Research map / Reinforcement learning / Imitation learning
Imitating Past Successes can be Very Suboptimal
Benjamin Eysenbach, Soumith Udatha, Ruslan Salakhutdinov, Sergey Levine
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add normalization (remove averaging step) to OCBC | OCBC/GCSL · GCSL benchmark original tasks | normalized return: +3\% | No clear effect |
| add normalization to OCBC under skewed action distribution | OCBC/GCSL · GCSL benchmark skewed actions | normalized return: +149\% | Improved |
| add normalization to OCBC under biased goal distribution | OCBC/GCSL · GCSL benchmark biased goals | normalized return: +62\% | Improved |
| add normalization to iterated OCBC | iterated OCBC · 3-action bandit | policy optimality for tasks e2 and e3 | Improved |
| add normalization to OCBC (skewed commanded goals 9:1) | OCBC · 2D navigation, orange goal | steps to reach goal | Improved |
| add normalization to OCBC | OCBC · 2D navigation, blue goal | steps to reach goal | No clear effect |
| label many states as failures (discard trials) in OCBC (ablation) | OCBC · 1D gridworld, high-data regime | average reward | Improved |
| label many states as failures (discard trials) in OCBC (ablation) | OCBC · 1D gridworld, low-data regime | average reward | Hurt |
| filtered relabeling with importance-weight clipping epsilon (ablation) | Normalized OCBC · Pointmass rooms (random and biased-goal datasets) | performance | No clear effect |
Closest papers by meaning
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.87)
- Blending Imitation and Reinforcement Learning for Robust Policy Improvement (ICLR 2024 · similarity 0.87)
- EvIL: Evolution Strategies for Generalisable Imitation Learning (ICML 2024 · similarity 0.87)
- SEABO: A Simple Search-Based Method for Offline Imitation Learning (ICLR 2024 · similarity 0.87)
- Learning When to Stop: Selective Imitation Learning Under Arbitrary Dynamics Shift (NeurIPS 2026 · similarity 0.87)
- Enhancing Online Reinforcement Learning with Meta-Learned Objective from Offline Data (AAAI 2025 · similarity 0.87)
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.
