Re: [computer-go] Pattern matching - example play

[Heikki Levanto <heikki@xxxxxx>]

On Thu, Dec 02, 2004 at 03:33:10PM +0100, Vincent Diepeveen wrote:
> 
> A deeper brute force search is an absolute necessity to make it into the
> dan levels with go software. Without that, the weakest chain is and will be
> just some life&death tactics.

Some searching is obviously necessary, but I am not at all convinced
that we need a global full-board search. Unlike chess, go has many
fairly (but usually not completely) localized problems, that IMHO need
to be analyzed locally. 

It may be possible to combine the local results into a global full-board
search tree, but it is not (yet?) obvious how that should be done.

I would not completely write off higher-level planning and neural nets,
and other fancy theories. Many of them have shown their values in
other games. Perhaps some features of go are closer to some other game
than chess...

But of course, unless and until we see really strong programs, we can
only speculate. From my part it is quite idle speculation, as I am not
even writing a go program.

Best regards

  Heikki


-- 
Heikki Levanto   "In Murphy We Turst"     heikki (at) lsd (dot) dk

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/