EFFICIENCY IMPROVEMENT OF LEARNING AND REDUCTION OF LEARNING DATA TO BE STORED IN LEARNING TYPE REVERSI GAME PROGRAM

摘要

PPROBLEM TO BE SOLVED: To improve the efficiency of learning and reduce learning data to be stored in a learning program where a user changes a position to place a stone and weighting states of respective squares in a reversi game program. PSOLUTION: In the game state of the reversi, 7 substantially similar game states can be created from a single game state besides the game state using a symmetric property. When a user changes the position to place the stone and the weighting of the respective squares to allow the learning of the single game state, the learning efficiency is improved by simultaneously learning the seven game states created using the symmetric property as well. As for the data created by learning the game states using the 7 symmetric properties, if the position to place the stone out of the areas divided into eight and the state of the respective squares are stored, the original data can be created by converting the stored data using the seven symmetric properties so that the data to be stored can be reduced. PCOPYRIGHT: (C)2006,JPO&NCIPI