MétaCan
Menu
Back to cohort
Record W4312891993 · doi:10.1609/icaps.v32i1.19804

Biased Exploration for Satisficing Heuristic Search

2022· article· en· W4312891993 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.
fundA Canadian funder is recorded on the work.

Bibliographic record

VenueProceedings of the International Conference on Automated Planning and Scheduling · 2022
Typearticle
Languageen
FieldComputer Science
TopicAI-based Problem Solving and Planning
Canadian institutionsUniversity of Toronto
FundersNatural Sciences and Engineering Research Council of Canada
KeywordsSatisficingIncremental heuristic searchHeuristicMathematical optimizationBeam searchBest-first searchNode (physics)Computer scienceConsistent heuristicNull-move heuristicGreedy algorithmMathematicsSearch algorithmAlgorithmArtificial intelligenceEngineering

Abstract

fetched live from OpenAlex

Satisficing heuristic search such as greedy best-first search (GBFS) suffers from local minima, regions where heuristic values are inaccurate and a good node has a worse heuristic value than other nodes. Search algorithms that incorporate exploration mechanisms in GBFS empirically reduce the search effort to solve difficult problems. Although some of these methods entirely ignore the guidance of a heuristic during their exploration phase, intuitively, a good heuristic should have some bound on its inaccuracy, and exploration mechanisms should exploit this bound. In this paper, we theoretically analyze what a good node is for satisficing heuristic search algorithms and show that the heuristic value of a good node has an upper bound if a heuristic satisfies a certain property. Then, we propose biased exploration mechanisms which select lower heuristic values with higher probabilities. In the experiments using synthetic graph search problems and classical planning benchmarks, we show that the biased exploration mechanisms can be useful. In particular, one of our methods, Softmin-Type(h), significantly outperforms other GBFS variants in classical planning and improves the performance of Type-LAMA, a state-of-the-art classical planner.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.681
Threshold uncertainty score0.438

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0010.000
Scholarly communication0.0000.000
Open science0.0010.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.082
GPT teacher head0.309
Teacher spread0.227 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it