AI News

FAER: Auditable Utility-Aligned Trajectory Replay for Language Model Post-Training (arxiv.org)

EvoGen-Harness: Learning Where and How to Evolve Image-Generation Harnesses (arxiv.org)

OmniMed-Jev: Calibrating LVLM Confidence for Trustworthy Medical Multimodal Decisions via System One (arxiv.org)

Fusion techniques of time frequency-based images to predict the outcome of rTMS depression therapy (arxiv.org)

STCFormer: Adaptive Spatio-Temporal Modeling with Dynamic Cluster Transformer for Station-based Weather Forecasting (arxiv.org)

A First Glance at Jev for Network Traffic Classification: Accuracy, Processing Time, and Cost (arxiv.org)

When Do Attention-Head Ablations Support Causal Claims? Projection-Level Confounds, Floor Effects, and Matched Controls (arxiv.org)

M$^2$Weather: A Benchmark for Joint Multi-Station and Multi-Variable Weather Forecasting (arxiv.org)

MoRA: MoE Pruning via Router Bias Learning and Expert Approximation (arxiv.org)

Manifold-Constrained Initial Noise Optimization for Efficient Generative Model Alignment (arxiv.org)

Deep Learning for Anomaly Detection in Railway Systems: A Structured Survey (arxiv.org)

Benchmarking System One decision models against trained classifiers and language models for automated decision gates (arxiv.org)

The Weakest Link: Distilling LLM Reasoning with Worst-Case Constrained Reinforcement Learning (arxiv.org)

Beyond Diagonal State Space Models: Exact Non-Abelian Group Tracking, Solvability Barriers, and Geometric Physical Manifolds (arxiv.org)

Bellman-Certified Rounding for Sparse Policy Deployment in MDPs (arxiv.org)

Partial AUC Maximization from Positive-unlabeled Data (arxiv.org)

Stable and Counterfactually Robust Physical World Models from Imposed Structure and Learned Physics (arxiv.org)

Coupling Perception and Reasoning in Federated Multimodal Graph Foundation Models (arxiv.org)

MOVE: Multimodal Open-world Verification and Expansion for Graph Learning (arxiv.org)

Large Language Bayes Is Not Reparameterisation-Invariant (arxiv.org)

Attention Manifolds: Steering or Blocking Language Models by Editing Learned B-Spline Surfaces (arxiv.org)

Verification Pulses and the Cost of Escaping Wrong Consensus (arxiv.org)

The Null Is the Hard Part: Exact Tests for Memorization in Generative Models (arxiv.org)

Contingent Exposure Routing for Financial AI: Outage Risk and the Cost of Indivisible Decisions (arxiv.org)

How Many Categories Are Enough? Distribution-Free Certification Limits for Few-Shot Anomaly Thresholds (arxiv.org)

Constant-Memory Recall: Learned Associations in a Fixed Matrix State (arxiv.org)

Useful to Whom? Sample Value Is Defined Only Relative to the Learner (arxiv.org)

Sharp Oracle-Regret Tradeoffs for Projection-Free Online Convex Optimization (arxiv.org)

Energy Time-Series Imputation with Differentially Private Diffusion Models via Clipping-Aware Objective Conditioning (arxiv.org)

Uncertainty-Aware RL-Controlled Adaptive 3D Mapping (arxiv.org)

Classification Based on Association Rules Algorithm for Breast Cancer (arxiv.org)

Generalized Biomedicine Discovery (arxiv.org)

The Hidden Costs of 99% Accuracy: A Trustworthiness Audit of the Telco Customer Churn Benchmark (arxiv.org)

Uncertainty-Aware Learning from Multi-Expert Interval Targets (arxiv.org)

One Mastery Threshold Does Not Fit All Knowledge Tracing Models (arxiv.org)

Nous: Learning and Certifying Memory Decisions Before Source Calibration (arxiv.org)

"very likely" Means "uncertain"? How LLMs Diverge from Humans in Linguistic Uncertainty Quantification (arxiv.org)

SW-KAN: Kolmogorov-Arnold Networks with Stieltjes-Wigert q-Orthogonal Polynomials (arxiv.org)

Fast Polynomial Transcendentals for LLMs (arxiv.org)

Integrating Fairness and Explainability in a Multiple Instance Reinforcement Learning System (arxiv.org)

FourierQK: Filter Shape, Admissibility and the Leakage-Coverage Law (arxiv.org)

How Far is Adam from Natural Gradient Descent? (arxiv.org)

Supervise What Decides Success: Criterion-Aligned Auxiliary Losses for Latent World-Model Planning (arxiv.org)

Reverse Item Response Theory for Sparsity-Robust Ranking in Fragmented Cancer Drug-Response Matrices (arxiv.org)

Langevin-Informed Transfer Learning: Replacing Target Samples by Black-Box Feedback (arxiv.org)

Auto-Formalizing Neuro-Symbolic Predictors (arxiv.org)

FedMIX-P: Mixing Local and Global Preconditioners for Federated Vision and Language Model Training (arxiv.org)

Learned End-to-End Guidance Schedules for Diffusion Models (arxiv.org)

Let the Heads Talk: Beyond Diagonal Graph Attention (arxiv.org)

Tight Transition Time Bounds for Separable Logistic Regression at the Edge of Stability (arxiv.org)