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

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

> this? Could we make a set of patterns like the one above that I think are
so
> fundamental you can't afford to miss one and see if your system contains
> them? Or would it be too much work...?

No this is trivial to test, just make separate SGF files or put them a few
empty points apart, doesn't matter much.

I will then let the system show the best moves in order of importance and I
want to bet the system will classify the cuts as highest importance as long
as there are no "bigger" moves.

The thing is, the pattern system bases pattern values on the "bigness" of
the move.
Meaning, if you put a simple cutting pattern on the centre of the board, my
pattern system will prefer to move on a 4-4 point and do a whole lot of
other moves first :)

The pattern system is intelligent enough to see how many stones the chains
have and how many liberties, so it can even judge the importance of a cut
(whoops, which makes my previous calculation quite wrong about the
likelihood I have missed a pattern, it becomes MUCH worse, not 1::3700 but
worst case about 1::200, which is still negligeable, as these are
theoretical values only, as I said when a pattern occurs more than once per
500,000 games, I have it.


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