AI News

Our Own Jev: Typed Decisions in One Forward Pass, from an In-House Judge to Open LoRA Models (arxiv.org)

OccJEPA: Occlusion as Masking for Amodal Video Segmentation (arxiv.org)

Dual-Channel Contrastive Encoding of Sperm Whale Coda (arxiv.org)

DriftNeRF: Cross-Ray Density Drifting for Sparse-View Neural Radiance Fields (arxiv.org)

SU(2)-Equivariant Spinor Message Passing for Molecular Magnetism (arxiv.org)

Recursive Self-Improvement on a Budget: Scoring Code Edits Before They Run (arxiv.org)

To Keep or Not to Keep: Learning KV Cache Retention in Disaggregated LLM Serving Systems (arxiv.org)

Extending ROC Analysis to Uncertainty-Aware Risk Prediction with an Interval-Based AUC (iAUC) (arxiv.org)

Machine Unlearning in Low-Dimensional Feature Subspace (arxiv.org)

From Fuzzy to Exact: The Halo Architecture for Infinite-Depth Reasoning via Rational Arithmetic (arxiv.org)

ChemBART: Pre-training a Reaction Language Model for Comprehensive Synthesis Planning (arxiv.org)

Energy-Based Transfer for Reinforcement Learning (arxiv.org)

In-Context Bias Propagation in LLM-Based Tabular Data Generation (arxiv.org)

Understanding and Mitigating Under-Confidence in GNNs from the Final Layer (arxiv.org)

Insights on back marking for the automated identification of animals (arxiv.org)

Quantifying Symptom Causality in Clinical Decision Making: An Exploration Using CausaLM (arxiv.org)

LongZero: Enhancing Long-Context Reasoning in LLMs from Zero Data (arxiv.org)

Think Harder Where It Matters: Item-Level Adaptive Recursion for Sequential Recommendation (arxiv.org)

A Quasipolynomial Algorithm for $\ell_\infty$ Contractions (arxiv.org)

Minimax Optimality of the NPMLE for Gaussian Mixtures in Hellinger Distance (arxiv.org)

Asymmetric On-Policy Distillation Learns In-Context Exploration (arxiv.org)

VX-One: Typed, Calibrated Decision Heads for Real-Time Perimeter Triage - An empirical study of unknown-attack honesty, cascade gray zones, and calibration under drift (arxiv.org)

Looped Uniform Diffusion Language Models (arxiv.org)

HGM-Dual: Lineage-Aware Guided Agent Harness Optimization for Long-Horizon Planning (arxiv.org)

Code Comprehensor: Hierarchical Documentation and Cross-Modal Retrieval for Large Repositories (arxiv.org)

TIDE: Tree-Induced Fine-Grained Diagnostic Evaluation from Model Behaviors (arxiv.org)

Implicit Feature Reweighting for Label-Free Graph Generalization (arxiv.org)

SOAR: Social and Opinion-Aware Recommendations (arxiv.org)

WRDO: Weighted Refusal Direction Optimization in Aligned Language Models (arxiv.org)

Stochastic Pruning of the DPO Computation Graph: 1.4x Faster Training and Lower Peak Memory (arxiv.org)

Block Diffusion for Longitudinal EHR: Accuracy, Efficiency, and Representation Trade-offs (arxiv.org)

Phase Transitions in Learning Driven by Extremal Data (arxiv.org)

Are Joint-Embedding Predictive Objectives Effective for Longitudinal EHRs? (arxiv.org)

Geometric knowledge spaces, Attention pairing and Architectures (arxiv.org)

Shortcut-Resistant CAM Distillation for Long-Tailed Recognition (arxiv.org)

Rethinking Agentic Model Orchestration for Time Series Forecasting (arxiv.org)

Asymmetry Tunneling Beyond Adaptive Equivariance Relaxation (arxiv.org)

Harness Scaling for Generalizable Agentic Reinforcement Learning (arxiv.org)

How Far Can Foundation Models Be Distilled for Efficient Jet Tagging? (arxiv.org)

BIRD-RL: Scaling Agentic Reinforcement Learning over Stateful Data-Centric Environments (arxiv.org)

Generalization Bounds via Vector-Valued $L_p$ Correlation $V_\alpha$ (arxiv.org)

Alternative Strategies for Active Learning with Bayesian Bradley Terry Model (arxiv.org)

Progressive Permutation Alignment with Temperature-Controlled Spectral Sharpening (arxiv.org)

Nominate-then-Adjudicate: LLM-Assisted One-Pass and Per-Instance MIP Solver Tuning (arxiv.org)

Parallelizable Probabilistic Latent Reasoning for Diverse Reasoning Trajectories (arxiv.org)

Compositional neural networks for multimodal X-ray and neutron scattering inversion (arxiv.org)

Inverse RL Helps Align AI by Imitating Humans (arxiv.org)

EPIC: Rethinking Trust Regions for Autoregressive LLM Policies via Exact-KL Prefix Importance-Ratio Clipping (arxiv.org)

Preventing Model Extraction in Contextual Linear Optimization (arxiv.org)

Reliability-Aware Probabilistic Fusion for Multimodal Learning with Incomplete Evidence (arxiv.org)