Research map / Reinforcement learning / Visual and representation RL
Knowledge Transfer for Deep Reinforcement Learning with Hierarchical Experience Replay
Haiyan Yin, Sinno Pan
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 |
|---|---|---|---|
| proposed multi-task distillation architecture vs DIST | multi-task policy distillation · Atari 10 games | % of teacher score (geometric mean): 92.41 -> 103.8 | Improved |
| proposed multi-task distillation architecture vs AMN | multi-task policy distillation · Atari 10 games | % of teacher score (geometric mean): 93.5 -> 103.8 | Improved |
| proposed architecture vs DIST/AMN | multi-task policy distillation · Atari 10 games | steps to converge: 1.5M vs at least 2.5M steps | Improved |
| use hierarchical prioritized replay instead of uniform sampling (ablation) | proposed architecture · Breakout, Q*bert | steps to reach target score: approximately 50% of the steps | Improved |
| use hierarchical prioritized replay instead of uniform sampling (ablation) | proposed architecture · Freeway, Pong | score | No clear effect |
| vary partition size p (5, 10, 15) (ablation) | H-PR · Atari multi-task | learning speed | No clear effect |
Closest papers by meaning
- In-context Reinforcement Learning with Algorithm Distillation (ICLR 2023 · similarity 0.88)
- Progressive Reinforcement Learning with Distillation for Multi-Skilled Motion Control (ICLR 2018 · similarity 0.88)
- Transfer Learning in Deep Reinforcement Learning: A Survey (TPAMI 2023 · similarity 0.88)
- Multi-Task Deep Reinforcement Learning with PopArt (AAAI 2019 · similarity 0.88)
- Actor Prioritized Experience Replay (Abstract Reprint) (AAAI 2024 · similarity 0.87)
- Turning Sand to Gold: Recycling Data to Bridge On-Policy and Off-Policy Learning via Causal Bound (NeurIPS 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.
