On Mon, Apr 27, 2009 at 01:25:29PM -1000, Mark Boon wrote:
> On Sun, Apr 26, 2009 at 11:54 PM, Łukasz Lew <[email protected]> wrote:
> > Hi,
> >
> > Have any one of you tried uct + capture heuristic?
> > Is it stronger? How much?
> >
>
> >From memory I'd say it was winning about 80% against plain UCT. But
> only capture if it can escape (which means it can make three
> liberties.)
>
> Mark
Just to clarify, is this a capture heuristic in the playout or in the
tree ? If it's in the tree, how did you include this heuristic in the
uct computation ?
And, can make three liberties is just an approximation of "can escape".
In this example, the two whites stone can escape even if they can only
make two liberties and is very comon in ladders sequances. Did you
handle thoses ?
......
.ox...
.xoox.
..xx..
......
This example show the opposite behaviour, even if the two white stones
can make three liberties, they can't escape.
.......
.xx..x.
.xoo...
..xxx..
.......
It ssems to me that this heuristic will put a very strong bias toward
making ladders shapes even if there is a blocker or atari cause the
inner player is considered the looser in case it can't localy escape.
--
Thomas Lavergne "Entia non sunt multiplicanda praeter
necessitatem." (Guillaume d'Ockham)
[email protected] http://oniros.org
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/
