Behold, a thread where people who wish some machine learning-based webservice, library/framework or application existed meet people who are looking for a fun project idea to implement! Pure magic.
I have an idea of a project involving Machine Learning, applied to the game of chess:<p>You would need a large database of chess games and a program that can detect some features in a position (for example: "white are a pawn up", "black pawn structure is better", "black king is not safe", etc.).<p>The idea would be to see if some features in a position are making one side more likely to win. For example, we might find that being a piece up makes one side more likely to win, since the largest majority of the cases when one side had a piece up, it ended up winning.<p>But that's not an interesting result. What I would like to see if it's possible to discover features that are good for one side, and which we didn't know about.<p>I didn't have time to start anything about it, though.
I posted some a little while ago: <a href="http://blog.andrewcantino.com/blog/2012/04/22/machine-learning-project-ideas/" rel="nofollow">http://blog.andrewcantino.com/blog/2012/04/22/machine-learni...</a>
Here are the final projects for CS 229: Machine Learning at Stanford.<p><a href="http://cs229.stanford.edu/projects2012.html" rel="nofollow">http://cs229.stanford.edu/projects2012.html</a><p>Take any of these studies, and try to replicate or extend it. They're all pretty interesting and come with the advantage that you get to see how someone else approached the problem as well.