The March 2007 KGS computer Go tournament will be this Sunday, in the
Asian night, European evening and American morning, starting at 17:00
UTC and ending at about 21:00 UTC.
It will use full-size boards, Chinese rules with 7.5 points komi, and
fast time limits of eighteen minutes absolute.
Hi,
I thought I'd report a small change I have made to the plain MC
algorithm. I have been unhappy with the fact that the result of the game
gets pressed into one bit (win/loose), and all other information is
discarded. This leads to silly endgame moves, since the algorithm sees
no difference
Heikki,
This is very similar to what AnchorMan on CGOS does. At the end of
each random simulation I keep the same statistcs on each point of
the board and I use it to improve the move selection algorithm. I
call this special board an ownership map for obvious reasons. I
just divide each value
Chaos theory has been said to suffer from eerily reminiscent syndrome:
you do some tests, generate some graphical results, the significance of it is
uncertain, but the images are so eerily reminiscent of something or other
So, in that fine tradition:, I'm temporarily posting some
That's not the actual one I had in mind. The program I'm talking about
is a couple of years older than this experiment.It basically used
PBIL as a mechanism to find a good move list ordering for playing the
game. This is not particularly flexible and has some serious
limitations,
but it's
On Tue, 2007-02-27 at 14:46 -0800, terry mcintyre wrote:
From: Don Dailey [EMAIL PROTECTED]
I have an older program (which I cannot find) that was quite
interesting, it evolved a playing strategy using PBIL,
What is PBIL?
(P)opulation (B)ased (I)ncremtal (L)earning. It related to
I wonder whether you could save time by not doing this during the
opening? It seems like ownership maps will be meaningless for opening
moves and gradually become more important the closer you get to the
end. It would be interesting to see how many moves into the game you
have to be before it