arXiv · cs.LG
Recent machine learning papers from arXiv.
Read from its RSS feedAppears in:KI
Latest from arXiv · cs.LG
- Evolving to Match Capabilities: Improving Harness Self-Evolution via Skill Adaptationvor 11 Tg.
- Can Language Models Learn While They Think? Testing Gradient Adaptation Across Architecturesvor 11 Tg.
- Lessons from Luck: Do LLM-Written Lessons Carry Chance into an Agent's Next Decision?vor 11 Tg.
- Feature Scaling Beats Architecture Choice in Offline Respiratory Screening for Low-Resource Clinicsvor 11 Tg.
- Inference-Time Prior Adaptation for Imbalanced Data Streams under Concept Driftvor 11 Tg.
- Physics Informed Neural Networks for Solving the Drift-Diffusion Equationsvor 11 Tg.
- Neural-DISCO: Source-Conditioned Counterfactual Editing of Neural Population Activityvor 11 Tg.
- How Much Fresh Data Should CTR Recalibration Use? A Temporal Evaluation of Post-hoc Calibrationvor 11 Tg.
- Lower expected Fisher rank implies better PAC generalizationvor 11 Tg.
- Model-agnostic predictability quantification of multivariable time seriesvor 11 Tg.
- Vision Token Pruning via Query-Vision Interaction Decompositionvor 11 Tg.
- Neyman Orthogonal Activation Steeringvor 11 Tg.
- Reinforcement Learning for Controller Parameters in the Auto-Bidding System at Pinterestvor 11 Tg.
- RadonBench: A Benchmark of Classical and Learning-Guided Search Techniques for Problems in Combinatorial Geometryvor 11 Tg.
- Individual Fairness via Local Predictive Consistencyvor 11 Tg.
- All-in-One: A Composable Attack on LLM Transformationsvor 11 Tg.
- Tree Depth and Sample-Quality-Associated Features: Two Hidden Factors in Boruta Selection for RNA-Seq Data Classificationvor 11 Tg.
- When Does Pre-Trained Capacity Help Causal Learning? From Scalar Prediction to Counterfactual Post-Trainingvor 11 Tg.
- MLIPDiff: Guiding Denoising Diffusion Models for Structure-Based Drug Design via Machine-Learned Interatomic Potentialsvor 11 Tg.
- Augmentation-Guided Adversarial Training via Local Stability-Based Sample Selectionvor 11 Tg.
- Gaussian Approximation for Composition and Subsampling of Differentially Private Algorithmsvor 11 Tg.
- Information-Theoretic Scaling Laws for Generative Diffusion Trainingvor 11 Tg.
- The Clustering Occupation Kernel (CLOCK) Method for Time Series Arising from Dynamical Systemsvor 11 Tg.
- LSTM-Style Looped Transformers with Shared KV Cachesvor 11 Tg.
- Multiple Importance Sampling for Guided Speculative Inferencevor 11 Tg.
- One Call Ahead: Packed Self-Speculative Decoding with Diffusion Draftingvor 11 Tg.
- The Cost of an Unknown Horizon for the Last Iterate of Clipped SGD under Heavy-Tailed Noisevor 11 Tg.
- Faithful Federated Tabular Data Synthesisvor 11 Tg.
- COPE: Composable Performance Emulation for Mixture-of-Experts Offloading Policiesvor 11 Tg.
- Performance Gaps Between Open-Loop, Certainty-Equivalent, and Optimal Control Under Action-Dependent Observationsvor 11 Tg.