>
> We tried looking at local patterns and at board locations in 3x3 or
> large-knight's-move neighborhoods. Disappointingly, neither of these things
> helped.
>

I imagine that including patterns would have to use prior knowledge from
game records (or wherever). Maybe they should not look like input terms, but
like bias terms. If you are not using prior knowledge and you are just
telling your classifier that this pattern exists on the board in this game I
don't see how it can effectively learn from it.


> Of course, it's possible that one of these ideas is valid and we just did
> it wrong. We welcome experiments by others!
>

I'm still playing catch-up with everyone else after a lot of time off from
go programming, but once I get there this should be near the top of my list
of ideas to try.
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to