On 12/27/14, Aja Huang <[email protected]> wrote:
> We were doing exploratory work that optimized performance not training
> time, so we don't know how symmetry affects training time. In terms of
> performance it seems not have an effect.

You are using 3x3.  Clarke and Storkey are using 5x5 (section 4.2,
first sentence).  So, each of your 3x3 filters contains about the same
amount of information (9 weights) as Clarke and Storkey's symmetrical
5x5 filters (triangle 3+2+1 = 6 weights).  If you made your 3x3
filters symmetrical, they each only have 3 weights, which is a bit
small perhaps?

I think an interesting question could be: better to have symmetrical
5x5 (6 weights) or no-symmetries 3x3 (9 weights)?

Hugh
_______________________________________________
Computer-go mailing list
[email protected]
http://computer-go.org/mailman/listinfo/computer-go

Reply via email to