Research map / Reinforcement learning / Imitation learning
Chain of Thought Imitation with Procedure Cloning
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 |
|---|---|---|---|
| predict BFS visited-array procedure observations before the action (PC) instead of vanilla BC | Procedure Cloning · Discrete maze 32x32 test mazes | success rate: 0\% -> better (vanilla BC completely fails) | Helped |
| use procedure as auxiliary prediction task (Aux BC) or data augmentation (Aug BC) instead of PC (ablation) | Behavioral cloning on discrete maze · Discrete maze 16x16 vs 32x32 | success rate | Mixed |
| PC instead of vanilla BC, Aug BC and Aux BC | AntMaze 10x10 and 13x13 · AntMaze test mazes | success rate | Helped |
| PC instead of implicit BC with early stopping | Bimanual sweep · 1000 expert demonstrations | success: 78.2\% -> 83.9\% | Helped |
| PC instead of BC on MinAtar with sticky actions and difficulty ramping | MCTS-expert imitation · MinAtar test environments | average episode reward | Helped |
| add a Value Iteration Network style planning module to the policy instead of PC (ablation) | BC on discrete maze · Discrete maze test mazes | success rate | No clear effect |
Closest papers by meaning
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.89)
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.89)
- Diffusion Model-Augmented Behavioral Cloning (ICML 2024 · similarity 0.88)
- A Smooth Sea Never Made a Skilled SAILOR: Robust Imitation via Learning to Search (NeurIPS 2025 · similarity 0.88)
- When Does Interaction Help? Representational Tradeoffs in Value-Based Imitation Learning (NeurIPS 2026 · similarity 0.87)
- Chain-of-Thought Predictive Control (ICML 2024 · 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.