Re: Fuzzy Logic & Neural Nets

Will Dwinnell (predictor@compuserve.com)
Fri, 17 Dec 1999 09:39:35 +0100 (MET)

bv wrote:
"Is there anything in fuzzy logic neural nets can't already handle ?
RBF, Probabilistic, Kohonen nets and the likes already handle fuzzy
classification quite neatly, don;'t they?"

All modeling methods are, at a sufficiently abstract level, equivalent.
I disagree with the general sense of your proposition, however. These
things are techniques, and despite the fact that they may
(theoretically) be used to construct equivalent mechanisms, one or the
other may be preferable in any given situation for other reasons
(convenience, etc.).

This question also gets to the heart of the question "What is fuzzy
logic?" While the simplest fuzzy logical systems employ fuzzy truth
values (as outputs or intermediate results), more sophisticated ones use
fuzzy numbers or sets. Assuming that we do not defuzzify the output,
conventional, scalar neural networks would need to be modified in an
important way to generate distibutions equivalent to fuzzy numbers or
sets. Even taking scalar (defuzzified) output fuzzy systems into
account, fuzzy rule bases can be crafted (by human designers) rather
than learned (as from training data).

Will Dwinnell
predictor@compuserve.com

############################################################################
This message was posted through the fuzzy mailing list.
(1) To subscribe to this mailing list, send a message body of
"SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
(2) To unsubscribe from this mailing list, send a message body of
"UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
to listproc@dbai.tuwien.ac.at
(3) To reach the human who maintains the list, send mail to
fuzzy-owner@dbai.tuwien.ac.at
(4) WWW access and other information on Fuzzy Sets and Logic see
http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
(5) WWW archive: http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html