Did you try playing out semeais at the border of mc tree and playout?
Or would that be too expensive?
I think at that point even a fairly naive trading of moves could improve
the accuracy of the playout.
Stefan
Am 13.01.2011 22:44, schrieb Olivier Teytaud:
By the way, has anybody tried the most basic way to settle semeais in
playouts? I mean:
* Spot semeais (not basic at all, but probably necessary for
almost all
approaches).
* Play all the moves in the semeai before going to usual playouts.
We've tried several things around that:
1) play the semeai first
- spot semeais (not that hard after all, from statistics on simulations)
- solve them with a dedicated solver
- play the moves which simplify the semeai before the rest of the
Monte-Carlo part
2) discard simulations
- spot semeais
- solve them with a dedicated solver
- discard simulations which propose a result which is not consistent
with the dedicated solver
3) add expertise (positive for semeais that should be played
immediately, negative for others)
Some of the experiments are here:
http://hal.inria.fr/inria-00512854/fr/
and there's nothing which works, even on very simple semeais :-) as
many people in this mailing list have tried similar things and failed
as well I start thinking that I'll die of
old age before someone solves this :-) (and I congratulate in
advance the person who will
solve that :-) )
Best regards,
Olivier
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
