Research map / Reinforcement learning / Visual and representation RL
A Novel Automated Curriculum Strategy to Solve Hard Sokoban Planning Instances
Dieqiao Feng, Carla P. Gomes, Bart Selman
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 curiosity-driven reward (RND) to MCTS (ablation) | BL · 225 hard Sokoban instances | instances solved: 30 -> 52 | Improved |
| replace CNN with graph network (ablation) | BL+CR · 225 hard Sokoban instances | instances solved: 52 -> 72 | Improved |
| add bandit (difficulty quantum momentum) subcase selection (ablation) | BL+CR+GN · 225 hard Sokoban instances | instances solved: 72 -> 115 | Improved |
| mix subcases of all 225 hard instances (MIX) (ablation) | BL+CR+GN+BD · 225 hard Sokoban instances | instances solved: 115 -> 146 | Improved |
| add MIX with CNN rather than graph net (ablation) | BL+CR+BD · 225 hard Sokoban instances | instances solved: 103 -> 105 | No clear effect |
| add 3137 extra unrelated instances to curriculum pool (ablation) | best strategy · 225 hard Sokoban instances | instances solved: 146 -> 179 | Improved |
| add 500 extra instances (ablation) | best strategy · 225 hard Sokoban instances | instances solved: 146 -> 148 | Improved |
| add 1000 extra instances (ablation) | best strategy · 225 hard Sokoban instances | instances solved: 148 -> 156 | Improved |
| bandit selection instead of uniform random selection (ablation) | curriculum RL · 225 hard Sokoban instances | instances solved: 72 -> 115 | Improved |
| train on data from failed tries (ablation) | RL agent · 225 hard Sokoban instances | final performance | No clear effect |
Closest papers by meaning
- Self-Paced Deep Reinforcement Learning (NeurIPS 2020 · similarity 0.88)
- Understanding the Complexity Gains of Single-Task RL with a Curriculum (ICML 2023 · similarity 0.88)
- A Probabilistic Interpretation of Self-Paced Learning with Applications to Reinforcement Learning (JMLR 2021 · similarity 0.88)
- Learning to Search and Searching to Learn for Generalization in Planning (ICML 2026 · similarity 0.88)
- Reinforcement learning with combinatorial actions for coupled restless bandits (ICLR 2025 · similarity 0.86)
- Diffusion-based Curriculum Reinforcement Learning (NeurIPS 2024 · similarity 0.85)
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.
