Dirk Brand (Bsc Hons Computer Science).

Investigation and Design of AI players for the Risk board game.

Risk is a modern board game based on war strategies and world-domination produced by Parker Brothers. It was originally released in France in 1957. The game can be played by two to six players and has been made into countless variations on the original. The most familiar version, is the original world-domination orientated game where players aim to eliminate their opponent’s forces and seize their territories to eventually conquer the world.

The game has previously been approached from an artificial intelligence (AI) perspective by Michael Wolf and Franz Hahn. As with all classic and modern board games, the question
is always “How strong an AI player can be made for a game as complex as this?”.

The strength of an AI player is usually measured by its ability to win ‘most of the time’ when playing against a certain level of human player. This strength is intimately linked with how many moves an AI player can think ahead. I will investigate two approaches to artificial intelligence design, Expectiminimax with Alpha-Beta pruning and Monte Carlo Tree Search. I will also design a Risk game to allow players to play against the AI that I develop.

My contact details: dirkbrand@ml.sun.ac.za