On Tue, May 14, 2013 at 5:37 PM, Hiroshi Yamashita <[email protected]> wrote: >> I tried ladder in play-outs before, but I could not get good result.
I find that very surprising. In the few experiments I've done some years back, the biggest boost I've seen of any kind came from tactics. Simple UCT with just two rules: 1) when a player puts himself in atari and can escape (meaning make more than two liberties) the next move then capture. 2) when a player plays a move that has two liberties and can be captured in a ladder, play the capturing move. beats UCT without those rules something like 90% of the time. I found the 'don't play self-atari' rule only to give a very marginal gain, but this is a rule that's a lot more fungible. Mark
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
