AI News

ELF-REG: Scaling Continuous Diffusion Language Models to Reasoning Tasks (arxiv.org)

TraceGuard: Adaptive Multimodal Poison Filtering through Cross-Feature Rank Agreement (arxiv.org)

A Rapid Pipeline for Training and Deploying ML Models on WeBe Band (arxiv.org)

SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL (arxiv.org)

Where Hallucinations Live: A Cross-Architecture Circuit in VQ-Tokenized Vision-Language Models (arxiv.org)

Pistis Technical Report (arxiv.org)

PAWS: Policy-driven Agentic World Simulation (arxiv.org)

TW3Cast: A Frozen Router of Lightly Fine-Tuned Foundation Models for Time-Series Forecasting on GIFT-Eval, Selected Entirely on the Training Split (arxiv.org)

Generating Interesting Scientific Ideas using Knowledge Graphs and LLMs: Evaluations with 100 Research Group Leaders (arxiv.org)

LLM Agents Can Easily Tamper With Their Own Traces (arxiv.org)

RAPID: Robot Agentic Programming from Demonstrations (arxiv.org)

Rolling-WAM: World Action Models with Rolling Imagination (arxiv.org)

Coding Agents for Generalized Task and Motion Planning Problems (arxiv.org)

TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations (arxiv.org)

Paging the Experts: A Reproducible Characterization of Flash-Backed MoE Inference on iPhone (arxiv.org)

Exploiting answer-invariant redundancies in satellite imagery for efficient VLM inference on edge (arxiv.org)

EvoTreeNAD: Genealogy-Guided Evolution for LLM-Driven Neural Architecture Discovery (arxiv.org)

Learning and interpreting policies for simultaneous entanglement requests in quantum networks (arxiv.org)

GRASP: Generating, Revising, and Assessing for Strategic Planning with Agentic AI (arxiv.org)

Orbital Error Dynamics: Self-Organized Criticality, Ephemeral Parameter Resonance, and Non-Linear Biological Ontologies in Zero-Storage Neural Synthesis (arxiv.org)

MQSS-Selector: RL-Guided Pass Selection for an MLIR Compilation Pipeline (arxiv.org)

Requirement-Bound Verified Commissioning: A Frozen Four-Billion-Parameter Local Model as a Candidate Generator under an External Acceptance Layer with Verification and Release Authority (arxiv.org)

Return or Revise? Learning When Revision Helps Retrieval-Augmented QA (arxiv.org)

When Should Forecasting Agents Reason? Behavioral Stress Tests for Reliability Routing (arxiv.org)

Detecting Glaucoma Across Multi-ethnic Myopic and Non-Myopic Populations Using an Uncertainty-Aware Vision Transformer: A Multicentre Model Development and Validation Study (arxiv.org)

Generative Evolutionary Design of Voxel-Based Soft Robots with Provable Optimality (arxiv.org)

RoboLDA: A Probabilistic Generative Model for Uncovering Embodied Hierarchical Structures in Voxel-based Soft Robots (arxiv.org)

agentic-ger: terminology recovery in long-form speech using global context (arxiv.org)

Rufus-Air: An Open LLM Post-Training Recipe (arxiv.org)

Segment-Level Risk Discovery in Online Handwriting for Alzheimer's Disease Detection (arxiv.org)

From Policy Documents to Structured Survey Responses: Evaluating Large Language Models for Policy Monitoring (arxiv.org)

Spot, Separate, and Enhance: Fully Generative Approach for Audio Mixing (arxiv.org)

HarnessPAI: An Evolving Harness for Physical AI (arxiv.org)

Med-AR: Autoregressive Vision-Language Pretraining for Long-Tailed Chest X-Ray Classification and Uncertainty-Aware Evaluation (arxiv.org)

AI-Moderated Interviews for Market Research and Digital Twins Calibration (arxiv.org)

On the SoS Certifiability of Log-Concave Distributions (arxiv.org)

From Processing to Functionality: Engineering Accessible Material States in Cu-Embedded SiO$_x$ Memristive Devices (arxiv.org)

AERIAL: Adversarial Evaluation of Robustness in Accuracy-Preserving Low-Precision EEG Decoders (arxiv.org)

Deep learning of longitudinal visual fields predicts glaucoma progression rate and identifies fast progressors (arxiv.org)

Self-Play Pretraining with Zero Data (arxiv.org)

Residual Correlation as a Diagnostic for Joint-Uncertainty Gains from GP Coregionalisation (arxiv.org)

Spatio-temporally complementary feature propagation on graphs for longitudinal AADT estimation (arxiv.org)

Classifier-Dependent Benefits of Pseudo-Labeling for Semi-Supervised Android Malware Attribution (arxiv.org)

NNV3: Expanding Neural Network Verification to New Architectures and Domains (arxiv.org)

Style, Not Self: Surface Cues Explain Zero-Shot Code Attribution by Large Language Models (arxiv.org)

How does Adversarial Influence Scale in Multi-Agent Systems? (arxiv.org)

Synthetic Hospital: An Open, Verifiable, Physician-Validated Longitudinal EHR Benchmark (arxiv.org)

Advancing Model Research in AgentX: Long-Horizon Autonomy for Industrial Recommender Systems (arxiv.org)

Certified Predictive Value-of-Advice Gating for Cost-Aware Language-Model Guidance in Reinforcement Learning (arxiv.org)

Common Covariance Geometry and Certification for Brownian Kernel Ladders (arxiv.org)