AI News

Silent Failures at the $2^{32}$ Boundary: A Technical Report on Large-Tensor Matrix Multiplication in PyTorch's Apple MPS Backend (arxiv.org)

Matched-Input Estimates Differ in Sign Across Architectures: Auditing EEG Foundation Models on Motor Imagery (arxiv.org)

Resist, Update, Reject: Preference Optimization Installs a Prior-Dependent Reliability Switch (arxiv.org)

A discrete generative model of neuronal spiking activity on microelectrode arrays (arxiv.org)

PSD: Pseudo Self-Distillation of Memory Representation Capabilities for LLM Agents (arxiv.org)

Multivariate quantile regression via Kolmogorov-Arnold Networks (arxiv.org)

GDN Tree-Scan: Served Tree Verification for Recurrent-Hybrid Language Models (arxiv.org)

Spatial-Interactor: Learning Spatial Reasoning through Interaction with the Observable Physical World (arxiv.org)

Reconstructed holograms and explanation-aware evaluation for low-cost computational pollen analysis in veterinary cytology (arxiv.org)

Common Cause, Not Cross-Attention: Blocking Visual Shortcuts in Audio-Video Generation (arxiv.org)

Contrastive Siamese Representation Learning for Predictive Maintenance of Electrical Submersible Pumps (arxiv.org)

SPeaR: Test-Time Adaptation with Steering Primitives for Realigning Representations (arxiv.org)

KV-COBRA: KV Cache Compression via Co-Optimized Bit-Rank Allocation (arxiv.org)

Streaming Deep Reinforcement Learning Finally Works (arxiv.org)

ESLM: Risk-Averse Selective Language Modeling for Efficient Pretraining (arxiv.org)

Memento No More: Coaching AI Agents to Master Multiple Tasks via Hints Internalization (arxiv.org)

Poisson Exchange Beyond Submodularity: Effective Approximation Algorithms for Offline and Online Subset Selection over Matroids (arxiv.org)

Scalable Incremental Robustness Analysis of Neural Network Feedback Systems (arxiv.org)

Do Student LLMs Inherit OOD Robustness? Invariance-Weighted Distillation for Reliable Knowledge Transfer (arxiv.org)

Logits are All We Need to Adapt Closed Models (arxiv.org)

Beyond the Raw Waveform: Fusing Visual Representations of EDA for Stress Detection (arxiv.org)

Token Signatures of Code: Comparing Coding Behaviors Across Large Language Models (arxiv.org)

Semantics Delivery Network: Rethinking Web Retrieval Infrastructure for LLM Agents (arxiv.org)

Defusing Explosive Prompts: Understanding and Preventing Trigger-Based Prompt Injections in LLM Agents (arxiv.org)

Rethinking Diffusion Segmentation: When Does It Rely on Its Noisy State, and Does Diffusion Matter? (arxiv.org)

A Mathematical Framework and a Suite of Learning Techniques for Neural-Symbolic Systems (arxiv.org)

Not All Task Vectors Need Equal Rank: Energy-Proportional Allocation for Model Merging (arxiv.org)

AutoGym: Blueprint-First Generation of Verifiable Agent Gyms (arxiv.org)

SPIBER: Reconstructing Free Energy Landscapes from Short, Unconverged Trajectories with Generative Flow Networks (arxiv.org)

Mask-Aware Execution for Efficient JEPA Training (arxiv.org)

StateMem: Single-State Residual Memory with Adaptive Inference for Vision-Language-Action Policies (arxiv.org)

Clinical Domain Classification from Medical Transcriptions (arxiv.org)

Towards Robust Classroom Attendance: A Comprehensive Evaluation of Face Detection and Recognition Models (arxiv.org)

Algorithmic Collusion and the Complexity of Information-Value-Free Equilibria (arxiv.org)

CTSpinoPelvic1K: spine, pelvis, ribs and femora in one coordinate frame, annotated for lumbosacral transitional anatomy (arxiv.org)

Beyond Final-Token Classification: Heterogeneous Readouts for Evidence-Grounded Suicide Risk Detection (arxiv.org)

ParA-LLM: A Unified Approach to Paralinguistic and Acoustic Speech Understanding (arxiv.org)

GraphToolbox: A Configurable Python Framework for Graph Neural Network Forecasting (arxiv.org)

Augmented Hypothesis Testing with Persona-Based LLM Simulations (arxiv.org)

iSDFT: Information-Proximal Self-Distillation for Continual Learning in LLMs (arxiv.org)

STEVE: Stabilizing Textual Gradient-Based Prompt Optimization via Error-Driven Refinement and Regularized Verification (arxiv.org)

Financial Language Models as Applied Artificial Intelligence Systems for News-Based Trading under Market Frictions (arxiv.org)

The shape of quark flavors (arxiv.org)

A Horizon-Independent Regret Bound for Optimistic Hedge in General-Sum Games (arxiv.org)

Marginal Calibration Does Not Compose: Hidden Dependence in Modular Robot Navigation (arxiv.org)

OnlineWM: Causality-Aware Active Online Learning for Effective World Modeling (arxiv.org)

TriFleetRCA: On-Premise LLM Root Cause Analysis for Kubernetes (arxiv.org)

SAGE: Optimal-Stopping Peer Selection for Decentralised Federated Learning (arxiv.org)

If You Hear It, Help Find It: Orthogonal Knowledge Distillation for Open-Vocabulary Audio-Visual Event Localization (arxiv.org)

One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents (arxiv.org)