Show More


Computer Science Department

Evolutionary Gambler: A human vs. computer game with the elements of artificial intelligence
and statistical analysis

Ketevan Parlagashvili

This human vs. computer game is an extension of the classic
rock-paper-scissors game with more than three objects and additional
constraints. The objects corresponding to rock, paper and scissors
are animals, some of which eat others. Both players (human and
computer) can place a bet on exactly one animal at a time. They can
bet with one or more coins, which they will be provided with at the
beginning of the game. Once one of the players runs out of coins,
the game ends. The goal for the program is to memorize and analyze
the behavior of the human player. The program will respond to the
actions of the human player in a way that guarantees to play, on
average, at least at the same level as the human. The longer the
game lasts, the more likely will the program be to increase its
rate of winning. This assumption is based on the fact that humans
are generally unable to produce completely random solutions and
their actions mostly involve some kind of a pattern. It is the
program's goal to analyze these patterns and use them to its own
advantage.