Re: Another Domain

[mueller@xxxxxxxxx (Martin Mueller)]

>> A few weeks ago I posted a message introducing a roughly defined idea,
>> the gist of which was to try to sample the search space of go using
>> "atomic moves" made up of multiple single stone moves and "searching"
>> using these moves.  The main goal being to find the hot spots on the
>> board.  I would have liked to have gotten more feedback from everyone

Maybe the following article is of interest:

@ARTICLE{hu97,
 AUTHOR       = "Hu, S. and Lehner, P.",
 TITLE        = "Multipurpose Strategic Planning in the Game of {Go}",
 JOURNAL      = "IEEE Transactions on Pattern Analysis and
 Machine Intelligence",
 VOLUME       = "19",
 NUMBER       = "9",
 PAGES        = "1048-1051",
 YEAR         = "1997"}


The approach is a bit different from yours. It describes how to combine
different goals (which might not work well in isolation) into a successful
overall strategy.

>I am
>trying to find or devise a simpler problem domain that does not require
>as difficult a base but that exemplifies characteristics similar to
>certain aspects of Go that I could use to experiment and determine
>whether the idea has merit.
>
>What would the requirements be of this simpler domain?
>        1.   Should be a non-trivial problem.
>        2.   An easy evaluation function would be nice.
>        3.   The problem should have natural subdivisions or sub goals
>which are not strictly defined and which are partly independent from
>each other.
>        4.   The moves (change from one node to the next) can be
>combined to satisfy a sub goal.
>        5.   Large enough tree to be sampled like this without
>blanketing the tree.

Maybe proving the safety of territory is a good candidate. See my paper on
that last year for GPW.

http://www.etl.go.jp/etl/suiron/~mueller/publications.html

@INPROCEEDINGS{mueller97b,
 AUTHOR       = "M{\"u}ller, M.",
 TITLE        = "Playing it Safe: Recognizing Secure Territories in
{Computer Go}
by Using Static Rules and Search",
 BOOKTITLE    = "Proceedings of the Game Programming Workshop in Japan '97",
 EDITOR       = "Matsubara, H.",
 ADDRESS      = "Computer Shogi Association, Tokyo, Japan",
 PAGES        = "80--86",
 YEAR         = "1997"}

        Martin