On Wed, Apr 6, 2011 at 10:11 AM, fastgo <[email protected]> wrote: > In StoneGrid, it has built in the Benson algorithm in the playout. It does > not seem to help much on overall strength. But I think it should solve the > case here. > > I think you can implement the Benson Algorithm in the late stage of the > playout. > > It has another advantage finishing the playout in shorter length. In > StoneGrid, the playout on average finishes in a little bit more than 70 > stones on 9x9. > > You need quite a lot of effort to get it fast as well as correct. So the > return of the effort is low. >
This is the exact type of thing I was rambling on about in a recent message. What you need to do is carefully measure this to see if you got a real improvement, and then be happy if you did. If the improvement is more than 5 or 10 ELO you should be ecstatic. You should also check to make sure it did not regress the strength of the program. The bad news is that you will probably need 10,000 games to check this with high certainty, unless the improvement is a lot (more than 15 or 20 ELO.) Use bayeselo to check your results. Do that and then move on to the next thing. Do that 10 times and you have 1 dan. Don > > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
