Research map / Reinforcement learning / Offline RL
Bayes Adaptive Monte Carlo Tree Search for Offline Model-based Reinforcement Learning
Jiayu Chen, Le Xu, Wentse Chen, Jeff Schneider
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 |
|---|---|---|---|
| add Bayesian belief adaptation over ensemble members (ablation) | offline MBRL world model · D4RL imagined rollouts | prediction error: 10 of 12 tasks lower error | Improved |
| add belief adaptation (ablation) | world model · D4RL Hopper random | overall prediction error: 0.106 -> 0.122 | Hurt |
| add belief adaptation (ablation) | world model · D4RL medium Hopper | overall prediction error: 34.40 -> 9.727 | Improved |
| add belief adaptation to Optimized (BA-MBRL) (ablation) | Optimized · Tokamak tracking (average) | average return: -70.98 -> -28.03 | Improved |
| add continuous BAMCP planner on top of BA-MBRL (ablation) | BA-MBRL · Tokamak tracking (average) | average return: -28.03 -> -20.61 | Improved |
| replace SAC distillation with supervised learning (ablation) | BA-MCTS · Tokamak tracking (average) | average return: -20.61 -> -24.11 | Mixed |
| replace SAC distillation with supervised learning (ablation) | BA-MCTS · D4RL Walker2d | stability | Hurt |
| remove pessimistic reward penalty (lambda=0) (ablation) | BA-MCTS · D4RL MuJoCo (average) | normalized score | Hurt |
| use BA-MCTS instead of MOBILE | offline MBRL · D4RL MuJoCo (average) | normalized score: 76.50 -> 80.25 | Improved |
| use BA-MCTS instead of APE-V (model-free BAMDP) | offline RL · D4RL MuJoCo | normalized score: 72.65 -> 80.25 | Improved |
| use BA-MCTS instead of Sampled EfficientZero search | offline RL · D4RL MuJoCo | normalized score | Improved |
Closest papers by meaning
- OCEAN-MBRL: Offline Conservative Exploration for Model-Based Offline Reinforcement Learning (AAAI 2024 · similarity 0.90)
- Online Robust Reinforcement Learning Through Monte-Carlo Planning (ICML 2025 · similarity 0.88)
- SUMO: Search-Based Uncertainty Estimation for Model-Based Offline Reinforcement Learning (AAAI 2025 · similarity 0.88)
- Bayesian Exploration Networks (ICML 2024 · similarity 0.88)
- Bayesian Design Principles for Offline-to-Online Reinforcement Learning (ICML 2024 · similarity 0.87)
- Reflect-then-Plan: Offline Model-Based Planning through a Doubly Bayesian Lens (ICML 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.
