[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] influence and more
From: |
Gunnar Farneback |
Subject: |
Re: [gnugo-devel] influence and more |
Date: |
Sun, 24 Nov 2002 20:11:46 +0100 |
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Inge wrote:
> I agree about this. We might, however, want to use non-linear
> relation between .weakness and need to enforce the dragon, The
> sigmoid function is standard in such cases.
This might be slightly off-topic, but I'm curious why you would
consider sigmoid functions as more natural than other non-linearities
in this application. I was of the impression that the most interesting
aspects of sigmoids are to map an unbounded interval to a bounded one
and having nice derivatives which come in handy for neural net
applications. Neither of these seem very relevant for the weakness
to value mapping.
/Gunnar