Show More


Computer Science Department

Machine Learning with Various Strategy Games

Peter Correia

This project intends to use a branch of artificial intelligence,
machine learning, to learn how to play and to eventually excel at
two strategy games, The 100 Game and Grundy's Game. The principle
of the 100 game is relatively simple, there are two players that
start from 0 and alternatively add a number from 1 to 10 to the
sum. The player who reaches 100 wins. Once one of the players reaches
89 the opponent has lost because the only other option is to will
inevitably end in a loss. In this version there will be one human
opponent and the other opponent will be the computer. The other
game is Grundy's Game and it is also a simple game. There are two
players where instead of adding numbers to a heap two players are
splitting the heap. On each turn, a player may split the heap up
so there is at least one element in both of the newly created heaps.
One player wins when they split the last heap up and there are no
more left. For these games there will again only be one human
opponent and the other opponent will be the computer. This second
computer opponent will progressively become more and more competent
in both strategy games. This project will work in the Java Programming
language.