> So you say that: "...I'm observing that most of the increase in level > comes from the selection during exploration and only in small part > from the selection during simulation.", could you elaborate at all? > This is very interesting. That almost suggests it might be fruitful > to use the patterns in the tree, but keep lighter playouts.
If I understand David Fotland [1] (and you) correctly this is the same approach Many Faces uses. Keep the playouts light, but it is worth spending effort to guide the MCTS tree. I also think Remi was saying the same in his paper [2] on using move patterns. E.g. in section 4.1 he says he just used a few features in the simulations, but used all the features to control the "progressive widening of the monte-carlo search tree" (section 4.2). Darren [1]: http://www.mail-archive.com/computer-go@computer-go.org/msg09759.html [2]: http://remi.coulom.free.fr/Amsterdam2007/ -- Darren Cook, Software Researcher/Developer http://dcook.org/mlsn/ (English-Japanese-German-Chinese-Arabic open source dictionary/semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/