AI News

Revisiting Simple Baselines for Relational Deep Learning (arxiv.org)

Generative Representation Learning for Scalable Latent Tokenization (arxiv.org)

The Physics Is the Network: Fast, Training-Free Neural Emulation (arxiv.org)

Forget About It: Mitigating Forgetting Without Replay or Auxiliary Objectives (arxiv.org)

Integrating AIRL-Based Reward Estimation into a Control Method Using Fuzzy Inference and Policy Gradient (arxiv.org)

Vela: User Behavior as Supervision Sparks LLM Intelligence (arxiv.org)

Lingdan-V2: Large language models for clinically grounded reasoning in traditional Chinese medicine (arxiv.org)

Towards Merging-conditional Backdoor Unalignment against Large Language Models (arxiv.org)

A Two-Level Clinical Novelty Prioritization Framework for Disease Trajectories Discovered from Longitudinal Electronic Health Records (arxiv.org)

Graph-Enhanced Disease Trajectory Discovery and Temporal Endpoint Forecasting from Longitudinal Electronic Health Records (arxiv.org)

Hindsight-Privileged RLVR: Single-Rollout Token Credit Without Group Sampling (arxiv.org)

Decoupled Marginal Sharpening for Training-Free Inference-Time Scaling (arxiv.org)

Selective Growing Q-Networks (SGQN): An Analysis of Adaptive Action Grids in Decoupled Q-Learning (arxiv.org)

Learning by Examples: Power Grid Steady State Analysis with a Tabular Foundation Model (arxiv.org)

Self-improving data synthesis can scale embedding models across domains and tasks (arxiv.org)

AlignAudio: A Unified Dual-Alignment Framework for Noise-Robust Speech-Conditioned Generation (arxiv.org)

DexCast: Learning Real-World Dexterous Control from Human Videos (arxiv.org)

Prospective Forcing: Learning to Look Ahead for Autoregressive Long Video Diffusion (arxiv.org)

EarthAlign: Learning Language-Aligned Representations of Earth System Data (arxiv.org)

The Accuracy Well: Nonmonotone Accuracy Dynamics at Vanishing Initialization (arxiv.org)

Rotate the Loop: Blockwise Conformality for Looped Transformers (arxiv.org)

Can We Trust This Evaluation? Instance-Level Reliability of LLM-as-a-Judge (arxiv.org)

Continual Learning for Reliable In-Sequence Molecular Communication (arxiv.org)

OrionMoE-v1.0: A Mixture-of-Experts Foundation Model for Tabular In-Context Learning (arxiv.org)

One Size Does Not Fit All: Curvature-Aware Adaptive Perturbation for SAM (arxiv.org)

FinGREAT: Evaluating LLMs on Financial Graph-guided Retrieval And Typing (arxiv.org)

Learning to Compute Is Not Learning to Retain or Reuse in Looped Transformers (arxiv.org)

CAST: History-Adaptive Spectral Clipping for LLM Pretraining (arxiv.org)

ProtModernBERT: Compute-Efficient Protein Language Modeling (arxiv.org)

RespFocus: Learning Temporal Evidence for Weakly Supervised Respiratory Sound Classification (arxiv.org)

TabMosaic: Unified In-Context Diffusion for Tabular Generative Tasks (arxiv.org)

Rethinking Human Intent-to-CAD: Parametric CAD Model Generation via Cooperative Multi-Task Alignment and Spatial-Aware Reinforcement Learning (arxiv.org)

Dynamic Conflict Clustering with Lightweight Learning and Strategic Repair for Distributed Constraint Satisfaction (arxiv.org)

ELECT: Settling Identity Before Rendering for Scalable Story Visualization (arxiv.org)

A Decade of PhotoMappers: A Longitudinal Study of Crowdsourced Disaster Photos Geolocalization with Vision-Language Models and Geospatial Reasoning (arxiv.org)

Counterfactual Regret for Reflective Agents with Lean-Verified Guarantees (arxiv.org)

ClawLoop: Self-Improving Agents via Harness-Native Training Loop (arxiv.org)

The Silent Failures in Agentic Data Science (arxiv.org)

Artificiality's First Exam: An LLM-Referenced C-test for Supraintelligence (arxiv.org)

Last Secure Code Benchmark: Is the Code You Vibe Secure? (arxiv.org)

Amortized ratio-estimation importance sampling and localized simulation-based calibration for intractable likelihoods (arxiv.org)

GLL: A Differentiable Graph Learning Layer for Neural Networks (arxiv.org)

Bellman Conformal Inference: Calibrating Prediction Intervals For Time Series (arxiv.org)

Principled Design of Expected Improvement for Continuous Simulation Optimization (arxiv.org)

The Statistical Benefits of Multiple Responses for Learning from Demonstrations (arxiv.org)

Testing Conditional Distribution Equality Using Generative Models (arxiv.org)

Tracking a Small Set of Experts in Online Convex Optimization (arxiv.org)

Fractional is Better: Learnable Derivative Orders in Neural Operator Learning (arxiv.org)

Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning (arxiv.org)

Beyond Expert Rankings: Contrast Calibration for Mixed-Precision MoE Quantization (arxiv.org)