AI News

Event-Centric Earthquake Monitoring with a Multi-Station Transformer (arxiv.org)

Learning to Detect Shifts in Unstructured Data Streams via Monitoring-by-Betting (arxiv.org)

Mitigating Black-Box Score-Based Query Attacks via Gradient Confusion (arxiv.org)

MRFlow: Mechanism-Structured Residual Flow Matching for Virtual Cell Perturbation Prediction (arxiv.org)

Dependency, Compression, and Synergy: A Unified Information-Theoretic View of Multimodal Learning (arxiv.org)

Selective Knowledge Control for Continual GUI Agent Learning over Application Streams (arxiv.org)

AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewards (arxiv.org)

MFG: Multi-turn Feedback Guided Molecular Optimization (arxiv.org)

Who Improves the Robot? Toward Physical Recursive Self-Improvement (arxiv.org)

Read the Room Before Your Model Generates (arxiv.org)

Self-Evolving Agent Harness (arxiv.org)

Cross-Market Meta-Learning for Electricity Price Forecasting (arxiv.org)

Pack Light, CarryOn: Building Object-Oriented Context for Long-Horizon Agents (arxiv.org)

BLADE: Auditing Information Use for Failure Attribution in LLM Multi-Agent Systems (arxiv.org)

Jeva: System One Models from First Principles (arxiv.org)

Beyond the Verdict: Agentic Code Judging with Verifiable Feedback (arxiv.org)

Agentic4PDM: Host-Governed Evidence Admissibility for Industrial Fault Diagnosis (arxiv.org)

Order-Consistent Latent Geometry for Reliable Short-Horizon Planning in World Models (arxiv.org)

Configuring RAG Systems: What Breaks, Where Originates, and How to Fix (arxiv.org)

Rethinking Quality in Large Language Model Routing: Security-Conditioned Model-and-Budget Selection (arxiv.org)

Input-Conditioned Search for Efficient Test-Time Scaling of Recursive Reasoners (arxiv.org)

Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale (arxiv.org)

The Smart Buildings Challenge: Learning Foundation Models for Real-World Optimization at Scale (arxiv.org)

Posterior Flow Matching (arxiv.org)

In-Context Batch Black-box Optimization via Flow Matching (arxiv.org)

KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks (arxiv.org)

From Spectral to Frobenius: When Muon Loses Its Spectral Implicit Bias (arxiv.org)

Exhaustive Canonical Skeletons Generation: A Formal Dataset for Decomposable Symbolic Regression (arxiv.org)

What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books (arxiv.org)

Rethinking Prediction Horizons: Decoupling Model Queries from State Updates in Few-Step MeanFlow (arxiv.org)

Can You Trust Your Preference Models? Stackelberg Games for Alignment (arxiv.org)

Certify the Decision, Not Every Score: Exact Top-k Selection from Few Evaluations (arxiv.org)

Single-Sample Decoding of a Two-Layer Sound-Word Canvas Model for Thai Klon Pad (arxiv.org)

FleetPilot: Fine-Grained Steering Empowers Real-World LLM Personalization (arxiv.org)

SkillHorizon: Constructing Long-Horizon Agent Tasks through Skill Composition (arxiv.org)

What Survives When Activation Steering Is Used as Memory (arxiv.org)

MCPDisCo: Seesaw-Mitigated Tuning for MCP-Native Agents (arxiv.org)

VC-OPD: Visual Counterfactual On-Policy Distillation for Grounded Vision-Language Reasoning (arxiv.org)

Predicting how model propensities generalize out of distribution (arxiv.org)

Belief Embedding Tree Search (arxiv.org)

Trainable at Depth, Solvable at Small Size: What the Loss Surface of a Polynomial Activation Dictates for Designing, Initializing and Training Networks (arxiv.org)

GBTSVR: Twin Support Vector Regression via Granular Ball Computing (arxiv.org)

Sequential Learning-Based Physics-Informed Neural Network Modeling for Parameter Estimation on a Power Converter (arxiv.org)

On the Convergence of Semi-Gradient TD: A Mean Field Perspective (arxiv.org)

ChemArgus: Automated Scientific Diagnostics for Lab-Grade Multimodal Chemistry Reasoning: A Frontier Benchmark (arxiv.org)

Provable Transformer Context Compression for In-Context Reinforcement Learning (arxiv.org)

On the Complexity Gap Between POMDPs and Multi-Environment POMDPs (arxiv.org)

SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications (arxiv.org)

WorkBorne: Discovering Agentic Tasks from Real-World Artifact Transitions (arxiv.org)

Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling (arxiv.org)