Sutton & Barto Book Reinforcement Learning An Introduction
Andres Perez - Reinforcement Learning & Autonomous Robots page
Reinforcement Learning Repository at University of Massachusetts, Amherst
Temporal Difference Learning and TD-Gammon - This site gives a good introduction to Richard Sutton's Temporal Differencing algorithms, along with an eplanation of how the TD-Gammon project used it very successfully to train a Backgammon program.
KnightCap is a chess program which uses Richard Sutton's TD(lambda) algorithm to modify its evaluation function based on the outcome of its own games.
NeuroChess is another chess program which uses Richard Sutton's TD(0) algorithm to modify its evaluation function, which is a Artificial Neural Network, based on both the outcome of its own games and the outcome of thousands of expert level games used for training.
Last updated: 27/01/04