Learning to search: From weak methods to domain-specific heuristics (1985)
| Venue: | Cognitive Science |
| Citations: | 37 - 3 self |
BibTeX
@ARTICLE{Langley85learningto,
author = {Pat Langley},
title = {Learning to search: From weak methods to domain-specific heuristics},
journal = {Cognitive Science},
year = {1985},
pages = {217--260}
}
Years of Citing Articles
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
Learning from experience involves three distinct components-generating behavior, assigning credit, and modifying behavior. We discuss these com-ponents in the context of learning scorch heuristics, olong with the types of learning that con occur. We then focus on SAGE, o system that improves its search strategies with practice. The program is implemented OS o production system, and learns by creating and strengthening rules for proposing moves. SAGE incorporates five different heuristics for assigning credit and blame, and employs a discrimination process to direct its search through the space of rules. The system has shown its generality by leorning heuristics for directing search in six different task domoins. In oddition to improving its search behavior on practice problems, SAGE is able to tronsfer its expertise to scaled-up versions of o task, and in one case. transfers its acquired search strategy to problems with different initial and goal stotes.
