Research map / Reinforcement learning / Imitation learning
Hierarchical Imitation and 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 hierarchical guidance to restrict low-level learning to relevant states and query low-level labels only for failed subtasks | Maze navigation imitation learning · Maze navigation | expert cost for a given success rate | Helped |
| Use hierarchical behavioral cloning instead of flat behavioral cloning | Maze navigation · Maze navigation | success rate | Helped |
| Use hg-DAgger instead of flat DAgger | Maze navigation · Maze navigation | low-level expert cost | Helped |
| Use high-level imitation guidance with low-level reinforcement learning instead of hierarchical RL without a high-level expert | Hierarchical learning · Montezuma's Revenge | external rewards | Helped |
| Use high-level imitation guidance with low-level reinforcement learning instead of hierarchical RL without a high-level expert | Hierarchical learning · Maze navigation | success rate | Helped |
| Delay accumulating low-level replay-buffer experience until the first positive pseudo-reward (ablation) | hg-DAgger/Q · Montezuma's Revenge | learning speed | Helped |
| Delay accumulating low-level replay-buffer experience until the first positive pseudo-reward (ablation) | h-DQN · Montezuma's Revenge | learning performance | Helped |
| Stop training a subpolicy once its success rate exceeds 90% (ablation) | hg-DAgger/Q · Montezuma's Revenge | learning stability | Helped |
Closest papers by meaning
- Hybrid Inverse Reinforcement Learning (ICML 2024 · similarity 0.90)
- Blending Imitation and Reinforcement Learning for Robust Policy Improvement (ICLR 2024 · similarity 0.90)
- Beyond-Expert Performance with Limited Demonstrations: Efficient Imitation Learning with Double Exploration (ICLR 2025 · similarity 0.89)
- RLIF: Interactive Imitation Learning as Reinforcement Learning (ICLR 2024 · similarity 0.89)
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.89)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.88)
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.