AI News

Low-Resource Safety Failures Are Action Failures, Not Representation Failures (arxiv.org)

Reinforcement Learning for Optimal Experiment Design in Parameter Identification of Mechatronic Systems (arxiv.org)

Computation-Aware Kalman Filtering with Model Selection for Neural Dynamics (arxiv.org)

MENTIS: What Belief Changes Under Alignment? Measuring Multi-Scale Latent Torsion in Language Models (arxiv.org)

Two-Fidelity Best-Action Identification for Stochastic Minimax Tree (arxiv.org)

Riemannian Optimization for Hadamard Products of Low-Rank Matrices (arxiv.org)

A Lightweight Hybrid MLP-Based Framework for Real-Time Phishing URL Detection Using Structural URL Features (arxiv.org)

3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code (arxiv.org)

Understanding LLM Behavior in Multi-Target Cross-Lingual Summarization (arxiv.org)

Riemannian Stochastic Optimization for Sufficient Dimension Reduction (arxiv.org)

Early Prediction of Liver Cirrhosis Up to Two Years in Advance: A Machine Learning Study Benchmarking Against the FIB-4 and APRI Scores (arxiv.org)

When Softmax Fails at the Top: Extreme Value Corrections for InfoNCE (arxiv.org)

Probabilistic Performance Guarantees for Multi-Task Reinforcement Learning (arxiv.org)

Speculative Sampling For Faster Molecular Dynamics (arxiv.org)

Forget Attention: Importance-Aware Attention Is All You Need (arxiv.org)

Structure and Scale in Simplicial Sequence Modelling (arxiv.org)

AdaKernel: Learning Adaptive Kernel Parameters for Spatiotemporal Graph Neural Networks (arxiv.org)

ShaplEIG: Bayesian Experimental Design for Shapley Value Estimation (arxiv.org)

DraDDP: A Multimodal Multi-Party Dialogue Discourse Parsing Dataset (arxiv.org)

Unplugging a Seemingly Sentient Machine Is the Rational Choice -- A Metaphysical Perspective (arxiv.org)

TN-SHAP-G: Graph-Structured Tensor Network Surrogates for Shapley Values and Interactions (arxiv.org)

Interpretability in Deep Time Series Models Demands Semantic Alignment (arxiv.org)

Sharpness-Aware Hybrid Model Learning for Architecture-Agnostic Parameter Estimation (arxiv.org)

Expressivity of congruence-based architectures for DNNs on positive-definite matrices (arxiv.org)

Minimax-Optimal Policy Regret in Partially Observable Markov Games (arxiv.org)

FIRM: Federated In-client Regularized Multi-objective Alignment for Large Language Models (arxiv.org)

When Data Is Scarce: Scaling Sparse Language Models with Repeated Training (arxiv.org)

Efficient Synthetic Network Generation via Latent Embedding Reconstruction (arxiv.org)

Understanding Stigmatizing Language in Clinical Documentation: A Paired Comparison of Ambient AI Drafts and Clinician Finalized Notes (arxiv.org)

Equilibrium Propagation for Non-Conservative Systems (arxiv.org)

Beyond Sinusoids: A Morlet Wavelet Framework for Transformer Positional Encoding (arxiv.org)

Can Vision Language Models Learn Intuitive Physics from Interaction? (arxiv.org)

Reason, Retrieve, Re-rank: A Zero-Shot Reasoning-Aware Framework for Composed Video Retrieval (arxiv.org)

LRAgent: Efficient KV Cache Sharing for Multi-LoRA LLM Agents (arxiv.org)

Knowledge-Informed Kernel State Reconstruction from Heterogeneous Partial Observations (arxiv.org)

Lookahead Sample Reward Guidance for Test-Time Scaling of Diffusion Models (arxiv.org)

When Does Predictive Inverse Dynamics Outperform Behavior Cloning? (arxiv.org)

It does what it says on the tin: safe synthetic data from coarsened margins (arxiv.org)

Linear Strategic Classification with Endogenous Improvements (arxiv.org)

From Zero to Hero: Advancing Zero-Shot Foundation Models for Tabular Outlier Detection (arxiv.org)

Polar Depth for Potentially Heavy-Tailed Data (arxiv.org)

Large-scale Uncertainty Quantification for Latent Variable Models Using Subsampling Markov Chain Monte Carlo (arxiv.org)

ERICA: Quantifying Replicability of Cluster Analysis (arxiv.org)

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation (arxiv.org)

When Hard Negatives Hurt: Bridging the Generative-Discriminative Gap in Hard Negative Synthesis for Retrieval (arxiv.org)

Tracing GenAI Literacy: Uncovering Student-AI Interaction Patterns in Academic Writing through Epistemic Network Analysis (arxiv.org)

Consistency Deep Equilibrium Models (arxiv.org)

Benchmarking Recursive-Collapse Warning Claims Under Matched False-Positive Control (arxiv.org)

Implicit Geographic Inference in LLM Medical Triage: Language-Driven Disparities in Emergency Recommendations (arxiv.org)

Evaluating Interactive Reasoning in Large Language Models: A Hierarchical Benchmark with Executable Games (arxiv.org)