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

["Mark Boon" <tesujisoftware@xxxxxxxxx>]

> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx]On Behalf Of Frank de Groot
> Sent: Monday, November 29, 2004 23:10
> To: computer-go
> Subject: Re: [computer-go] Pattern matching - example play
>
> Not when you use the right type of search.
> Humans can do it, so why not a computer?
>
> What might help is huge pattern libraries of all kinds to direct
> the search
> and evaluate positions in terms of connectivity, shape, eye space etc.

Yes, that can help a big deal. Humans use patterns for this as well.
However, the patterns are backed up by some kind of procedural knowledge as
well, so when in doubt the result of the pattern can be overruled. It's this
kind of 'fuzzy' evaluation that makes humans good at Go.

With connectivity and eyes and such, you don't need professional games to
'learn' the patterns. That's because it's easy to verify the result of a
pattern by doing evaluation or search. I don't think you can use patterns
solely to determine eye-space, as any approximation isn't going to be good
enough. But you can probably use patterns as a short-cut.

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