RPS - Frank Salim

Frank Salim, Fall 2006

Using a Neural Network to Play a Recurrent Matrix Game

Rock Paper Scissors is a game with simple rules. There are patterns one can exploit against any opponent that is not a random number generator. There is precedence in using computer learning, and in particular neural networks to play rock paper scissors. I want to use a neural network to exploit the inevitable recurring short term patterns that are present in the play of both human and AI rock paper scissors competitors.

Paper Presentation

This paper deals with environments containing antagonistic, intelligent learning agents and why it is necessary to use a probabilistic strategy.
Markov games as a framework for multi-agent reinforcement learning

Slides for presentation to be given on 12/12/06

presentation in OO.org format
presentation in PowerPoint format