Re: Assumption free (neuro)-fuzzy modeling????

Sjonnie (Sjonnie@holland.nl)
Mon, 16 Nov 1998 06:04:13 +0100 (MET)

Thanks Robert,

I agree with you, the problem that kind of pops up is:

You said: Actually, I would not want a modeling technique completely free of
assumptions. The trick is to make the assumptions match your knowledge
of how the process works -- that is, assumptions are an opportunity
to build your knowledge into the model.

This means: I generate my assumtions in such a way that they fit my data, if
data is my only knowledge. This means for every assumtion to make my model
fit, I loose a degree of freedom. => If I use enough assumptions (MFs and
rules) I can make any model fit my data.
The Fang and Chen guys used clustering to find MFs. They ended up with in
total 42 MFs (7 rules) and they had only 43 data records. If you say fuzzy
is assumption free, your home free, model is perfect. If this is not the
case (they used on average 4 points to define their MFs 4*42=168 parameters
in the model) their model must be crap.

What is your opinion?

PS I hope this is all clear.

############################################################################
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