Can a Bayesian Spam Filter Play Chess?

martin-boundary writes "The typical Bayesian spam filters learn to distinguish ham from spam just by reading thousands of emails, but is this all they can do? This essay shows step by step how to teach a Bayesian filter to play chess against a human, on Linux, with XBoard."

Comments

Re:Results?

[aug24]

..and from reading to the end of it, the answer is...

Not really. But it does work, and it would be possible from someone to take this and expand on it quite neatly.

For example, it currently uses entire games to compare. So if it comes across an unusual opening, even one close to a standard one, it's not able to decide effectively. Perhaps something using game fragments would be possible, then it might reproduce structured plays even when the previous game play has been unusual.

Really though, it is a successful tiny step in a direction that no-one else has thought of going. That's worth congratulating in and of itself.

So... anyone got any other suggestions for improvements?

Justin.