archive
Every paper Pith has read. Search by title, abstract, or pith.
14903 papers in cs.LG · page 18
-
Adaptive rates lower energy use in humanoid robot teleoperation
Domain-Adaptive Communication-Rate Optimization for Sim-to-Real Humanoid-Robot Wireless XR Teleoperation
-
Decoupled recursion cuts interference in MLLM edits
Modality-Decoupled Online Recursive Editing
-
Factor-augmented SGD converges with streaming high-dimensional data
Factor Augmented High-Dimensional SGD
-
LLMs learn redundant copies of concepts across languages
Language models struggle with compartmentalization
-
Trajectory selection beats sampling in delayed disambiguation
EviTrack: Selection over Sampling for Delayed Disambiguation
-
High-pass spectral filter fixes Muon failures in VLA and RLVR
Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVR
-
Best volatility forecast model differs from best portfolio model
Do Better Volatility Forecasts Lead to Better Portfolios? Evidence from Graph Neural Networks
-
Three different models win at forecast error
Do Better Volatility Forecasts Lead to Better Portfolios? Evidence from Graph Neural Networks
-
Modular platform enables concurrent LLM evaluation
OpenCompass: A Universal Evaluation Platform for Large Language Models
-
Transformers rewrite non-attention ops as GEMM epilogues
CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs
-
Transformers rewritten as GEMM epilogue programs
CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs
-
Small abstract spaces enable RL generalization to larger tasks
Smaller Abstract State Spaces Enable Cross-Scale Generalization in Reinforcement Learning
-
GMM curriculum cuts PINN errors on PDEs by up to 98%
From Simple to Complex: Curriculum-Guided Physics-Informed Neural Networks via Gaussian Mixture Models
-
Backdoor attack hits near-100% success on masked diffusion LMs
Backdooring Masked Diffusion Language Models
-
Python framework unifies XAI methods for ECG models
ExECG: An Explainable AI Framework for ECG models
-
Proxy of post-target continuations boosts time series forecasts
Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecasting
-
Local distance graphs recover global Euclidean embeddings
Euclidean Embedding of Data Using Local Distances
-
Post-training lifts video models' physical consistency
PhyWorld: Physics-Faithful World Model for Video Generation
-
Centralized critic removes action-sampling variance in self-play RL
GAE Falls Short in Imperfect-Information Self-Play Reinforcement Learning
-
Quantum hybrid raises F1 when UAV detectors drop contextual proxies
Quantum Machine Learning for Cyber-Physical Anomaly Detection in Unmanned Aerial Vehicles: A Leakage-Free Evaluation with Proxy-Audited Feature Sets
-
Regime gate improves time series forecast accuracy under shifts
DeRegiME: Deep Regime Mixtures for Probabilistic Forecasting under Distribution Shift
-
Method reduces age bias in medical image classification by decorrelating difficulty
Robust Mitigation of Age-Dependent Confounding Effects via Sample-Difficulty Decorrelation
-
Step-level scores flag reasoning errors in closed LLMs
Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution
-
LLM Uncertainty Scores Only Measure Output Consistency
Position: Uncertainty Quantification in LLMs is Just Unsupervised Clustering
-
Regularizer cuts demographic gaps in medical image AI
Worst-Group Equalized Odds Regularization for Multi-Attribute Fair Medical Image Classification
-
RL on All of Us data prescribes steadier higher daily steps
Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions
-
Quantized model cuts brain tumor AI size by 6x with same accuracy
Quantized Machine Learning Models for Medical Imaging in Low-Resource Healthcare Settings
-
PneumoNet hits 86.6% accuracy with 1.4% forgetting across device shifts
On-Device Continual Learning with Dual-Stage Buffer and Dynamic Loss for Point-of-Care Pneumonia Diagnosis
-
Multi-head attention error falls as subspaces decorrelate
Multi-Head Attention as Ensemble Nadaraya-Watson Estimation: Variance Reduction, Decorrelation, and Optimal Head Diversity
-
SPRT cuts LLM debate calls 3.7x on GSM8K at 97% accuracy
Sequential Consensus for Multi-Agent LLM Debates: A Wald-SPRT compute governor with calibration-based failure detection
-
Action-gap certificate certifies greedy goal reach in sparse planning
Planner-Admissible Graph-PDE Value Extensions for Sparse Goal-Conditioned Planning
-
Drones with machine learning aid meteorite recovery
A Cloud-Based Tool for Meteorite Recovery Using Drones and Machine Learning
-
Exponential activations let RBMs capture strong higher-order terms
Activation Functions, Statistics and Learning of Higher-Order Interactions in Restricted Boltzmann Machines
-
Retrieval memory sharpens forecasts for new delivery zones
Bridge: Retrieval-Augmented Spatiotemporal Modeling for Urban Delivery Demand
-
Higher-order Langevin dynamics reduce memorization in diffusion models
Reducing Diffusion Model Memorization with Higher Order Langevin Dynamics
-
Reward heuristics tune quadrotor RL policies for fast or slow settling
A Heuristic Approach for Performance Tuning in RL-based Quadrotor Control via Reward Design and Termination Conditions
-
AI agents produce 117 papers but none clear top-tier bar
How Far Are We From True Auto-Research?
-
Wrapper gives pathwise risk control for updating LLMs
Conformal Selective Acting: Anytime-Valid Risk Control for RLVR-Trained LLMs
4 Piths -
Total capacity of stationary physical systems predicts ML performance
Information Processing Capacity of Stationary Physical Systems: Theory, Data-efficient Estimation Methods, and Photonic Demonstration
-
Total IPC of stationary systems bounds to readout count and predicts ML results
Information Processing Capacity of Stationary Physical Systems: Theory, Data-efficient Estimation Methods, and Photonic Demonstration
-
Sparse matrix bank gives SSMs dense-model expressivity
Flash PD-SSM: Memory-Optimized Structured Sparse State-Space Models
-
Low-rank bandits recover drifting subspaces from scalar rewards
Catching a Moving Subspace: Low-Rank Bandits Beyond Stationarity
4 Piths -
Benign rewriting lifts LLM safety against poisoning by 51 percent
Be Kind, Rewrite: Benign Projections via Rewriting Defend Against LLM Data Poisoning Attacks
-
Pareto points minimize forgetting on conflicting tasks
PMF-CL: Pareto-Minimal-Forgetting Continual Learner for Conflicting Tasks
-
Local attack and support calls stabilize global argument rankings
GRASP: Deterministic argument ranking in interaction graphs
-
One model trained on text and time series matches both specialists
Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding
-
Smartphone teleop rivals specialized hardware for robot demos
COBALT: Crowdsourcing Robot Learning via Cloud-Based Teleoperation with Smartphones
-
Smartphones collect 7500 robot demos in five days
COBALT: Crowdsourcing Robot Learning via Cloud-Based Teleoperation with Smartphones
-
Causal latents shown identifiable in multimodal partial-sharing setups
Identifiable Multimodal Causal Representation Learning under Partial Latent Sharing
-
Text-encoded context boosts ECG pathology classification
CLIC: Contextual Language-Informed Cardiac Pathology Classification