Re: [Computer-go] learning patterns for mc go

[Hendrik Baier]

Thank you Olivier, that is a very interesting paper and exactly what I 
was looking for. I believe there are indeed possibilities for relatively 
domain-independent playout improvements like yours, also in Go.
One question: If you use, in cmc(s), always the argmax over all tiles 
including your *last* move - shouldn't you then also, in the update 
step, only update the tiles of moves directly following each other in 
the playout? As far as I understand your paper, you update all tiles of 
all possible move pairs from the playout.

Best,
Hendrik
_______________________________________________
Computer-go mailing list
Computer-go@dvandva.org
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go