Another spin on this idea: if you have winrates at several different
komi level for all move candidates, you could piece this together
weighting correct komi the strongest and large fantasy komi the least.
In a position that benefits from dynamic komi, the high komi results
will still prevail, because the low komi results will be almost identical.
Otoh, maybe this could also help in those high handicap positions where
the bot otherwise would give away the game by giving the opponent too
much certain territory in the beginning. Moves that do well at a high
komi level, but do even worse than other moves without komi, could be
sorted out.
Stefan
Von: Stefan Kaitschick<[email protected]>
...
A good alternative might be to let each computer evaluate the position
at a different komi level.
Then you get a risk/reward profile for the current position.
As this is more about the risk/reward characteristics, and not so much
about the best move, it might also be a good idea to do this kind of
search instead of "pondering".
I like this idea.
Does someone have ideas how to use such komi profiles in concrete?
Ingo.
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
