Re: computer-go: Results on self play vs recorded games

[Peter Drake <drake@xxxxxxxxxx>]

On Thursday, August 21, 2003, at 10:17  AM, Erik van der Werf wrote:

> Interesting results, I'm curious, how good are the recorded games?
>  Thinks like average rank of say the weakest 10% of the players would 
> be interesting. (I assume these were human players, right?).
I got them from Nici Schraudolph, who would have to make any such 
comments.

> Did you do any fancy things like using a modified TD to deal with bad 
> moves?
No.  The self-play used a sort of best-first search to choose the best 
move.  I waited until the end of the game and then taught the network 
that (a) all moves made were the correct moves and (b) the final board 
should be predicted from every previous board.

Peter Drake
Assistant Professor of Computer Science
Lewis & Clark College
http://www.lclark.edu/~drake/