Move Ordering using Neural Networks

L. Kocsis, J. Uiterwijk, J. Van Den Herik

Research output: Chapter in Book/Report/Conference proceedingChapterAcademic

Abstract

The efficiency of alpha-beta search algorithms heavily depends on the order in which the moves are examined. This paper focuses on using neural networks to estimate the likelihood of a move being the best in a certain position. The moves considered more likely to be the best are examined first. We selected Lines of Action as a testing ground. We investigated several schemes to encode the moves in a neural net- work. In the experiments, the best performance was obtained by using one output unit for each possible move of the game. The results indicate that our move-ordering approach can speed up the search with 20 to 50 percent compared with one of the best current alternatives, the history heuristic.
Original languageEnglish
Title of host publicationEngineering of Intelligent Systems. IEA/AIE 2001
EditorsL. Monostori, L. Váncza, M. Ali
PublisherSpringer, Berlin, Heidelberg
Pages45-50
Number of pages6
Volume2070
ISBN (Electronic)978-3-540-45517-2
ISBN (Print)978-3-540-42219-8
DOIs
Publication statusPublished - 2001

Publication series

SeriesLecture Notes in Computer Science
Volume2070
ISSN0302-9743

Fingerprint

Dive into the research topics of 'Move Ordering using Neural Networks'. Together they form a unique fingerprint.

Cite this