Learning to Predict Life and Death from Go Game Records

ECD van der Werf*, MHM Winands, HJ van den Herik, JWHM Uiterwijk

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

5 Citations (Web of Science)

Abstract

This article presents a new learning system for predicting life and death in the game of go. It is called gone. The system uses a multi-layer perceptron classifier which is trained on learning examples extracted from game records. Blocks of stones are represented by a large amount of features which enable a rather precise prediction of life and death. On average, gone correctly predicts life and death for 88% of all the blocks that are relevant for scoring. Towards the end of a game the performance increases up to 99%. A straightforward extension for full-board evaluation is discussed. Experiments indicate that the predictor is an important component for building a strong full-board evaluation function.
Original languageEnglish
Pages (from-to)258-272
JournalInformation Sciences
Volume175
Issue number4
DOIs
Publication statusPublished - 15 Nov 2005

Keywords

  • Go
  • learning
  • game records
  • neural net
  • life and death

Cite this