Research map / Reinforcement learning / Offline RL
A Policy-Guided Imitation Approach for Offline Reinforcement Learning
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 |
|---|---|---|---|
| use POR instead of IQL | offline RL · D4RL antmaze mean | normalized score: 63.6 -> 76.2 | Helped |
| use POR instead of best RL baseline | offline RL · D4RL antmaze mean | normalized score: 58.4 -> 76.2 | Helped |
| use POR instead of best RL baseline | offline RL · D4RL locomotion mean | normalized score: 63.5 -> 65.6 | Helped |
| POR vs TD3+BC | offline RL · D4RL walker2d-medium | normalized score: 83.7 -> 81.1 | Hurt |
| use sparse guide policy variant (ablation) | POR · D4RL antmaze | normalized score: 86.3 vs 71.3 on umaze-diverse; 53.4 vs 73.4 on large-diverse | Mixed |
| use small network for execute-policy (ablation) | POR · D4RL antmaze mean | normalized score: 76.2 -> 63.8 | Hurt |
| use small network for execute-policy (ablation) | POR · D4RL locomotion mean | normalized score: 65.7 -> 57.0 | Hurt |
| train execute-policy on top-X% trajectories only (ablation) | POR · D4RL antmaze mean | normalized score: 76.2 -> 52.4 | Hurt |
| train execute-policy on top-X% trajectories only (ablation) | POR · D4RL locomotion mean | normalized score: 65.7 -> 60.9 | Hurt |
| retrain guide-policy on expert plus additional action-free data (mix) (ablation) | POR · hopper/walker2d with extra data | normalized score | Helped |
Closest papers by meaning
- Is Value Learning Really the Main Bottleneck in Offline RL? (NeurIPS 2024 · similarity 0.92)
- A Clean Slate for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.92)
- Enhancing Online Reinforcement Learning with Meta-Learned Objective from Offline Data (AAAI 2025 · similarity 0.92)
- Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL (ICML 2024 · similarity 0.92)
- Dual RL: Unification and New Methods for Reinforcement and Imitation Learning (ICLR 2024 · similarity 0.92)
- 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.