Re: Is this an application for Fuzzy Logic

David A. Fuess (fuess@llnl.gov)
Mon, 29 Jul 1996 18:10:15 +0200


arms@cs.ualberta.ca (Bill Armstrong) wrote:

>The answer is yes, but the ttweaking or training aspect can
>be so important that you might do better with a neural net.
>You could use a net which is compatible with fuzzy logic though.

I agree with the YES part of Peter's reply, but part company when he
recommends NN as the first approach. I ten dto use NN when the
boundries are unknown not just ill-defined.

In this case the boundries are known but imprecise. It is a classical
"textbook" case for FL.

Dave