On Oct 1, 2010, at 8:03 PM, Petr Baudis wrote:

>  Hi!
> 
> On Fri, Oct 01, 2010 at 12:49:59PM +0200, Rémi Coulom wrote:
>>> * More importantly - how do you handle tenuki? Does the playout
>>> *always* play a local move if any is available, or do you include tenuki
>>> with some value within the probability distribution?
>> 
>> The same 3x3 patterns are used both for tenuki and local moves. Local moves 
>> are made more probable than tenuki thanks to the "contiguous" feature.
> 
>  Oh, that's very interesting; I thought your probability distribution
> is based purely on local moves. I guess I misunderstood some earlier
> hints.
> 
>  Did you choose to use only the 3x3 pattern feature for whole-board
> probability distribution for performance reasons or ease of
> implementation, or because considering the other features as well
> reduced strength?

I never really tried anything else. In order to get a reasonable speed, tenuki 
distribution should be easily updated incrementally. Having to recompute 
features for every point of the board at every move would probably be too 
costly.

> 
>  Anyway, congratulations to your great performance!

Thanks. I am very proud of my student.

Rémi
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to