Hybrid Optimization for Developing Human Like Chess Playing System

Chess is considered as the “Drosophila of Artificial Intelligence (AI)” since the research in AI began with the concept of game playing and the game of chess became favourite topic in the Artificial Intelligence research history. This article introduces various traditional techniques used for design and development of chess playing system and also proposes the improved hybrid optimization technique for deciding the best move in chess. Most state of the art chess engines relied upon more traditional search techniques like Minimax and Alpha beta till the recent success of systems which are based on machine learning concept. Most chess engines are designed in such a way that they do not play intuitively like humans, but calculate as much variations as possible and base their decision upon that. Recent developments in game playing showed how AlphaZero used new approach based on deep reinforcement learning. Chess engines can be improved in terms of intuitive feel for the position by training the model with top human games and best move can be derived using our proposed hybrid optimization technique. Our hybrid based optimization model is evaluated in terms of parameters like accuracy, sensitivity, specificity and found to be more effective when compared with other techniques.