CHESS AI: Machine learning and Minimax based Chess Engine

Abstract

Designing Chess Engine has been a main focus of research for a long time. The paper employs a novel combination approach of Machine learning based estimator with artificial intelligence (AI) to build chess AI. The Minimax Algorithm is a decision theory-based technique implemented for reducing the load on the chess engine’s hardware. Also, Alpha-Beta Pruning algorithm is implemented to eliminate any nodes in the search tree that aren’t essential and hence makes the AI efficient. Trained estimators achieved a high accuracy of 96.77% for calculating the probability of ‘good move’. A variable depth of search tree based on the number of legal moves has also been employed for the minimax algorithm.