AI News

Sim-to-Real Transfer of Vision-Language Navigation in Continuous Environments Using an Ackermann-Steered Mobile Robot (arxiv.org)

Energy-Conditioned Noise Schedule and Whitening for Spectral Diffusion (arxiv.org)

Cascadia: Resident 975B MoE Inference on Eleven AI PCs (arxiv.org)

A Trust Layer for Agent Evaluation (arxiv.org)

An Information-Theoretic Evaluation Framework for Benchmark and Model Diagnosis in Knowledge Tracing (arxiv.org)

Repair Lot Skyline: A Weighted Constraint Satisfaction Approach to Pavement Repair Optimization from Geospatial Hazard Density (arxiv.org)

DART-ES: Difficulty-Aware Reweighting and Targeted Replay for Fine-Tuning LLMs with Evolution Strategies (arxiv.org)

Mask-Guided KV Cache Eviction in Block Diffusion Language Models (arxiv.org)

STOCK-JEPA: Prior-Anchored Latent Revision Representation Learning in Equity Markets (arxiv.org)

Identifiable World Models from Pretrained Diffusion Representations (arxiv.org)

Shaping the Wind: Nested Potentials for Kinematically Admissible Urban Wind Prediction (arxiv.org)

TRIAGE: Direction-Aware Mismatch Stabilization of Native NVFP4 Reinforcement Learning (arxiv.org)

SPECTRUM: Proximal Spectral Modulation for Looped Self-Distillation (arxiv.org)

Can Semantic Geometry Teach an AI Judgement? (arxiv.org)

Internalizing Agent Experience into Diffusion Model Weights via On-Policy Context Distillation (arxiv.org)

MemMux: Runtime Verification and Honest Resource Attribution for Fleets of Parallel Coding Agents (arxiv.org)

Verifying Coordination in Parallel Coding Agents: NP-Bench and a Scheduling Planner (arxiv.org)

Does the Model Use the Feature? Separating Steering from Mechanism in LLMs (arxiv.org)

The Right Memory in the Wrong Context: Verifying Retrieval Admissibility in Long-Term Agent Memory (arxiv.org)

Skillful Data-Driven Subseasonal Soil Moisture Forecasting: Prospects and Limits for Flash Drought Prediction (arxiv.org)

Learning to Simulate Individuals from Macro Social Signals (arxiv.org)

What Must Replay Preserve? Separating Correctable Bias from Class Correspondence (arxiv.org)

Few-Shot Bioactivity Prediction with Meta-Learning under Assay Heterogeneity (arxiv.org)

Demo: Vision-Language Model-Guided Online Calibration of an Electromagnetic Digital Twin (arxiv.org)

Rationale-Guided Policy Optimization: Learning to Reason with Adaptive Rationale Scaffolding (arxiv.org)

Trajectory-Retrieval Speculative Decoding: When Does a Model's Own History Help? (arxiv.org)

Evaluating Escalation Signals for LLM Routing: Targets, Controls, and Five Ways to Fool Yourself (arxiv.org)

Evaluate the Stack, Not the Layer: Do Deterministic and LLM Gates for Agent Actions Fail Independently? (arxiv.org)

MemCo: Memory-Centric Collaboration for Generalizing LLM Agents to Unseen Environments (arxiv.org)

When Attention Does Not Explain the Peak: Temporal Reference vs. Forecast Output in Attention-Based Time-Series Forecasting (arxiv.org)

SchemaFill: Efficient LLM Tool Calling via Slot-Parallel Speculative Decoding (arxiv.org)

Muon Is Theoretically Wrong For Convolutions, But Empirically Effective (arxiv.org)

LiLib: Lifelong Air-to-Ground Path-Loss Prediction on UAVs via a Drift-Triggered Model Library (arxiv.org)

Auditable Claims about AI Agents (arxiv.org)

Defense-in-Depth for LLMs: Evaluating Memory Gates Against Activation-Induced and Memory-Induced Sycophancy (arxiv.org)

2d-fet-bench: from spatial reasoning to fet design on flakes (arxiv.org)

Adaptive Gait Biofeedback With Participant-Held-Out Modeling and Participant-Specific Updating in Chronic Ankle Instability (arxiv.org)

When Does AI Supervision Help? A Role-Aware Study of Network Fraud Decision Management with Blockchain Auditability (arxiv.org)

Constant-Curvature Sliced Gromov-Wasserstein for Heterogeneous Cross-Curvature Alignment (arxiv.org)

Will the Judge Flip? Predicting Position-Sensitive LLM Judgments from Residual Stream Activations (arxiv.org)

Learnable Spectral Activations (arxiv.org)

Benchmarking Label-Revealed Online Updates for EEG BCI Decoding (arxiv.org)

StaFIR: Convex Learning of Stationarity-Aware Causal Filters (arxiv.org)

Decoupling What from Where: How Should a Small GUI Grounding Model Receive the Action Type? (arxiv.org)

COMPASS: Finding Where Reasoning Lives in Language Models (arxiv.org)

Memory Prediction Excess: A Probabilistic Quantity for Predictive Gain and Memory Length in Stochastic Processes (arxiv.org)

Low-Rank and Structured Sparse Tensor Decomposition for Anomaly Detection in Multivariate Functional Data (arxiv.org)

Data Fusion for Errors-in-Variables (arxiv.org)

How Many Independent Samples Does a Satellite Image Contain? Generalization Bounds for Spatially Dependent Data (arxiv.org)

sHAIL-Causal: A Sequential Staircase Procedure for Invariant Causal Predictor Discovery (arxiv.org)