On Thu, Oct 9, 2008 at 11:39 AM, Jason House <[EMAIL PROTECTED]> wrote: > On Oct 9, 2008, at 11:08 AM, "Eric Boesch" <[EMAIL PROTECTED]> wrote: > [...] > Does anyone allow passing at random in their playouts??? A game stopped from > two premature passes is tough to score, if not completely meaningless.
I haben't done this, but I have a plan that includes passes "at random"(with non-uniform probabilities). With the rules I use, a game stopped by two premature passes is very easy to score, and it does have some meaning. I'll explain. For any move played during a playout we use some information (the 3x3 pattern around it, the location on the board, the number of liberties, the number of stones captured...) to estimate its score. The sum of the estimates of all the moves in a playout should come close to the actual score at the end. We use the difference between those two numbers to adjust some weights that were used in computing the estimates. Over time this system will learn which moves are good and which aren't. Now we can bias the playouts so moves with high estimated scores are played more often (in some cases, much more often). A system like that would learn that playing almost any move is generally better than passing. The hope is that this would recognize sekis because moves there would have scores that are much worse than passing, so both players would pass without moving there. It is even possible that we can remove the definition of what points look "eyeish", and have this algorithm figure out that moves with those 3x3 patterns are usually worse than passing. I don't know how workable that is, but it's about the only plan I have to get MC to handle sekis correctly. Álvaro. _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
