Title(KR)
한중일 프로바둑기사 포석 분석
Title(ENG)
Analysis of Korean, Chinese and Japanese Pro Go Players' Openings
Keywords(KR)
Go opening, Related-samples t-test, SPSS, PCA, IndependentLDA, DependentLDA, Classification
Keywords(ENG)
Go opening, Related-samples t-test, SPSS, PCA, IndependentLDA, DependentLDA, Classification
Author
Byung-Doo Lee
Abstract(ENG)
One of the oldest board games in the world is the game of Go which is originated at least more than 2,500 years ago. In spite of its long history, the theoretical studies concerning to Go openings are still insufficient. We firstly used related-samples t-test using SPSS to the three countries' Go mean openings to find out their similarities and differences. Experimental result shows that there are no significant differences (p=.959) between Korean and Chinese mean openings, but meanwhile, there are just some slight similarities (p=.061) between Korean and Japanese mean openings. We secondly applied PCA and LDA classifiers with Euclidean distance to correctly classify a pro player's opening into his/her class obtained from the training openings. Result shows that the recognition rate (42.1%) of dependent LDA classifier is much better than that (27.9%) of PCA classifier; and also that dependent LDA classifier outperforms independent LDA classifierwith the recognition rate of 35.0%.
Page 17-26