A Differential Evolution for the Tuning of a Chess Evaluation Function

We describe an approach for tuning a chess program evaluation function. The general idea is based on the differential evolution algorithm. The tuning of the evaluation function has been implemented using only final outcomes of games. Each individual of the population represents a chess program with specific (different) parameters of its evaluation function. The principal objective is to ascertain fitness values of individuals in order to promote them into successive generations. This is achieved by competition between individuals of two populations which also changes the individuals of both populations. The preliminary results show that less generations are necessary to obtain good (tuned) parameters. Acquired results have exhibited the fact that population individuals (vectors) are not as diverse as they would be if no changes were made during the competition.