[computer-go] Territory & nfluence

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

From: "Mark Boon" <tesujisoftware@xxxxxxxxxxxxxxxxx>
Subject: RE: [computer-go] Pattern matching - example play



> as such. If someone sees a flaw in my analysis, or can somehow show me
that
> such a system in fact would greatly improve the play of a Go program, then
I
> would be very interested to hear about it. Previously I think this
> discussion has had too much confusion to what it's all about and in some
> cases has been taken too personally. That is a pity, as I think it's
> actually one of the most interesting subjects in this group.


All good and well, I think we are done discussing this pattern topic anyway.
I will post some final results when the nastiest of bugs are ironed out so
one can make a fair judgement.

I have to admit that I don't feel encouraged by unfair critiscism that has
religious undertones.
I don't really am interested in "beliefs". Mark never gives any arguments,
but voices "beliefs" as if his beliefs somehow carry more weight than
another person's measurements.

My new "Bouzy 21/7" delta-updater uses only ONE single morphological
operation of a morphological kernel.
No "generations" that dilate and erode, just a single, hyperfast lense
operation.
(It is not really a genuine Bouzy 21/7 though.)

I would like to get a discussion started about Influence/territory instead.
My idea is that such a thing is not required to be accurate.

For example it is rather irrelevant that it knows how much territory a
player really has, as long as the PROPORTIONS between the players are
correct. So that an engine can do a search to optimize territory or to
optimize a percentual increase or optimize an absolute increase or optimize
a difference with the opponent or keep the territory at least the same etc.

The actual correct territory, is it really of paramount importance?
I ask this because my territory algo is only LIKE Bouzy 21/7, the results
are in fact a bit different.
So I wonder how much time I should spend on making it more accurate. It
seems that it is accurate "proportionally", but not "absolutely". Bouzy's
21/7 neither is completely accurate and as far as I understand an order of
magnitude slower to implement.

For an example of my influence function:

http://www.moyogo.com/influence.htm

This is done with a single pass over the board.
Some "territory" points appear at unlikely places, but I still think those
are "points" (as in score) that are reasonable/relevant.
They indicate "strength". What do people "believe"? :)

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