Research map / Probabilistic and scientific modeling / Molecules and proteins
On the Analysis of Complex Backup Strategies in Monte Carlo Tree Search
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 off-policy MaxMCTS(lambda=0) instead of Monte Carlo backup | MCTS with uniform selection · IPC Elevators | mean reward: -109.67 -> -83.95 | Helped |
| use MaxMCTS(lambda=0.3) instead of Monte Carlo backup | MCTS with uniform selection · IPC Recon | mean reward: 0.0 -> 1.67 | Helped |
| use MaxMCTS(lambda=0.7) instead of Monte Carlo backup | MCTS with uniform selection · IPC Academic Advising | mean reward: -131.88 -> -127.78 | Helped |
| use on-policy MCTS(lambda) or MCTS-gamma instead of Monte Carlo backup | MCTS · IPC domains | mean reward | No clear effect |
| use parameter-free MaxMCTS-gamma instead of Monte Carlo backup | MCTS · 12 IPC domains | mean reward | Helped |
| vary lambda in MaxMCTS(lambda) from 1 to low values (ablation) | MaxMCTS(lambda) · IPC domains | mean reward | Mixed |
| change backup to MaxMCTS(0) instead of tuning UCT action selection with Monte Carlo (ablation) | MCTS · IPC Elevators and Recon | mean reward | Helped |
| use MaxMCTS(lambda=0.4) instead of Monte Carlo with 3 zero-reward terminal states (ablation) | MCTS in grid world · 9x9 grid, 3 terminal states | average episodic reward: 11.3 -> 85.1 | Helped |
| use MaxMCTS(lambda=0.4) instead of Monte Carlo with no obstacles (ablation) | MCTS in grid world · 9x9 grid, 0 terminal states | average episodic reward: 90.4 -> 88.7 | Hurt |
| use complex backups instead of Monte Carlo | MCTS · computation time per episode | seconds: 196 -> 202 | No clear effect |
Closest papers by meaning
- Monte-Carlo Tree Search with Uncertainty Propagation via Optimal Transport (ICML 2025 · similarity 0.91)
- Online Robust Reinforcement Learning Through Monte-Carlo Planning (ICML 2025 · similarity 0.91)
- Monte Carlo Tree Search in the Presence of Transition Uncertainty (AAAI 2024 · similarity 0.90)
- Graph-Based Stochastic-Power-UCT: Monte-Carlo Graph Search with Power Mean Estimation (NeurIPS 2026 · similarity 0.88)
- Monte Carlo Tree Diffusion for System 2 Planning (ICML 2025 · similarity 0.88)
- Power Mean Estimation in Stochastic Continuous Monte-Carlo Tree Search (ICML 2025 · 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.