AI News

Efficient Reasoning via Constrained Optimization in Latent Space (arxiv.org)

READ-Bench: Benchmarking Historical Instance Retrieval for Time-Series Diagnosis (arxiv.org)

Reading the Whole Heart: Latent-Attention Masked Autoencoders for Multimodal Cardiac Representation Learning (arxiv.org)

Phase-HDC: Replacing Optimizer History with Gradient Thresholds in Discrete Phase Learning (arxiv.org)

Recurrent Self-Improvement: Dynamic Cross-Loop On-Policy Distillation for Looped Language Models (arxiv.org)

Self-Organization from Constrained Geometric Radiation (arxiv.org)

When Routing Reveals Membership: Privacy Leakage from MoE Router Telemetry (arxiv.org)

NanoProof: Open and Efficient Automated Theorem Proving in Lean 4 (arxiv.org)

Temporal transformer CAN encoder with federated lightweight heads for anomaly detection (arxiv.org)

SPERA: Spherical Prior EEG Foundation Model with Geometry- and Frequency-Aware Latent Prediction (arxiv.org)

Freeze the Decoder, Heal the Encoder: Parameter-Efficient Adaptation for SVD-Based KV-Cache Compression (arxiv.org)

Coverage, Not Difficulty, Sets How Much Synthetic Data an Activation Probe Needs (arxiv.org)

RIFT: Relative Isolation From Trees For Anomaly Detection (arxiv.org)

HI3D 3.0 (Twinkle3D): Object-specific 3D Asset Generation with High Resolution (arxiv.org)

One Permutation Is All You Need: Fast, Deterministic Feature Importance and Model Stress-Testing (arxiv.org)

Conformal Data Contamination Tests for In-distribution Data Acquisition (arxiv.org)

Equilibrium Distribution for t-Distributed Stochastic Neighbor Embedding with Generalized Kernels (arxiv.org)

Networks with Finite VC Dimension: Pro and Contra (arxiv.org)

SOL: Measuring Gaps between Text Distributions by Double Sliced Wasserstein Metrics (arxiv.org)

Measuring Learned Monotone Temporal Aggregation at Matched Admissibility (arxiv.org)

Towards Scalable Meta-Learning of near-optimal Interpretable Models via Synthetic Model Generations (arxiv.org)

Deep reinforcement learning for optimal trading with partial information (arxiv.org)

Convergence of graph Dirichlet energies and graph Laplacians on intersecting manifolds of varying dimensions (arxiv.org)

Handling Covariate Mismatch in Collaborative Linear Prediction (arxiv.org)

KDFP: A first-principles approach to knowledge distillation in large language models (arxiv.org)

A Survey on Archetypal Analysis (arxiv.org)

On the Tightness and Computational Tractability of Higher-Dimensional Confidence Sequences (arxiv.org)

What can linear attention learn from nonlinear teachers in-context? (arxiv.org)

From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier (arxiv.org)

Ghost tasking for parametrized Gaussian Processes solving linear differential equations (arxiv.org)

On the Clock: Towards Punctual and Productive Time-Budgeted AI Agents (arxiv.org)

AI-Mediated Self: How HCI Defines and Relates to the Self (arxiv.org)

Whose Ground Truth? Embracing Ambiguity in Human-Centered AI (arxiv.org)

Plan-and-Patch: Diffusion Language Models for Agentic Planning (arxiv.org)

Speaking the Navigator's Language: Trajectory-Grounded Instruction Translation for Frozen Aerial VLN Agents (arxiv.org)

GameCommBench: A Unified Benchmark and Type-Aware Evaluation for AI-Generated Game Commentary (arxiv.org)

Bilevel optimization for data-driven learning of Koopman embeddings using kernel-based autoencoders (arxiv.org)

Self-sufficient Independent Component Analysis for Demixing Flows (arxiv.org)

Closing the Horizon Gap in Policy Optimization for Adversarial MDPs (arxiv.org)

Composite Online-to-Nonconvex Conversion with Optimal Oracle Complexity (arxiv.org)

The Harness as the Only Mutable Surface: Compliance-Bounded Self-Evolution of LLM Agents in Credit Pipelines, with a Measured Admission Gate (arxiv.org)

Verification with Transfer: Exact Information Frontiers and Their Price in Calls (arxiv.org)

Credal Machine Learning for Risk-Averse Decision Making (arxiv.org)

Exploiting Gradients in Bayesian Inference of Expensive Simulators (arxiv.org)

Optimal random quantisers for spherically symmetric distributions (arxiv.org)

Verification and Self-Improvement in Agentic AI: Foundations and Limits (arxiv.org)

Automated Detection of Match Phases in Football from Spatio-Temporal Tracking Data Using Graph Neural Networks (arxiv.org)

Rare Gate Disagreements Can Limit Plasticity: When Gradient Flow Mispredicts Finite-Batch SGD (arxiv.org)

From Geometry to Generalization: Why Row Normalization Can Beat Adam and Muon (arxiv.org)

Regularized Small Area Estimation with Graph Laplacian Benchmarking priors (arxiv.org)