|
 Learning
to Play Go
The complexity of game of Go is greater than that of Chess; the
most advanced Go computer players reach at best the level of a human
amateur. We believe that machine learning can radically improve
computer Go. |
|
Current Projects
- Computer Go
The game of Go is an ancient Chinese game of strategy for two
players. By most measures of complexity it is more complex than
Chess. While Deep
Blue (and more recently
Deep Fritz) play Chess at the world champion's level no
Go-playing program has yet even reached the level of play of an
average amateur Go player. The reason for the failure to reproduce
the impressive results in chess for the game of Go appear to lie in
its greater complexity, both in terms of the number of different
positions and in the difficulty of defining an appropriate
evaluation function for Go positions.
We take the view, that only an automated way of acquiring Go
knowledge - machine learning - can radically improve on the current
situation in computer Go. Numerous Go servers in the internet offer
thousands of game records of Go played by players that are very
competent as compared to today's computer Go programs. The great
challenge is to build machine learning algorithms that extract
knowledge from these data-bases such that it can be used for playing
Go well.
|
|
Publications
- David Stern, Thore Graepel, David MacKay. Modelling Uncertainty in the Game of Go. Advances in Neural Information Processing Systems 16, 2004
(PDF)
- David Stern, Ralf Herbrich, Thore Graepel. Bayesian Pattern Ranking for Move Prediction in the Game of Go. International Conference on Machine Learning 2006
- David Stern, Ralf Herbrich, Thore Graepel. Learning to Solve Game Trees. Submitted to International Conference on Machine Learning 2007
|
|
|
|
