Presentation 1997/1/22
Stochastic Node Caching for memory bounded search
Teruhisa Miura, Toru Ishida,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Linear space search algorithms only keep nodes on the current search path. But they may revisit the same node repeatedly, and thus it may take unpractically long time to find a solution. In this paper, we propose a simple and effective algorithm called Stochastic Node Caching for reducing revisits. It keeps a node in memory with some fixed probability when the node is expanded. To demonstrate the effectiveness of Stochastic Node Caching, we compared it with MREC. We applied them to the 15 puzzle and the multiple sequence alignment. On the multiple sequence alignment, the achievement was about 70% reduction in the total number of visits compared to MREC.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) linear-space search / memory-bounded search / 15 Puzzle / IDA^* / MREC
Paper # AI96-35
Date of Issue

Conference Information
Committee AI
Conference Date 1997/1/22(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Artificial Intelligence and Knowledge-Based Processing (AI)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Stochastic Node Caching for memory bounded search
Sub Title (in English)
Keyword(1) linear-space search
Keyword(2) memory-bounded search
Keyword(3) 15 Puzzle
Keyword(4) IDA^*
Keyword(5) MREC
1st Author's Name Teruhisa Miura
1st Author's Affiliation Department of Information Science, Kyoto University()
2nd Author's Name Toru Ishida
2nd Author's Affiliation Department of Information Science, Kyoto University
Date 1997/1/22
Paper # AI96-35
Volume (vol) vol.96
Number (no) 453
Page pp.pp.-
#Pages 8
Date of Issue