sample efficiency
Sample efficiency indicates how effectively a learning algorithm makes use of the available training data to achieve high performance. High sample efficiency means that a model can learn well from fewer examples.
- 3D Equivariant Visuomotor Policy Learning via Spherical Projection
- A Differential and Pointwise Control Approach to Reinforcement Learning
- A Plug-and-Play Query Synthesis Active Learning Framework for Neural PDE Solvers
- A Practical Guide for Incorporating Symmetry in Diffusion Policy
- A Smooth Sea Never Made a Skilled SAILOR: Robust Imitation via Learning to Search
- Adaptive Prediction-Powered AutoEval with Reliability and Efficiency Guarantees
- Amortized Active Generation of Pareto Sets
- Avoiding exp(R) scaling in RLHF through Preference-based Exploration
- Bootstrap Off-policy with World Model
- Bridging Equivariant GNNs and Spherical CNNs for Structured Physical Domains
- Composite Flow Matching for Reinforcement Learning with Shifted-Dynamics Data
- Convergent Functions, Divergent Forms
- Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning
- DecoyDB: A Dataset for Graph Contrastive Learning in Protein-Ligand Binding Affinity Prediction
- EDELINE: Enhancing Memory in Diffusion-based World Models via Linear-Time Sequence Modeling
- Efficient Bayesian Experiment Design with Equivariant Networks
- Flow Equivariant Recurrent Neural Networks
- From Kolmogorov to Cauchy: Shallow XNet Surpasses KANs
- Generalization or Hallucination? Understanding Out-of-Context Reasoning in Transformers
- Imitation Beyond Expectation Using Pluralistic Stochastic Dominance
- Imitation Learning with Temporal Logic Constraints
- Improving Reward Models with Proximal Policy Exploration for Preference-Based Reinforcement Learning
- LaRes: Evolutionary Reinforcement Learning with LLM-based Adaptive Reward Search
- Learning (Approximately) Equivariant Networks via Constrained Optimization
- Learning (Approximately) Equivariant Networks via Constrained Optimization
- Learning Interactive World Model for Object-Centric Reinforcement Learning
- Learning and Planning Multi-Agent Tasks via an MoE-based World Model
- Learning from Demonstrations via Capability-Aware Goal Sampling
- Meta-D2AG: Causal Graph Learning with Interventional Dynamic Data
- Mitra: Mixed Synthetic Priors for Enhancing Tabular Foundation Models
- Nabla-R2D3: Effective and Efficient 3D Diffusion Alignment with 2D Rewards
- On scalable and efficient training of diffusion samplers
- PHYBench: Holistic Evaluation of Physical Perception and Reasoning in Large Language Models
- Parameterized Synthetic Text Generation with SimpleStories
- Preference Learning with Response Time: Robust Losses and Guarantees
- Preference-based Reinforcement Learning beyond Pairwise Comparisons: Benefits of Multiple Options
- Provable Sample-Efficient Transfer Learning Conditional Diffusion Models via Representation Learning
- REASONING COMPILER: LLM-Guided Optimizations for Efficient Model Serving
- Rebalancing Return Coverage for Conditional Sequence Modeling in Offline Reinforcement Learning
- Regret-Optimal Q-Learning with Low Cost for Single-Agent and Federated Reinforcement Learning
- Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach
- ResponseRank: Data-Efficient Reward Modeling through Preference Strength Learning
- Retrospective In-Context Learning for Temporal Credit Assignment with Large Language Models
- Revisiting Multi-Agent World Modeling from a Diffusion-Inspired Perspective
- Sample-Efficient Multi-Round Generative Data Augmentation for Long-Tail Instance Segmentation
- Scaling Laws for Robust Comparison of Open Foundation Language-Vision Models and Datasets
- Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization
- Self-Improving Embodied Foundation Models
- ShiQ: Bringing back Bellman to LLMs
- Social World Model-Augmented Mechanism Design Policy Learning
- Staggered Environment Resets Improve Massively Parallel On-Policy Reinforcement Learning
- Succeed or Learn Slowly: Sample Efficient Off-Policy Reinforcement Learning for Mobile App Control
- Synthetic-powered predictive inference
- The Computational Advantage of Depth in Learning High-Dimensional Hierarchical Targets
- Thompson Sampling in Function Spaces via Neural Operators
- Time Reversal Symmetry for Efficient Robotic Manipulations in Deep Reinforcement Learning
- Turning Sand to Gold: Recycling Data to Bridge On-Policy and Off-Policy Learning via Causal Bound
- Uncertainty-Guided Exploration for Efficient AlphaZero Training
- Understanding Generalization in Physics Informed Models through Affine Variety Dimensions
- Uni-RL: Unifying Online and Offline RL via Implicit Value Regularization
- Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations
- Zero-shot World Models via Search in Memory