, RPNT - Robust Pretrained Neural Transformer, designed to achieve robust generalization through pretraining, which in turn enables effective finetuning for downstream motor decoding tasks. We achieved the proposed RPNT architecture by systematically investigating which transformer building blocks could be suitable for neural spike activity modeling, since components from models developed for other modalities, such as text and images, do not transfer directly to neural data. The final RPNT architecture incorporates three unique enabling components: 1) Multidimensional rotary positional embedding to aggregate experimental metadata such as site coordinates, session ids and behavior types; 2) Context-based attention mechanism via convolution kernels operating on global attention to learn local temporal structures for handling non-stationarity of neural population activity; 3) Robust self-supervised learning objective with stochastic causal masking strategies and contrastive representations. We pretrained two versions of RPNT on distinct datasets that present significant generalization challenges: a) Multi-session, multi-task, and multi-subject microelectrode benchmark; b) Multi-site recordings using high-density Neuropixel 1.0 probes from many cortical locations. After pretraining, we evaluated RPNT generalization on cross-session, cross-type, cross-subject, and cross-site downstream behavior decoding tasks. Our RPNT consistently outperforms the existing decoding models on these tasks.
Read the full abstract

All authors Hao Fang, Ryan A. Canfield, Tomohiro Ouchi, Beatrice Macagno, Eli Shlizerman, Amy L. Orsborn

Saturday 24 January

Machine Learning

Generating Counterfactual Patient Timelines from Real-World Data

Counterfactual simulation - exploring hypothetical consequences under alternative clinical scenarios - holds promise for transformative applications such as personalized medicine and in silico trials. However, it remains challenging due to methodological limitations. Here, we show that an autoregressive generative model trained on real-world data from over 300,000 patients and 400 million patient timeline entries can generate clinically plausible counterfactual trajectories. As a validation task, we applied the model to patients hospitalized with COVID-19 in 2023, modifying age, serum C-reactive protein (CRP), and serum creatinine to simulate 7-day outcomes. Increased in-hospital mortality was observed in counterfactual simulations with older age, elevated CRP, and elevated serum creatinine. Remdesivir prescriptions increased in simulations with higher CRP values and decreased in those with impaired kidney function. These counterfactual trajectories reproduced known clinical patterns. These findings suggest that autoregressive generative models trained on real-world data in a self-supervised manner can establish a foundation for counterfactual clinical simulation.
Read the full abstract

All authors Yu Akagi, Tomohisa Seki, Toru Takiguchi, Hiromasa Ito, Yoshimasa Kawazoe, Kazuhiko Ohe

Machine Learning · Cross-listed from cs.CV

Lost in Translation: How Language Re-Aligns Vision for Cross-Species Pathology

Foundation models are increasingly applied to computational pathology, yet their behavior under cross-cancer and cross-species transfer remains unspecified. This study investigated how fine-tuning CPath-CLIP affects cancer detection under same-cancer, cross-cancer, and cross-species conditions using whole-slide image patches from canine and human histopathology. Performance was measured using area under the receiver operating characteristic curve (AUC). Few-shot fine-tuning improved same-cancer (64.9% to 72.6% AUC) and cross-cancer performance (56.84% to 66.31% AUC). Cross-species evaluation revealed that while tissue matching enables meaningful transfer, performance remains below state-of-the-art benchmarks (H-optimus-0: 84.97% AUC), indicating that standard vision-language alignment is suboptimal for cross-species generalization. Embedding space analysis revealed extremely high cosine similarity (greater than 0.99) between tumor and normal prototypes. Grad-CAM shows prototype-based models remain domain-locked, while language-guided models attend to conserved tumor morphology. To address this, we introduce Semantic Anchoring, which uses language to provide a stable coordinate system for visual features. Ablation studies reveal that benefits stem from the text-alignment mechanism itself, regardless of text encoder complexity. Benchmarking against H-optimus-0 shows that CPath-CLIP's failure stems from intrinsic embedding collapse, which text alignment effectively circumvents. Additional gains were observed in same-cancer (8.52%) and cross-cancer classification (5.67%). We identified a previously uncharacterized failure mode: semantic collapse driven by species-dominated alignment rather than missing visual information. These results demonstrate that language acts as a control mechanism, enabling semantic re-interpretation without retraining.
Read the full abstract

Machine Learning

Dynamic Adversarial Reinforcement Learning for Robust Multimodal Large Language Models

Despite their impressive capabilities, Multimodal Large Language Models (MLLMs) exhibit perceptual fragility when confronted with visually complex scenes. This weakness stems from a reliance on finite training datasets, which are prohibitively expensive to scale and impose a ceiling on model robustness. We introduce AOT-SFT, a large-scale adversarial dataset for bootstrapping MLLM robustness. Building on this, we propose AOT (Adversarial Opponent Training), a self-play framework that forges MLLM robustness by creating its own training data. Our method orchestrates a co-evolution between an image-editing Attacker and a Defender MLLM, where the Attacker generates a diverse and dynamic curriculum of image manipulations, forcing the Defender to adapt and improve. Extensive experiments demonstrate that AOT enhances the Defender's perceptual robustness and reduces hallucinations, establishing a scalable paradigm for training more reliable MLLMs.
Read the full abstract

All authors Yicheng Bao, Xuhong Wang, Qiaosheng Zhang, Chaochao Lu, Xia Hu, Xin Tan

Machine Learning

Kalman-Inspired Runtime Stability and Recovery in Hybrid Reasoning Systems

Hybrid reasoning systems that combine learned components with model-based inference are increasingly deployed in tool-augmented decision loops, yet their runtime behavior under partial observability and sustained evidence mismatch remains poorly understood. In practice, failures often arise as gradual divergence of internal reasoning dynamics rather than as isolated prediction errors. This work studies runtime stability in hybrid reasoning systems from a Kalman-inspired perspective. We model reasoning as a stochastic inference process driven by an internal innovation signal and introduce cognitive drift as a measurable runtime phenomenon. Stability is defined in terms of detectability, bounded divergence, and recoverability rather than task-level correctness. We propose a runtime stability framework that monitors innovation statistics, detects emerging instability, and triggers recovery-aware control mechanisms. Experiments on multi-step, tool-augmented reasoning tasks demonstrate reliable instability detection prior to task failure and show that recovery, when feasible, re-establishes bounded internal behavior within finite time. These results emphasize runtime stability as a system-level requirement for reliable reasoning under uncertainty.
Read the full abstract

Machine Learning · Cross-listed from cs.CL

Building Safe and Deployable Clinical Natural Language Processing under Temporal Leakage Constraints

Clinical natural language processing (NLP) models have shown promise for supporting hospital discharge planning by leveraging narrative clinical documentation. However, note-based models are particularly vulnerable to temporal and lexical leakage, where documentation artifacts encode future clinical decisions and inflate apparent predictive performance. Such behavior poses substantial risks for real-world deployment, where overconfident or temporally invalid predictions can disrupt clinical workflows and compromise patient safety. This study focuses on system-level design choices required to build safe and deployable clinical NLP under temporal leakage constraints. We present a lightweight auditing pipeline that integrates interpretability into the model development process to identify and suppress leakage-prone signals prior to final training. Using next-day discharge prediction after elective spine surgery as a case study, we evaluate how auditing affects predictive behavior, calibration, and safety-relevant trade-offs. Results show that audited models exhibit more conservative and better-calibrated probability estimates, with reduced reliance on discharge-related lexical cues. These findings emphasize that deployment-ready clinical NLP systems should prioritize temporal validity, calibration, and behavioral robustness over optimistic performance.
Read the full abstract

All authors Ha Na Cho, Sairam Sutari, Alexander Lopez, Hansen Bow, Kai Zheng

Machine Learning

Denoising diffusion networks for normative modeling in neuroimaging

Normative modeling estimates reference distributions of biological measures conditional on covariates, enabling centiles and clinically interpretable deviation scores to be derived. Most neuroimaging pipelines fit one model per imaging-derived phenotype (IDP), which scales well but discards multivariate dependence that may encode coordinated patterns. We propose denoising diffusion probabilistic models (DDPMs) as a unified conditional density estimator for tabular IDPs, from which univariate centiles and deviation scores are derived by sampling. We utilise two denoiser backbones: (i) a feature-wise linear modulation (FiLM) conditioned multilayer perceptron (MLP) and (ii) a tabular transformer with feature self-attention and intersample attention (SAINT), conditioning covariates through learned embeddings. We evaluate on a synthetic benchmark with heteroscedastic and multimodal age effects and on UK Biobank FreeSurfer phenotypes, scaling from dimension of 2 to 200. Our evaluation suite includes centile calibration (absolute centile error, empirical coverage, and the probability integral transform), distributional fidelity (Kolmogorov-Smirnov tests), multivariate dependence diagnostics, and nearest-neighbour memorisation analysis. For low dimensions, diffusion models deliver well-calibrated per-IDP outputs comparable to traditional baselines while jointly modeling realistic dependence structure. At higher dimensions, the transformer backbone remains substantially better calibrated than the MLP and better preserves higher-order dependence, enabling scalable joint normative models that remain compatible with standard per-IDP pipelines. These results support diffusion-based normative modeling as a practical route to calibrated multivariate deviation profiles in neuroimaging.
Read the full abstract

Machine Learning

GraphDancer: Training LLMs to Explore and Reason over Graphs via Curriculum Reinforcement Learning

Large language models (LLMs) increasingly rely on external knowledge to improve factuality, yet many knowledge sources on the Web, such as scholarly indexes and product catalogs, are organized as heterogeneous graphs. Reasoning over such graph-structured knowledge poses two key challenges: (1) navigating structured, schema-defined relations requires precise function calls beyond similarity-based retrieval, and (2) answering complex questions often demands multi-hop evidence aggregation through iterative information seeking. We study how LLMs can learn these interaction capabilities and transfer them to unseen domains. We propose GraphDancer, a reinforcement learning (RL) framework that teaches LLMs to navigate graphs by interleaving reasoning and function execution. To improve RL for moderate-sized LLMs, we introduce a graph-aware curriculum that schedules training by the structural difficulty of questions using an easy-to-hard biased sampler. We evaluate GraphDancer on the multi-domain GRBench benchmark, training only on Academic and Biomedical graphs. Testing covers held-out questions from these domains as well as unseen E-commerce, Literature, and Legal graphs. With either Qwen2.5-3B-Instruct or Qwen3-14B as the backbone, GraphDancer outperforms graph-agent baselines using the same backbone across all evaluation domains. These results indicate that the learned graph exploration and reasoning capabilities transfer across domains. On average across unseen domains, GraphDancer with a 14B backbone also outperforms baselines powered by the closed-source models GPT-4.1-mini and GPT-5-nano. Trajectory analyses further show that GraphDancer retrieves answer-bearing evidence more often and achieves higher accuracy when such evidence is found, while substantially reducing invalid-protocol outcomes.
Read the full abstract

All authors Yuyang Bai, Zhuofeng Li, Ping Nie, Jianwen Xie, Yu Wang, Yu Zhang

Machine Learning

What Drives Length of Stay After Elective Spine Surgery? Insights from a Decade of Predictive Modeling

Objective: Predicting length of stay after elective spine surgery is essential for optimizing patient outcomes and hospital resource use. This systematic review synthesizes computational methods used to predict length of stay in this patient population, highlighting model performance and key predictors. Methods: Following PRISMA guidelines, we systematically searched PubMed, Google Scholar, and ACM Digital Library for studies published between December 1st, 2015, and December 1st, 2024. Eligible studies applied statistical or machine learning models to predict length of stay for elective spine surgery patients. Three reviewers independently screened studies and extracted data. Results: Out of 1,263 screened studies, 29 studies met inclusion criteria. Length of stay was predicted as a continuous, binary, or percentile-based outcome. Models included logistic regression, random forest, boosting algorithms, and neural networks. Machine learning models consistently outperformed traditional statistical models, with AUCs ranging from 0.94 to 0.99. K-Nearest Neighbors and Naive Bayes achieved top performance in some studies. Common predictors included age, comorbidities (notably hypertension and diabetes), BMI, type and duration of surgery, and number of spinal levels. However, external validation and reporting practices varied widely across studies. Discussion: There is growing interest in artificial intelligence and machine learning in length of stay prediction, but lack of standardization and external validation limits clinical utility. Future studies should prioritize standardized outcome definitions and transparent reporting needed to advance real-world deployment. Conclusion: Machine learning models offer strong potential for length of stay prediction after elective spine surgery, highlighting their potential for improving discharge planning and hospital resource management.
Read the full abstract

All authors Ha Na Cho, Seungmin Jeong, Yawen Guo, Alexander Lopez, Hansen Bow, Kai Zheng

Machine Learning · Cross-listed from eess.IV

Radiomics in Medical Imaging: Methods, Applications, and Challenges

Radiomics enables quantitative medical image analysis by converting imaging data into structured, high-dimensional feature representations for predictive modeling. Despite methodological developments and encouraging retrospective results, radiomics continue to face persistent challenges related to feature instability, limited reproducibility, validation bias, and restricted clinical translation. Existing reviews largely focus on application-specific outcomes or isolated pipeline components, with limited analysis of how interdependent design choices across acquisition, preprocessing, feature engineering, modeling, and evaluation collectively affect robustness and generalizability. This survey provides an end-to-end analysis of radiomics pipelines, examining how methodological decisions at each stage influence feature stability, model reliability, and translational validity. This paper reviews radiomic feature extraction, selection, and dimensionality reduction strategies; classical machine and deep learning-based modeling approaches; and ensemble and hybrid frameworks, with emphasis on validation protocols, data leakage prevention, and statistical reliability. Clinical applications are discussed with a focus on evaluation rigor rather than reported performance metrics. The survey identifies open challenges in standardization, domain shift, and clinical deployment, and outlines future directions such as hybrid radiomics-artificial intelligence models, multimodal fusion, federated learning, and standardized benchmarking.
Read the full abstract

Machine Learning

Gauss-Newton Natural Gradient Descent for Shape Learning

We explore the use of the Gauss-Newton method for optimization in shape learning, including implicit neural surfaces and geometry-informed neural networks. The method addresses key challenges in shape learning, such as the ill-conditioning of the underlying differential constraints and the mismatch between the optimization problem in parameter space and the function space where the problem is naturally posed. This leads to significantly faster and more stable convergence than standard first-order methods, while also requiring far fewer iterations. Experiments across benchmark shape optimization tasks demonstrate that the Gauss-Newton method consistently improves both training speed and final solution accuracy.
Read the full abstract

All authors James King, Arturs Berzins, Siddhartha Mishra, Marius Zeinhofer

Machine Learning

NCSAM Noise-Compensated Sharpness-Aware Minimization for Noisy Label Learning

Learning from Noisy Labels (LNL) remains a fundamental challenge in deep learning because real-world datasets often contain corrupted annotations. Most existing methods rely on label correction or sample selection mechanisms. In contrast, we study LNL from an optimization perspective by establishing a theoretical connection between label noise and the flatness-seeking behavior of Sharpness-Aware Minimization (SAM). Based on this analysis, we propose Noise-Compensated Sharpness-Aware Minimization (NCSAM), which uses a noise-compensated perturbation to counteract the optimization bias induced by noisy labels. By correcting distorted SAM perturbations, NCSAM mitigates the memorization of noisy labels during training while preserving the simplicity of optimization-based learning. Experiments on synthetic and real-world noisy-label benchmarks show that NCSAM consistently improves over SAM-based optimization baselines and remains competitive with representative noisy-label learning methods.
Read the full abstract

Machine Learning · Cross-listed from eess.AS

MK-SGC-SC: Multiple Kernel Guided Sparse Graph Construction in Spectral Clustering for Unsupervised Speaker Diarization

Speaker diarization aims to segment audio recordings into regions corresponding to individual speakers. Although unsupervised speaker diarization is inherently challenging, the prospect of identifying speaker regions without pretraining or weak supervision motivates research on clustering techniques. In this work, we share the notable observation that measuring multiple kernel similarities of speaker embeddings to thereafter craft a sparse graph for spectral clustering in a principled manner is sufficient to achieve state-of-the-art performances in a fully unsupervised setting. Specifically, we consider four polynomial kernels and a degree one arccosine kernel to measure similarities in speaker embeddings, using which sparse graphs are constructed in a principled manner to emphasize local similarities. Experiments show the proposed approach excels in unsupervised speaker diarization over a variety of challenging environments in the DIHARD-III, AMI, and VoxConverse corpora. To encourage further research, our implementations are available at https://github.com/nikhilraghav29/MK-SGC-SC.
Read the full abstract

All authors Nikhil Raghav, Avisek Gupta, Swagatam Das, Md Sahidullah

Machine Learning

BrainDistill: Implantable Motor Decoding with Task-Specific Knowledge Distillation

Transformer-based neural decoders with large parameter counts, pre-trained on large-scale datasets, have recently outperformed classical machine learning models and small neural networks on brain-computer interface (BCI) tasks. However, their large parameter counts and high computational demands hinder deployment in power-constrained implantable systems. To address this challenge, we introduce BrainDistill, a novel implantable motor decoding pipeline that integrates an implantable neural decoder (IND) with a task-specific knowledge distillation (TSKD) framework. Unlike standard feature distillation methods that attempt to preserve teacher representations in full, TSKD explicitly prioritizes features critical for decoding through supervised projection. Across multiple neural datasets, IND consistently outperforms prior neural decoders on motor decoding tasks, while its TSKD-distilled variant further surpasses alternative distillation methods in few-shot calibration settings. Finally, we present a quantization-aware training scheme that enables integer-only inference with activation clipping ranges learned during training. The quantized IND enables deployment under the strict power constraints of implantable BCIs with minimal performance loss.
Read the full abstract

All authors Yuhan Xie, Jinhan Liu, Xiaoyong Ni, Fei Tan, Icare Sakr, Thibault Collin, Shiqi Sun, Alejandro Rodriguez Guajardo, Demon Fanny, Charles-francois Vincent Latchoumane, Henri Lorach, Jocelyne Bloch, Gregoire Courtine, Mahsa Shoaran

Machine Learning

Sparse Subspace-to-Expert Sharing for Task-Agnostic Continual Learning

Continual learning in Large Language Models (LLMs) is hindered by the plasticity-stability dilemma, where acquiring new capabilities often leads to catastrophic forgetting of previous knowledge. Existing methods typically treat parameters uniformly, failing to distinguish between specific task knowledge and shared capabilities. We introduce Mixture of Sparse Experts for Task-Agnostic Continual Learning (SETA), a framework that resolves the plasticity-stability conflict through adaptive sparse subspace decomposition into task-specific expert modules. Unlike standard updates, where tasks compete for the same parameters, SETA separates knowledge into unique experts, designed to isolate task-specific patterns, and shared experts, responsible for capturing common features. SETA shares an overlapping parameter block only when the new task's update passes a subspace-agreement test and a sign test against its prior expert; shared knowledge is protected by adaptive elastic anchoring, and independently gated routers retrieve the relevant experts at inference without task identifiers. Extensive experiments across diverse domain-specific benchmarks demonstrate that SETA achieves competitive or superior overall performance relative to state-of-the-art continual learning baselines, improving overall performance over the strongest baseline by 1.1 to 1.7 points on TRACE and by up to 5.5 points on the O-LoRA benchmark, with backward transfer between -2.2 and -1.3 on TRACE, against 4.6 to 6.3 points of forgetting for MoE-CL and 8 to 15 points for the other LoRA baselines.
Read the full abstract

All authors Fatema Siddika, Md Anwar Hossen, Tanwi Mallick, Ali Jannesari

Machine Learning · Cross-listed from cs.AR

Athena: Synergizing Data Prefetching and Off-Chip Prediction via Online Reinforcement Learning

Prefetching and off-chip prediction are two techniques proposed to hide long memory access latencies in high-performance processors. In this work, we demonstrate that: (1) prefetching and off-chip prediction often provide complementary performance benefits, yet (2) naively combining them often fails to realize their full performance potential, and (3) existing prefetcher control policies leave significant room for performance improvement behind. Our goal is to design a holistic framework that can autonomously learn to coordinate an off-chip predictor with multiple prefetchers employed at various cache levels. To this end, we propose a new technique called Athena, which models the coordination between prefetchers and off-chip predictor (OCP) as a reinforcement learning (RL) problem. Athena acts as the RL agent that observes multiple system-level features (e.g., prefetcher/OCP accuracy, bandwidth usage) over an epoch of program execution, and uses them as state information to select a coordination action (i.e., enabling the prefetcher and/or OCP, and adjusting prefetcher aggressiveness). At the end of every epoch, Athena receives a numerical reward that measures the change in multiple system-level metrics (e.g., number of cycles taken to execute an epoch). Athena uses this reward to autonomously and continuously learn a policy to coordinate prefetchers with OCP. Our extensive evaluation using a diverse set of memory-intensive workloads shows that Athena consistently outperforms prior state-of-the-art coordination policies across a wide range of system configurations with various combinations of underlying prefetchers, OCPs, and main memory bandwidths, while incurring only modest storage overhead. Athena is freely available at https://github.com/CMU-SAFARI/Athena.
Read the full abstract

All authors Rahul Bera, Zhenrong Lang, Caroline Hengartner, Konstantinos Kanellopoulos, Rakesh Kumar, Mohammad Sadrosadati, Onur Mutlu

Machine Learning · Cross-listed from eess.AS

ToS: A Team of Specialists ensemble framework for Stereo Sound Event Localization and Detection with distance estimation in Video

Sound event localization and detection with distance estimation (3D SELD) in video involves identifying active sound events at each time frame while estimating their spatial coordinates. This multimodal task requires joint reasoning across semantic, spatial, and temporal dimensions, a challenge that single models often struggle to address effectively. To tackle this, we introduce the Team of Specialists (ToS) ensemble framework, which integrates three complementary sub-networks: a spatio-linguistic model, a spatio-temporal model, and a tempo-linguistic model. Each sub-network specializes in a unique pair of dimensions, contributing distinct insights to the final prediction, akin to a collaborative team with diverse expertise. ToS has been benchmarked against state-of-the-art audio-visual models for 3D SELD on the DCASE2025 Task 3 Stereo SELD development set, consistently outperforming existing methods across key metrics. Future work will extend this proof of concept by strengthening the specialists with appropriate tasks, training, and pre-training curricula.
Read the full abstract

Machine Learning

A Thermodynamic Theory of Learning I: Irreversible Ensemble Transport and Epistemic Costs

Learning systems acquire structured internal representations from data, yet classical information-theoretic results state that deterministic transformations do not increase information. This raises a fundamental question: how can learning produce abstraction and insight without violating information-theoretic limits? We argue that learning is inherently an irreversible process when performed over finite time, and that the realization of epistemic structure necessarily incurs entropy production. To formalize this perspective, we model learning as a transport process in the space of probability distributions over model configurations and introduce an epistemic free-energy framework. Within this framework, we define the free-energy reduction as a bookkeeping quantity that records the total reduction of epistemic free energy along a learning trajectory. This formulation highlights that realizing such a reduction over finite time necessarily incurs irreversible entropy production. We then derive the Epistemic Speed Limit (ESL), a finite-time inequality that lower-bounds the minimal entropy production required by any learning process to realize a given distributional transformation. This bound depends only on the Wasserstein distance between initial and final ensemble distributions and is independent of the specific learning algorithm.
Read the full abstract

Machine Learning

Understanding Transformer Encoder-Decoder Representations through Bernoulli Dropout

We study Transformer overparameterization through the lens of angular similarity in high-dimensional encoder-decoder embeddings. We apply Bernoulli dropout between the encoder and the decoder, varying the keep probability $p$ to identify a sparsity-dependent threshold above which the Top-1 prediction is preserved. Theoretically, we prove that, if the effective sparsity embeddings is sufficiently large, and thus decoder performance, remain stable under moderate coordinate dropout. Empirically, we implement the Bernoulli dropout by constructing a new Transformer model augmented with Binary Erasure Channel (BEC) and test its performance on an English-French translation task. Experimental results visualize the trends for validation accuracies and BLEU scores, both decline sharply at some threshold.
Read the full abstract

Machine Learning · Cross-listed from cs.NE

Deep Intrinsic Surprise-Regularized Control (DISRC): A Biologically Inspired Mechanism for Efficient Deep Q-Learning in Sparse Environments

Deep reinforcement learning (DRL) has driven major advances in autonomous control. Still, standard Deep Q-Network (DQN) agents tend to rely on fixed learning rates and uniform update scaling, even as updates are modulated by temporal-difference (TD) error. This rigidity destabilizes convergence, especially in sparse-reward settings where feedback is infrequent. We introduce Deep Intrinsic Surprise-Regularized Control (DISRC), a biologically inspired augmentation to DQN that dynamically scales Q-updates based on latent-space surprise. DISRC encodes states via a LayerNorm-based encoder and computes a deviation-based surprise score relative to a moving latent setpoint. Each update is then scaled in proportion to both TD error and surprise intensity, promoting plasticity during early exploration and stability as familiarity increases. We evaluate DISRC on two sparse-reward MiniGrid environments, which included MiniGrid-DoorKey-8x8 and MiniGrid-LavaCrossingS9N1, under identical settings as a vanilla DQN baseline. In DoorKey, DISRC reached the first successful episode (reward > 0.8) 33% faster than the vanilla DQN baseline (79 vs. 118 episodes), with lower reward standard deviation (0.25 vs. 0.34) and higher reward area under the curve (AUC: 596.42 vs. 534.90). These metrics reflect faster, more consistent learning - critical for sparse, delayed reward settings. In LavaCrossing, DISRC achieved a higher final reward (0.95 vs. 0.93) and the highest AUC of all agents (957.04), though it converged more gradually. These preliminary results establish DISRC as a novel mechanism for regulating learning intensity in off-policy agents, improving both efficiency and stability in sparse-reward domains. By treating surprise as an intrinsic learning signal, DISRC enables agents to modulate updates based on expectation violations, enhancing decision quality when conventional value-based methods fall short.
Read the full abstract

Machine Learning · Cross-listed from cs.AI

Discovery of Feasible 3D Printing Configurations for Metal Alloys via AI-driven Adaptive Experimental Design

Configuring the parameters of additive manufacturing processes for metal alloys is a challenging problem due to complex relationships between input parameters (e.g., laser power, scan speed) and quality of printed outputs. The standard trial-and-error approach to find feasible parameter configurations is highly inefficient because validating each configuration is expensive in terms of resources (physical and human labor) and the configuration space is very large. This paper combines the general principles of AI-driven adaptive experimental design with domain knowledge to address the challenging problem of discovering feasible configurations. The key idea is to build a surrogate model from past experiments to intelligently select a small batch of input configurations for validation in each iteration. To demonstrate the effectiveness of this methodology, we deploy it for Directed Energy Deposition process to print GRCop–42, a high-performance copper–chromium–niobium alloy developed by NASA for aerospace applications. Within three months, our approach yielded multiple defect-free outputs across a range of laser powers dramatically reducing time to result and resource expenditure compared to several months of manual experimentation by domain scientists with no success. By enabling high-quality GRCop–42 fabrication on readily available infrared laser platforms for the first time, we democratize access to this critical alloy, paving the way for cost-effective, decentralized production for aerospace applications.
Read the full abstract

All authors Azza Fadhel, Nathaniel W. Zuckschwerdt, Aryan Deshwal, Susmita Bose, Amit Bandyopadhyay, Jana Doppa

Machine Learning · Cross-listed from cs.CV

Stylizing ViT: Anatomy-Preserving Instance Style Transfer for Domain Generalization

Deep learning models in medical image analysis often struggle with generalizability across domains and demographic groups due to data heterogeneity and scarcity. Traditional augmentation improves robustness, but fails under substantial domain shifts. Recent advances in stylistic augmentation enhance domain generalization by varying image styles but fall short in terms of style diversity or by introducing artifacts into the generated images. To address these limitations, we propose Stylizing ViT, a novel Vision Transformer encoder that utilizes weight-shared attention blocks for both self- and cross-attention. This design allows the same attention block to maintain anatomical consistency through self-attention while performing style transfer via cross-attention. We assess the effectiveness of our method for domain generalization by employing it for data augmentation on three distinct image classification tasks in the context of histopathology and dermatology. Results demonstrate an improved robustness (up to +13% accuracy) over the state of the art while generating perceptually convincing images without artifacts. Additionally, we show that Stylizing ViT is effective beyond training, achieving a 17% performance improvement during inference when used for test-time augmentation. The source code is available at https://github.com/sdoerrich97/stylizing-vit .
Read the full abstract

All authors Sebastian Doerrich, Francesco Di Salvo, Jonas Alle, Christian Ledig

Machine Learning · Cross-listed from q-bio.QM

GenAI-Net: A Generative AI Framework for Automated Biomolecular Network Design

Biomolecular networks underpin emerging technologies in synthetic biology-from robust biomanufacturing and metabolic engineering to smart therapeutics and cell-based diagnostics-and also provide a mechanistic language for understanding complex dynamics in natural and ecological systems. Yet designing chemical reaction networks (CRNs) that implement a desired dynamical function remains largely manual: while a proposed network can be checked by simulation, the reverse problem of discovering a network from a behavioral specification is difficult, requiring substantial human insight to navigate a vast space of topologies and kinetic parameters with nonlinear and possibly stochastic dynamics. Here we introduce GenAI-Net, a generative AI framework that automates CRN design by coupling an agent that proposes reactions to simulation-based evaluation defined by a user-specified objective. GenAI-Net efficiently produces novel, topologically diverse solutions across multiple design tasks, including dose responses, complex logic gates, classifiers, oscillators, and robust perfect adaptation in deterministic and stochastic settings (including noise reduction). By turning specifications into families of circuit candidates and reusable motifs, GenAI-Net provides a general route to programmable biomolecular circuit design and accelerates the translation from desired function to implementable mechanisms.
Read the full abstract

All authors Maurice Filo, Nicolò Rossi, Zhou Fang, Mustafa Khammash