Research map / Reinforcement learning / Imitation learning
Imitation Learning by 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 ILR instead of GAIL | SAC-based imitation learning · Hopper, 16 expert episodes, 500K iterations | runtime: 213m 56s -> 99m 16s | Helped |
| Use ILR instead of GMMIL | SAC-based imitation learning · Hopper, 16 expert episodes, 500K iterations | runtime: 448m 42s -> 99m 16s | Helped |
| Use ILR instead of SQIL | SAC-based imitation learning · Hopper, 16 expert episodes, 500K iterations | runtime: 94m 39s -> 99m 16s | No clear effect |
| Include the entropy term in the actor update (ablation) | SAC · PyBullet continuous control tasks | performance | No clear effect |
| Train ILR on 16 expert episodes rather than a smaller demonstration set | ILR · PyBullet continuous control tasks | extrinsic return | Helped |
| Use ILR rather than GMMIL with limited demonstrations | Imitation learning · Hopper and Walker | extrinsic return | Mixed |
| Use ILR rather than other imitation methods | Imitation learning · Ant | sample efficiency | Helped |
| Use ILR rather than other imitation methods | Imitation learning · Hopper, Walker, and HalfCheetah | sample efficiency | No clear effect |
| Use the continuous nearest-neighbor reward relaxation | ILR · Hopper | relationship between intrinsic and extrinsic return | Helped |
Closest papers by meaning
- Random Expert Distillation: Imitation Learning via Expert Policy Support Estimation (ICML 2019 · similarity 0.94)
- Imitation Learning by Estimating Expertise of Demonstrators (ICML 2022 · similarity 0.92)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.91)
- EvIL: Evolution Strategies for Generalisable Imitation Learning (ICML 2024 · similarity 0.90)
- Rethinking Inverse Reinforcement Learning: from Data Alignment to Task Alignment (NeurIPS 2024 · similarity 0.90)
- Hybrid Inverse Reinforcement Learning (ICML 2024 · similarity 0.90)
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.