CPYu: An Optimization of Chess Playing Through Game Tree Search Reduction and Supervised Learning

Abstract

It is infeasible to calculate all the different move sequences in a chess game, since the game tree is far too large for even modern day computers. CPYu, a chess engine implemented in Python and built from scratch, was created to ana- lyze the efficacy of game tree pruning algorithms and supervised learning meth- ods. The pruning algorithms, which induce game tree truncation and search space reduction, result in significant decreases in computation time. Supervised learning on grandmaster chess games was used to train CPYus evaluation of a position and increase its playing strength.