Relevance Cuts: Localizing the Search
Abstract
Humans can effectively navigate through large search spaces, enabling them to solve problems with daunting complexity. This is largely due to an ability to successfully distinguish between relevant and irrelevant actions (moves). In this paper we present a new single-agent search pruning technique that is based on a move’s influence. The influence measure is a crude form of relevance in that it is used to differentiate between local (relevant) moves and non-local (not relevant) moves, with respect to the sequence of moves leading up to the current state. Our pruning technique uses the m previous moves to decide if a move is relevant in the current context and, if not, to cut it off. This technique results in a large reduction in the search effort required to solve Sokoban problems.
Keywords
single-agent search heuristic search Sokoban local search IDA*Preview
Unable to display preview. Download preview PDF.
References
- [1]G. Adelson-Velskiy, V. Arlazarov, and M. Donskoy. Some methods of controlling the tree search in chess programs. Artificial Intelligence, 6(4):361–371, 1975.MATHCrossRefGoogle Scholar
- [2]J. Culberson. Sokoban is PSPACE-complete. Technical Report TR97-02, Department of Computing Science, University of Alberta, Edmonton, Alberta, Canada, 1997. ftp://ftp.cs.ualberta.ca/pub/TechReports/1997/TR97-02.Google Scholar
- [3]D. Dor and U. Zwick. SOKOBAN and other motion planning problems, 1995. At: http://www.math.tau.ac.il/~ddorit.
- [4]M. Ginsberg. Essentials in Artificial Intelligence. Morgan Kaufman Publishers, San Francisco, 1993.Google Scholar
- [5]G. Goetsch and M.S. Campbell. Experiments with the null-move heuristic. In T.A. Marsland and J. Schaeffer, editors, Computers, Chess, and Cognition, pages 159–181, New York, 1990. Springer-Verlag.Google Scholar
- [6]A. Junghanns and J. Schaeffer. Single-agent search in the presence of deadlock. In AAAI-98, pages 419–424, Madison/WI, USA, July 1998.Google Scholar
- [7]A. Junghanns and J. Schaeffer. Sokoban: Evaluating standard single-agent search techniques in the presence of deadlock. In Advances in Artificial Intelligence, pages 1–15. Springer Verlag, 1998.Google Scholar
- [8]R.E. Korf. Depth-first iterative-deepening: An optimal admissible tree search. Artificial Intelligence, 27(1):97–109, 1985.MATHCrossRefMathSciNetGoogle Scholar
- [9]R.E. Korf. Macro-operators: A weak method for learning. Artificial Intelligence, 26(1):35–77, 1985.MATHCrossRefMathSciNetGoogle Scholar
- [10]R.E. Korf. Finding optimal solutions to Rubik’s Cube using pattern databases. In AAAI-97, pages 700–705, 1997.Google Scholar
- [11]J. Schaeffer. Experiments in Search and Knowledge. PhD thesis, Univ. of Waterloo, Canada, 1986.Google Scholar
- [12]G. Wilfong. Motion planning in the presence of movable obstacles. In 4th ACM Symposium on Computational Geometry, pages 279–288, 1988.Google Scholar