AI News

In Defense of Information Leakage in Concept-based Models (arxiv.org)

How Does Reasoning Flow? Tracing Attention-Induced Information Flow for Targeted RL in LLMs (arxiv.org)

It\^o maps for any-step SDEs (arxiv.org)

ERAlign: Energy-based Representation Alignment of GNNs and LLMs on Text-attributed Graphs (arxiv.org)

Mitigating Bias in Low-SNR Financial Reinforcement Learning via Quantum Representations (arxiv.org)

SpenseGPT: Practical One-shot Pruning Enabling Sparse and Dense GEMMs for LLM Inference (arxiv.org)

Parallel Causal Associative Fields: Gated Sparse Memory for Long-Context Language Modeling (arxiv.org)

SHAPO: Sharpness-Aware Policy Optimization for Safe Exploration (arxiv.org)

Dual-Branch Gated Fusion for Open-Set Audio Deepfake Source Tracing (arxiv.org)

Fast Exact Nearest-Neighbor Learning for High-Frequency Financial Time Series (arxiv.org)

A Source Domain is All You Need: Source-Only Cross-OS Transfer Learning for APT Anomaly Detection via Semantic Alignment and Optimal Transport (arxiv.org)

Automated Pronunciation Evaluation for Korean Toddler Speech using Speech Diarization and Self-Supervised Learning (arxiv.org)

KG-SoftMAP: Soft Knowledge-Graph Priors for Bayesian Network Structure Learning from Sparse Discrete Data (arxiv.org)

Beyond Explaining Predictions: Logic-Based Explanations for Confidence in Machine Learning Models (arxiv.org)

Privacy-Preserving Credit Risk Prediction with Alternative Data (arxiv.org)

Rank Collapse, Fixed Points, and the Renormalization Group Structure of MLP Residual Networks (arxiv.org)

Dropout-GRPO: Variational Stochasticity for Continuous Latent Reasoning (arxiv.org)

Making Time Editable in Video Diffusion Transformers (arxiv.org)

Flow Control: Steering Vision-Language-Action Models with Simple Real-Time Inputs (arxiv.org)

Local Is Not a Sufficient Privacy Boundary: Governing OS-Integrated On-Device AI (arxiv.org)

Gaming AI-Assisted Peer Reviews Poses New Risks to the Scientific Community (arxiv.org)

Failure Modes of Deep Multi-Agent RL in Asynchronous Pricing: Reproducible Triggers, Trace Diagnostics, and a Partial Fix (arxiv.org)

Self-EmoQ: Plutchik-Guided Value-based Planning to Drive Streaming Emotional TTS (arxiv.org)

CollabSkill: Evaluating Human-Agent Collaboration On Real-World Tasks (arxiv.org)

Agentic Social Affordance Framework (ASAF): Agent Identity Design as a Collaboration Interface in Multi-Agent Systems (arxiv.org)

AI-Driven Analytics of Team-Teaching Talk: Acoustic Patterns across Experience, Cohorts and the Learning Design (arxiv.org)

TD-Grokking: Learning from Zero-Reward Problems by Training-Time Decomposition (arxiv.org)

Mitigating Manifold Departure: Uncertainty-Aware Subspace Rectification for Trustworthy MLLM Decoding (arxiv.org)

Uncertainty-aware Multi-fidelity Closure via Conditional Normalizing Flows (arxiv.org)

SynIB: Informational Bottleneck for Maximizing Synergy in Multimodal Learning (arxiv.org)

Representation-Aware Advantage Estimation: Your Reward Model Provides More Than A Scalar Output (arxiv.org)

Machine Learning Methods for Studying Latent Neural Activity Dynamics (arxiv.org)

Flexible Flows for Biological Sequence Design (arxiv.org)

Lost in Serialization: Invariance and Generalization of LLM Graph Reasoners (arxiv.org)

Towards Diverse Scientific Hypothesis Search with Large Language Models (arxiv.org)

Dirichlet-Guided Group Forecasting for Alleviating Over-smoothing in Time Series Forecasting (arxiv.org)

Embedding Hybrid Systems into Continuous Latent Vector Fields (arxiv.org)

Causal Ensemble Agent: Hierarchical Causal Discovery with LLM-guided Expert Reweighting (arxiv.org)

Geometry-Aware Reinforcement Learning for 2D Irregular Nesting (arxiv.org)

Catching One in Five: LLM-as-Judge Blind Spots in Production Multi-Turn Transaction Agents (arxiv.org)

Exploratory Responsiveness and Adaptive Rigidity under AI-Assisted Optimization (arxiv.org)

Predictive Assistance and the Temporal Dynamics of Exploratory Compression (arxiv.org)

From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs (arxiv.org)

IDP-Bench: Benchmarking ability of LLMs to protect personal information in interdependent privacy contexts (arxiv.org)

Fast and Highly Expressive Policy Learning for Offline Reinforcement Learning via Bootstrapped Flow Q-Learning (arxiv.org)

Is Fairness Truly Fair? Towards Reliable Lipschitz Fairness in Multi-Task Learning via Fixed-\texorpdfstring{$\delta$}{delta} Alignment (arxiv.org)

PhysMetrics.Weather: An Evaluation Framework for Physical Consistency in ML Weather Models (arxiv.org)

From Observation to Intervention: A Causal Audit of Expert Importance in Mixture-of-Experts Models (arxiv.org)

Bypassing Copyright Protection in Diffusion-based Customization via Two-Stage Latent Feature Optimization (arxiv.org)

Event-Driven Reinforcement Learning Enables Long-Horizon Control in Semiconductor Fabrication (arxiv.org)