[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Gnu Go 3.2 possibly correctable flaw in play
|
From: |
Gunnar Farneback |
|
Subject: |
Re: [gnugo-devel] Gnu Go 3.2 possibly correctable flaw in play |
|
Date: |
Mon, 28 Jul 2003 20:21:47 +0200 |
|
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Dan wrote:
> Right, move 12 is crucial in this game. Unfortunately 3.3.23 also plays
> W:12 at P16.
>
> [...]
>
> K3 is not considered, nor are other moves in this area!
>
> The patch below fixes that. It has to be written as several patterns
> due to the fact that every pattern has to have an anchor.
>
> After this patch W plays K3 in this game. The problem remains,
> that P16 is overvalued, so W will go back and play it fairly
> soon in this game.
It's clear that we have some tuning problems with the fuseki, and
we've had that for quite some time. I'd like to encourage people who
are interested in doing machine learning experiments with GNU Go to
work with the fuseki. To me it seems like one of the areas with most
potential for substantial and reasonably quick improvements.
/Gunnar