Research map / Learning theory and algorithms / Combinatorial and search optimization
Best-First Width Search: Exploration and Exploitation in Classical 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 |
|---|---|---|---|
| replace local GBFS with BFS(w) width-based local exploration (GBFS-LS to GBFS-W) (ablation) | GBFS with h_add · IPC-2014 | coverage: 57 -> 78 | Helped |
| replace local GBFS with BFS(w) exploration (ablation) | GBFS with h_ff · IPC-2014 | coverage: 79 -> 88 | Helped |
| add local GBFS exploration to plain GBFS (ablation) | GBFS with h_add · IPC-2014 | coverage: 47 -> 57 | Helped |
| put novelty first and heuristic as tie breaker: <w,h> instead of <h,w>-like GBFS-W (ablation) | BFWS with h_add · IPC-2014 | coverage: 78 -> 100 | Helped |
| put novelty first and heuristic as tie breaker: <w,h> instead of GBFS-W (ablation) | BFWS with h_ff · IPC-2014 | coverage: 88 -> 104 | Helped |
| add novelty as primary preference to GBFS <hL, hff> (ablation) | GBFS f3 · IPC-2014 | coverage: 99 -> 149 | Helped |
| use novelty over goal counters <w(#g,#r), #g> instead of landmark and FF heuristics | BFWS · IPC-2014 | coverage: 149 -> 192 | Helped |
| add incomplete polynomial front end (prune novelty>1) before complete BFWS back end (Dual-BFWS) | BFWS(f5) · IPC-2014 | coverage: 192 -> 225 | Helped |
| prune states with novelty > 1 in BFWS(f5) (ablation) | BFWS(f5) · IPC-2014 | coverage: 192 -> 151 | Hurt |
| swap lexicographic order to <h,w> (heuristic first) in BFWS (ablation) | BFWS(f1) · IPC-2014 | coverage | Hurt |
| use Dual-BFWS instead of Jasper | Dual-BFWS · IPC-2014 | avg plan quality: 423 -> 330 | Helped |
Closest papers by meaning
- On the Completeness of Best-First Search Variants That Use Random Exploration (AAAI 2016 · similarity 0.89)
- Width-based Lookaheads with Learnt Base Policies and Heuristics Over the Atari-2600 Benchmark (NeurIPS 2021 · similarity 0.88)
- Efficient Lookahead Encoding and Abstracted Width for Learning General Policies in Classical Planning (NeurIPS 2026 · similarity 0.87)
- Intelligent Go-Explore: Standing on the Shoulders of Giant Foundation Models (ICLR 2025 · similarity 0.87)
- Thought of Search: Planning with Language Models Through The Lens of Efficiency (NeurIPS 2024 · similarity 0.85)
- SeeA*: Efficient Exploration-Enhanced A* Search by Selective Sampling (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.