AI News

On the Collapse of Generative Paths: A Criterion and Correction for Diffusion Steering (arxiv.org)

Boosting RL-Based Visual Reasoning with Selective Adversarial Entropy Intervention (arxiv.org)

PolarMem: A Training-Free Polarized Latent Graph Memory for Verifiable Vision-Language Models (arxiv.org)

Structure Enables Effective Self-Localization of Errors in LLMs (arxiv.org)

Breaking the Reversal Curse in Autoregressive Language Models via Identity Bridge (arxiv.org)

Beyond End-to-End Video Models: An LLM-Based Multi-Agent System for Educational Video Generation (arxiv.org)

Prototype Transformer: Towards Language Model Architectures Interpretable by Design (arxiv.org)

REAL: Resolving Knowledge Conflicts in Knowledge-Intensive Visual Question Answering via Reasoning-Pivot Alignment (arxiv.org)

When AI Benchmarks Plateau: A Systematic Study of Benchmark Saturation (arxiv.org)

Retrieval-aligned Tabular Foundation Models Enable Robust Clinical Risk Prediction in Electronic Health Records Under Real-world Constraints (arxiv.org)

Perspective on Bias in Biomedical AI: Preventing Downstream Healthcare Disparities (arxiv.org)

RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography (arxiv.org)

ANDRE: An Attention-based Neuro-symbolic Differentiable Rule Extractor for Inductive Logic Programming (arxiv.org)

UrbanFusion: Stochastic Multimodal Fusion for Contrastive Learning of Robust Spatial Representations (arxiv.org)

Tractable Shapley Values and Interactions via Tensor Networks (arxiv.org)

Symbolic Neural Generation with Applications to Lead Discovery in Drug Design (arxiv.org)

DuetServe: Harmonizing Prefill and Decode for LLM Serving via Adaptive GPU Multiplexing (arxiv.org)

Scaling depth capacity via zero/one-layer model expansion (arxiv.org)

FIRM: Federated In-client Regularized Multi-objective Alignment for Large Language Models (arxiv.org)

WUSH: Near-Optimal Adaptive Transforms for LLM Quantization (arxiv.org)

Position: Current Benchmarking Hinders Real Progress in Deep Learning for Time Series Forecasting (arxiv.org)

Early Prediction of Liver Cirrhosis Up to Two Years in Advance: A Machine Learning Study Benchmarking Against the FIB-4 and APRI Scores (arxiv.org)

When Does Predictive Inverse Dynamics Outperform Behavior Cloning? (arxiv.org)

Avatar Forcing: Real-Time Interactive Head Avatar Generation for Natural Conversation (arxiv.org)

From Zero to Hero: Advancing Zero-Shot Foundation Models for Tabular Outlier Detection (arxiv.org)

Robust Learning of a Group DRO Neuron (arxiv.org)

Multi-Objective Reinforcement Learning for Tactical Decision Making for Trucks in Highway Traffic (arxiv.org)

Consistency Deep Equilibrium Models (arxiv.org)

SurrogateSHAP: Training-Free Contributor Attribution for Text-to-Image (T2I) Models (arxiv.org)

Knowledge-Informed Kernel State Reconstruction from Heterogeneous Partial Observations (arxiv.org)

GUDA: Counterfactual Group-wise Training Data Attribution for Diffusion Models via Unlearning (arxiv.org)

Adaptive Exploration for Latent-State Bandits (arxiv.org)

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation (arxiv.org)

LRAgent: Efficient KV Cache Sharing for Multi-LoRA LLM Agents (arxiv.org)

Zero-Shot Off-Policy Learning (arxiv.org)

Probabilistic Performance Guarantees for Multi-Task Reinforcement Learning (arxiv.org)

Interpretability in Deep Time Series Models Demands Semantic Alignment (arxiv.org)

naPINN: Noise-Adaptive Physics-Informed Neural Networks for Recovering Physics from Corrupted Measurement (arxiv.org)

Beware of the Batch Size: Hyperparameter Bias in Evaluating LoRA (arxiv.org)

The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer (arxiv.org)

Exposing Vulnerabilities in Explanation for Time Series Classifiers via Dual-Target Attacks (arxiv.org)

Lookahead Sample Reward Guidance for Test-Time Scaling of Diffusion Models (arxiv.org)

When Single Answer Is Not Enough: Rethinking Single-Step Retrosynthesis Benchmarks for LLMs (arxiv.org)

Equilibrium Propagation for Non-Conservative Systems (arxiv.org)

From Evaluation to Design: Using Potential Energy Surface Smoothness Metrics to Guide Machine Learning Interatomic Potential Architectures (arxiv.org)

Tempora: Characterising the Time-Contingent Utility of Online Test-Time Adaptation (arxiv.org)

Principle-Evolvable Scientific Discovery via Uncertainty Minimization (arxiv.org)

Vegas: Self-Speculative Decoding with Verification-Guided Sparse Attention (arxiv.org)

DeepIPCv2: LiDAR-powered Robust Environmental Perception and Navigational Control for Autonomous Vehicle (arxiv.org)

Recent Advances in Multi-modal 3D Intelligence: A Comprehensive Survey and Evaluation (arxiv.org)