> So: > - theoretically, I don't see any reason for mogo to be asymptotically > consistent
I think it is still asymptotically consistent: - at "each" (or every n) iteration you add one node. - if the node is at the end of the game, the evaluation is perfect. - you play the move with the highest combination of UCT/Rave So starting from the end of the game, I think by induction we can show that it is consistent (without forgetting that we start with a prior, == 0.5 if no pattern, so the RAVE value is never 0). Am I wrong? Sylvain _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
