| |
BGonline.org Forums
Luck Adjusted Result
Posted By: Tom Keith In Response To: Luck Adjusted Result (MaX)
Date: Wednesday, 12 August 2009, at 1:08 p.m.
... using variance reduction in the gnubg/Snowie sense ... do add some bias if the eval is not perfect.
That's the surprising thing about luck reduction. It is unbiased even if your NN is not. You get the same result (after enough trials) that you get without VR.
David Montgomery outlines why this is true in his article on variance reduction here: http://www.bkgm.com/articles/GOL/Feb00/var.htm
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.