On Tue, 5 Oct 2010, Erik van der Werf wrote:
Lukasz brings up an interesting point. Winrate may not be the ideal
statistic for all situations. Maybe the average score (as used in most
early work on MC), soft-max, or a median tracker would be better for
some situations.
Maybe a nice question for the academics:
If you were free to keep track of a histogram for all possible scores
in each node (so you have everything from winrate at every possible
komi to simply the average score), then what would be the optimal
selection strategy?
That's a very hard question for many reasons:
* what do we really simulate?
* what's the law of the playouts in this context?
* there is a kind of «dependance» between playouts. I mean, even if the
playouts moves are independent in each playout, if two different playouts play
a local situation in the same way, they get correlated. This is not
mathematical dependence, but it has some effect on the way I see things,
even in an ideal setting.
I'll try to write something in a very idealised situation. Using also
approximations like not using deconvolution, and so on. Even then, I
don't know if we can get anything.
And if the above can be answered, what would be the minimum set of
statistics needed to maintain similar performance?
Probably many...
Jonas
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
