Research map / Reinforcement learning / Offline RL
Beyond Uniform Sampling: Offline Reinforcement Learning with Imbalanced Datasets
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Apply Density-ratio Weighting (DW) to uniform-sampling offline RL | CQL, IQL, and TD3BC · Imbalanced D4RL datasets with similar initial states | normalized return | Helped |
| Apply DW to offline RL training | CQL, IQL, and TD3BC · Imbalanced D4RL datasets with diverse initial states | normalized return | Helped |
| Use DW-AW rather than AW | CQL · Small imbalanced D4RL datasets | normalized return | Helped |
| Apply DW rather than uniform sampling or percentage-filtering | IQL and TD3BC · Small imbalanced D4RL datasets | normalized return | Helped |
| Apply DW rather than AW or percentage-filtering | IQL and TD3BC · Small imbalanced D4RL datasets | normalized return | No clear effect |
| Initialize DW training with AW sampling instead of uniform sampling (ablation) | DW · Imbalanced D4RL datasets with similar initial states | normalized return | Helped |
| Weight the policy optimization objective as well as the regularization term (ablation) | DW with offline RL · Imbalanced offline RL datasets | policy performance | Helped |
| Use DW rather than trajectory-level AW or percentage-filtering | Offline RL in the four-room trajectory-stitching experiment · Four-room environment | expected return | Helped |
| Apply weighted sampling or importance weighting rather than uniform sampling | Offline RL algorithms · Original D4RL datasets | normalized return | No clear effect |
Closest papers by meaning
- Harnessing Mixed Offline Reinforcement Learning Datasets via Trajectory Weighting (ICLR 2023 · similarity 0.93)
- Efficient Online Reinforcement Learning with Offline Data (ICML 2023 · similarity 0.92)
- On Sample-Efficient Offline Reinforcement Learning: Data Diversity, Posterior Sampling and Beyond (NeurIPS 2023 · similarity 0.92)
- Tackling Data Corruption in Offline Reinforcement Learning via Sequence Modeling (ICLR 2025 · similarity 0.91)
- Simple Ingredients for Offline Reinforcement Learning (ICML 2024 · similarity 0.91)
- Beyond OOD State Actions: Supported Cross-Domain Offline Reinforcement Learning (AAAI 2024 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.