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

["Frank de Groot" <frank@xxxxxxxxxxxxx>]

From: "Heikki Levanto" <heikki@xxxxxxxxxxxxxxxxx>
Subject: Re: [computer-go] Pattern matching - example play


> Those concepts you mention are all explained in the first beginners
> books. The hard concepts are explained in more advanced books...

My ideaa is that you only need the beginner's books stuff.
If you apply that rigorously and to the max, and in a deep search, you end
up with a system that applies the very complex pro stuff all by itself.
Because the pro stuff is all simple concepts applied all toghether deep into
the future of the position.

> I still don't see how you can extract exactly the relevant information
> out of games.

Look at it this way.
I don't play Go.
I got a simple idea to extract "good shape".
That code is very small.
It predicts 46% of the moves a pro plays and 66% is in the top-3.

Imagine how it will improve when I add learned connectivity, learned
eyespace and learned territory.

The method of extracting the info is a trick of the trade but just because
it's trick doesn't mean it is the wrong method to develop a Go program, or
that having no knowledge of Go oneself makes one doomed to failure..

Enough talk, we will see.

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