AI News

Evo-B&B: Behavior-Aware LLM-Guided Joint Evolution of Branching and Node Selection Policies for MILP (arxiv.org)

World Action Critic (WAC) Model: Joint Foresight and Criticism for Policy Learning (arxiv.org)

Compression Precedes, Content Decides: Emergent Misalignment in Reward Models (arxiv.org)

When Can You Trust an Agent's Own Verifier? Stress-Testing Counterfactual Rationale Verification Under Search Pressure and Oracle Access (arxiv.org)

StoBasket: Joint Upsell and Cross-Sell Optimization for Contribution Margin (arxiv.org)

Deterministic Bounds on Accuracy and Stability of Metric-Space Regression and Classification (arxiv.org)

On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction (arxiv.org)

Don't Retrain, Remix: Adaptation of Web Agents using Mixture Hypernetworks (arxiv.org)

Predicting Privileged Information Effectiveness for On-Policy Self-Distillation (arxiv.org)

R$^2$Orch: Recursive Self-Improvement for Agent Orchestration via Self-Play within a Skill-Internalization Loop (arxiv.org)

Graph Serializations for Reasoning LLMs: A Study Across Knowledge Extraction and Structural Reasoning (arxiv.org)

LEMON: Equivariant Learning over Attention Maps for Zero-Shot Misbehavior Monitoring on New LLMs (arxiv.org)

GlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations (arxiv.org)

Brain-Aligned Multimodal LLMs for Social and Emotional Understanding of Cartoon Movies (arxiv.org)

Between Spoilers and Guesswork: Avoiding Over-Specification in Coding Benchmarks (arxiv.org)

Generative Value Classifier (arxiv.org)

Progressive Misalignment Can Emerge from Benign Recursive Self-Distillation (arxiv.org)

Lean4AgenticRL: Advancing Agentic RL with Verifiable Multi-level Process Rewards (arxiv.org)

Sparse Autoencoders Can Induce Local Geometry for Mitigating Catastrophic Forgetting (arxiv.org)

FlexUMM: Flexible Representation Learning for Unified Multimodal Models (arxiv.org)

Not All Graphs Are Maps: A Task-Agnostic Exploration Framework for Mobile GUI Navigation (arxiv.org)

From Relevant Memories to Sufficient Evidence: Rethinking Memory Utilization in LLM Agents (arxiv.org)

At the Limits of Computability: Steps Towards Autonomous AI Discovery on the Busy Beaver Frontier (arxiv.org)

Not All Test-Time Compute Is Evidence: Model-Specific Bias in LLM Reasoning (arxiv.org)

TRACE: Transformer Residual Activation Codebook Encoding for Communication-efficient Distributed Edge Inference (arxiv.org)

BOLT: Multi-Bottleneck Bound-Guided Learning for Flexible Job-Shop Scheduling with Time Lags and Resource-Coupled Transport (arxiv.org)

CGDT-CESN: Correlation-Guided and Dynamics-Trained Clustered Echo State Network for Multivariate Time-Series Prediction (arxiv.org)

Binding-Specific Relearning Savings for Test-Time Continual Learning (arxiv.org)

Repair, Retry, and the Cost of Hindsight: Verifier Feedback in Formal Theorem Proving (arxiv.org)

Rank Measures of Fine-Tuning Updates Need a Noise Baseline (arxiv.org)

CF-NODE: Counterfactual Spatial Node Intervention for Test-Time Adaptation Under Domain Shift, Evaluated Across Five Diabetic Retinopathy Datasets (arxiv.org)

Generative adversarial networks with Monte Carlo dropout uncertainty quantification for structural health monitoring: Fibre Bragg grating data augmentation comparison in Hawk T1A aircraft (arxiv.org)

Repairing LLM-Generated Specifications with Deterministic Diagnosis (arxiv.org)

Is the Policy Gradient a Gradient? (arxiv.org)

Reinforcement Learning in MDPs with Information-Ordered Policies (arxiv.org)

Missing value imputation with adversarial random forests - MissARF (arxiv.org)

What is causal about causal models and representations? (arxiv.org)

Exact Tail Laws for Nonlinear SGD with Infinite-Variance Gradients (arxiv.org)

Equivalence testing and run-count design for stochastic machine-learning experiments (arxiv.org)

No-Regret Language Generation (arxiv.org)

Boundary Detection for High-Dimensional Point Clouds (arxiv.org)

Beyond Expected Values: Risk-Sensitive Planning with Distributional Monte-Carlo Tree Search (arxiv.org)

Causal Discovery beyond Conditional Independence: Identifiability, Algorithms, and Statistical Inference for Directed Acyclic and Cyclic Graphs (arxiv.org)

Minimax-optimal Scaling Laws in Ridge Regression via Transfer Learning (arxiv.org)

A Class of R\'enyi-based Concentration Metrics for Kolmogorov-Arnold Geometry in Multilayer Perceptrons (arxiv.org)

Towards Hyperparameter Transfer for Differentially Private Optimization (arxiv.org)

Conformal Policy Selection: When to Deploy AI Decisions for Targeted Intervention (arxiv.org)

Flat Loss, Evolving Features: Grokking in Modular Addition (arxiv.org)

Computing Fr\'echet on Unknown Submanifolds of Riemannian Manifolds (arxiv.org)

Learning smooth dynamics via Kinematic Interpolants (arxiv.org)