2007/3/9, Eduardo Sabbatella <[EMAIL PROTECTED]>:
> Is it useful for MC Go?
> The idea is appealing for UCT, as the distribution
> of the arms is not
> stationary, and discounting is the simplest idea to
> deal with
> non-stationarity.
I think in Go outcome is stationary.
The "real" outcome for a move in a given board setup
will never change. Its a fixed.
I told in UCT. So in UCT, the distribution changes as you add nodes in
the subtree and the distribution converges to the max (was the average
when there is no subtree).
Sylvain
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/
