arXiv · cs.LG
Recent machine learning papers from arXiv.
Read from its RSS feedAppears in:IA
Latest from arXiv · cs.LG
- Evolving to Match Capabilities: Improving Harness Self-Evolution via Skill Adaptationil y a 11 j
- Can Language Models Learn While They Think? Testing Gradient Adaptation Across Architecturesil y a 11 j
- Lessons from Luck: Do LLM-Written Lessons Carry Chance into an Agent's Next Decision?il y a 11 j
- Feature Scaling Beats Architecture Choice in Offline Respiratory Screening for Low-Resource Clinicsil y a 11 j
- Inference-Time Prior Adaptation for Imbalanced Data Streams under Concept Driftil y a 11 j
- Physics Informed Neural Networks for Solving the Drift-Diffusion Equationsil y a 11 j
- Neural-DISCO: Source-Conditioned Counterfactual Editing of Neural Population Activityil y a 11 j
- How Much Fresh Data Should CTR Recalibration Use? A Temporal Evaluation of Post-hoc Calibrationil y a 11 j
- Lower expected Fisher rank implies better PAC generalizationil y a 11 j
- Model-agnostic predictability quantification of multivariable time seriesil y a 11 j
- Vision Token Pruning via Query-Vision Interaction Decompositionil y a 11 j
- Neyman Orthogonal Activation Steeringil y a 11 j
- Reinforcement Learning for Controller Parameters in the Auto-Bidding System at Pinterestil y a 11 j
- RadonBench: A Benchmark of Classical and Learning-Guided Search Techniques for Problems in Combinatorial Geometryil y a 11 j
- Individual Fairness via Local Predictive Consistencyil y a 11 j
- All-in-One: A Composable Attack on LLM Transformationsil y a 11 j
- Tree Depth and Sample-Quality-Associated Features: Two Hidden Factors in Boruta Selection for RNA-Seq Data Classificationil y a 11 j
- When Does Pre-Trained Capacity Help Causal Learning? From Scalar Prediction to Counterfactual Post-Trainingil y a 11 j
- MLIPDiff: Guiding Denoising Diffusion Models for Structure-Based Drug Design via Machine-Learned Interatomic Potentialsil y a 11 j
- Augmentation-Guided Adversarial Training via Local Stability-Based Sample Selectionil y a 11 j
- Gaussian Approximation for Composition and Subsampling of Differentially Private Algorithmsil y a 11 j
- Information-Theoretic Scaling Laws for Generative Diffusion Trainingil y a 11 j
- The Clustering Occupation Kernel (CLOCK) Method for Time Series Arising from Dynamical Systemsil y a 11 j
- LSTM-Style Looped Transformers with Shared KV Cachesil y a 11 j
- Multiple Importance Sampling for Guided Speculative Inferenceil y a 11 j
- One Call Ahead: Packed Self-Speculative Decoding with Diffusion Draftingil y a 11 j
- The Cost of an Unknown Horizon for the Last Iterate of Clipped SGD under Heavy-Tailed Noiseil y a 11 j
- Faithful Federated Tabular Data Synthesisil y a 11 j
- COPE: Composable Performance Emulation for Mixture-of-Experts Offloading Policiesil y a 11 j
- Performance Gaps Between Open-Loop, Certainty-Equivalent, and Optimal Control Under Action-Dependent Observationsil y a 11 j