www.cs.ox.ac.uk/teaching/studentprojects/12.html

You are here:

• Home
• Current Students
• UCT/Monte-Carlo Game-Playing Program

UCT/Monte-Carlo Game-Playing Program

Supervisor
Suitable for	Computer Science, Part B Computer Science, Part C Mathematics and Computer Science, Part C

Abstract

Fairly recently, new ideas have emerged in Artificial Intelligence that have led to the design of radically different game-playing programs. Originally developed for the game of Go, these algorithms (such as UCT or Monte-Carlo) appear to work quite well on various other games (typically games with a very large number of possible moves at each stage) for which minimax/alpha-beta performs poorly. The aim of this advanced project would be to implement and evaluate such algorithms on a particular game.