A Method Of Generating Sample Of Self-playing Chess Game Based On Competition System Organization

The study of computer game has been running through the development of artificial intelligence.Although the new generation of artificial intelligence players represented by Alpha Zero have been able to defeat human players,there are still many problems that can not be thoroughly analyzed in the study of board games,among which the problem of learning sample generation is one.It is well known that the quality of learning samples affects the quality of AI players.However,this problem has not attracted enough attention,and there is a lack of research on the generation and evaluation of samples.Therefore,this paper puts forward a method of sample generation of self-playing chess game based on competition system organization.Based on the analysis of various sports competition system,this paper puts forward an algorithm of generating sample of genetic variation self-playing chess game based on competition system organization and a comprehensive evaluation method for quality of standard weight samples.The main work of this paper is as follows:(1)In view of the generation process of self-playing chess game learning samples,a genetic variation self-playing chess sample generation algorithm based on match system organization,namely C-GASP sample generation algorithm,is proposed.The algorithm can effectively regulate the selection process of players by introducing mature and fair sports competition system,and make players evolve generation by generation by using genetic variation method,so as to further improve the generation efficiency of samples.(2)In order to quantitatively evaluate the quality of learning samples,a comprehensive evaluation analysis method of standard weight samples,namely SWE method,is proposed.The evaluation indexes directly related to the sample quality are selected to evaluate the quality of the sample from a quantitative way,and the rationality and correctness of the method are verified based on the analytic hierarchy process(AHP)and the comprehensive evaluation index of sample size.(3)Based on the analysis of sample generation method and extraction process,the experimental platform of sample generation of self-playing chess game is constructed.This platform not only supports self-playing checkers,but also meets the needs of this study to acquire learning samples,make qualitative and quantitative analysis,train AI players,provide AI interfaces and implement them into the game for actual combat detection.Several experiments on the platform of sample generation of self-playing chess games show that the method proposed in this paper is feasible and effective,and can generate high quality learning samples in a short time.