On Tue, 5 Oct 2010, Kahn Jonas wrote:
It is still few days away from being ready for submission, but I figured
that given the recent surge of interest in dynamic komi at the list, it
might be useful to make an earlier preprint version available:
http://pasky.or.cz/~pasky/go/dynkomi.pdf
Any comments and suggestions are welcome!
Nice article!
I have a question on the methods that were tried, however, specifically
Value-based dynamic komi. I would have expected different zones for the case
when komi is
increased and for when it is decreased. Namely, if you are winning, you
would still want to play somewhat conservatively (and that's why the
45-50 interval is surprising), and when you are losing, you would still
want to be aggressive.
Have you tried anything looking like having values red_low \leq red_high \leq
green_low \leq green_high
and maintain winning rate between:
* red_low and green_low if komi is decreased
* red_low and green_high if normal komi
* red_high and green_high if komi is increased ?
Or maybe the easier red \leq orange \leq green, and values between
* red and orange if komi is decreased
* red and green if normal komi
* orange and green if komi is increased ?
Jonas
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
