Project Overview
The goal of our project is to
implement learning in the game of checkers. Our chosen method of learning
is a genetic algorithm. The evaluation function has twelve properties
that are weighted. The weights for this function are translated into
strings of bits and then passed through a series of crossovers and
mutations. Selection of which genome to use next is achieved through
running a game of the already intelligent checkers, wischk, versus the
newly created weights, dumchk. Theoretically, we will let our program run
for several days and optimal weights should be discovered.
back to main