Research map / Probabilistic and scientific modeling / Bayesian optimization
Learning to Prune Dominated Action Sequences in Online Black-Box Planning
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 |
|---|---|---|---|
| add DASA2 dominated-sequence avoidance (vs default 18 actions) | p-IW(1) · 53 Atari games, 10000 frames | #Best games: 6 -> 22 | Helped |
| add DASA2 (vs hand-coded restricted action set) | p-IW(1) · 53 Atari games, 10000 frames | #Best games: 10 -> 22 | Helped |
| use 2-step sequences (DASA2) instead of 1-step (DASA1) (ablation) | p-IW(1) with DASA · 53 Atari games, 10000 frames | #Best games: 10 -> 22 | Helped |
| probabilistic avoidance (DASA1) instead of static pruning (DASP1) (ablation) | p-IW(1) · 53 Atari games, 10000 frames | #Best games: 4 -> 10 | Helped |
| add DASP1 static pruning vs default action set | p-IW(1) · 53 Atari games, 10000 frames | #Best games: 6 -> 4 | No clear effect |
| add DASA2 pruning | p-IW(1) · 53 Atari games, 10000 frames | avg nodes expanded per planning: 119.6 -> 254.9 | Helped |
| add DASA2 pruning | p-IW(1) · 53 Atari games, 10000 frames | avg search depth: 34.1 -> 82.8 | Helped |
| add DASA2 with small budget of 2000 frames | p-IW(1) · 53 Atari games, 2000 frames | #Best games: 5 -> 24 | Helped |
| add DASA2 | IW(1) · 53 Atari games, 10000 frames | #Best games: 7 -> 22 | Helped |
| add DASA2 | BrFS · 53 Atari games, 10000 frames | #Best games: 6 -> 18 | Helped |
| add DASA2 with 72 actions (spurious buttons) | p-IW(1) · 53 Atari games, extended action set | #Best games: 16 -> 39 | Helped |
| add DASA2 vs restricted set in avg nodes expanded | p-IW(1) · 53 Atari games, 10000 frames | avg nodes expanded per planning: 234.0 -> 254.9 | Helped |
Closest papers by meaning
- Scaling Offline Model-Based RL via Jointly-Optimized World-Action Model Pretraining (ICLR 2025 · similarity 0.83)
- Amortized Planning with Large-Scale Transformers: A Case Study on Chess (NeurIPS 2024 · similarity 0.82)
- Implicit Search via Discrete Diffusion: A Study on Chess (ICLR 2025 · similarity 0.82)
- The Update-Equivalence Framework for Decision-Time Planning (ICLR 2024 · similarity 0.82)
- Beyond-Expert Performance with Limited Demonstrations: Efficient Imitation Learning with Double Exploration (ICLR 2025 · similarity 0.82)
- Mastering Board Games by External and Internal Planning with Language Models (ICML 2025 · similarity 0.82)
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.