David Silver wrote:
I would be very interested to hear more about Dimwit's approach, and
also Remi's experiments with online learning in CrazyStone.
Hi,
My idea was very similar to what you describe. The program built a
collection of rules of the kind "if condition then move". Condition
could be anything from a "tree-search rule" of the kind "in this
particular position play x", or general rule such as "in atari, extend".
It could be also anything in-between, such as a miai specific to the
current position. The strengths of moves were updated with an
incremental Elo-rating algorithm, from the outcomes of random simulations.
I did not go very far in that direction, and my rule-based program is
still very weak. I found that I could bring very big improvements to
Crazy Stone with the techniques I described in my paper, so I focused on
that. I will incorporate my patterns into the rule-based program in the
future.
I found that my rule-based program scaled extremely well with larger
board sizes. What about yours ?
Rémi
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/
