12:00, Wed 11 Dec 1996, WeH 7220 Adversarial Reinforcement Learning Will Uther I'm planning to start out with a quick intro to my long term research interest: Bolo. Then we get on to the meat of the talk, my results using various reinforcement learning techniques in a small 'Hexagonal Soccer' game. This work started as a quick look at Michael Littman's Minimax-Q-Learning algorithm and grew from there. A number of RL algorithms are compared, including: Q-Learning, Minimax Q-Learning, Opponent Modeling Q-Learning, Prioritised Sweeping. I also look at some generalising algorithms: Fitted Prioritised Sweeping and UTree.