Research map / Reinforcement learning / Policy gradient and actor-critic
Self-Imitation 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 |
|---|---|---|---|
| Add self-imitation learning to A2C | A2C · 49 Atari games | median human-normalized score: 96.1% -> 138.7% | Helped |
| Add self-imitation learning to A2C | A2C · Hard-exploration Atari games | score | Helped |
| Use A2C+SIL instead of A2C | A2C · Venture | score: 0 -> 0 | No clear effect |
| Add self-imitation learning to A2C | A2C · Key-Door-Treasure | learning speed | Helped |
| Combine self-imitation learning with count-based exploration | A2C · Apple-Key-Door-Treasure | optimal policy learning | Helped |
| Add prioritized experience replay with ACER objective | A2C · 49 Atari games | median human-normalized score: 96.1% -> 46.8% | Hurt |
| Add self-imitation learning to PPO | PPO · MuJoCo: Swimmer, Walker2d, and Ant | reward | Helped |
| Add self-imitation learning to PPO | PPO · Delayed-reward MuJoCo tasks | reward | Helped |
| Reduce the number of SIL updates per iteration or lower the SIL loss weight later in training (ablation) | A2C+SIL · James Bond and Star Gunner | performance | Helped |
Closest papers by meaning
- Beyond-Expert Performance with Limited Demonstrations: Efficient Imitation Learning with Double Exploration (ICLR 2025 · similarity 0.88)
- IL-SOAR : Imitation Learning with Soft Optimistic Actor cRitic (ICML 2025 · similarity 0.87)
- Mimicking To Dominate: Imitation Learning Strategies for Success in Multiagent Games (NeurIPS 2024 · similarity 0.86)
- Adversarial Imitation Learning via Boosting (ICLR 2024 · similarity 0.86)
- Hybrid Inverse Reinforcement Learning (ICML 2024 · similarity 0.85)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.85)
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.