Re: ANFIS, Fuzzy Hardware, Online Self-Learning

Frank Hoffmann (fhoffman@cs.berkeley.edu)
Wed, 13 Oct 1999 23:11:39 +0200 (MET DST)

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************
At 01:55 AM 10/5/99 -0700, Frank Hoffmann wrote:
>*********************************************************************
>Berkeley Initiative in Soft Computing (BISC)
>*********************************************************************
>
>From: "Luis Enrique Cordova Sosa" <l.cordova@ieee.org>
>To: <bisc-group@cs.berkeley.edu>, <red-dig@uniandes.edu.co>,
> "Multiple recipients of list" <fuzzy-mail@dbai.tuwien.ac.at>
>Subject: ANFIS, Fuzzy Hardware, Online Self-Learning
>
>ANFIS, Fuzzy Hardware, Online Self-Learning
>
>Dear people involved in ANFIS, Self-Learning and Fuzzy Hardware
>
>I am an electronics engineer student here in National University of =
>Engineering
>from Lima, Peru. Regarding few papers of Professors Lotfi Zadeh, Frank
=
>Hoffman,
>J-S. R. Jang, M. Patyra and others, i would like someone reference me =

>some
>more papers about self-learning ONLINE and preferably without
predefined
>plant model. Better if there are physical implementations (of course on

>ASIC/FPGA).
>
>Thank you very very much to everyone.
>
>"Nowadays i have to deal with this problem, how to map on chip a high
>performance, low cost, universal FC with self-learning capabilities".
>
=
> FuzzyChip
>
>
These issues are discussed in GREAT detail in papers aimed more at
neural
network people.
For example, at www.iamcm.org, you could click on my recent review
paper. (">"
for next slide, and "A" for text.).

That paper cites the Handbook of Intelligent Control, which has some
more
details. David White had many exciting applications
of online learning without any plant model at all... but I think that
better
results are possible by TRAINING an empirical
model of the plant and using that. That is the basis, for example, of
the Ford
clean air controller
(to be implemented on a general purpose neural chip which I think should
cost
about $1), which will be on all Ford cars in the world by 2001
(if they can install it that fast). The chip -- from Mosaix LLC of
Pasadena,
California - can be used in other applications as well, and does
have online learning.

But what about the fuzzy case? This is an interesting story. Most people
doing
adaptive fuzzy systems use forms of adaptation which have very limited
capabilities.
(I would call them "static cloning" systems. See www.iamcm.org to
explain
what I
mean). But in principle, an appropriate kind of
fuzzy system can be trained in any way that a neural network can be,
without
turning into a "black box."
See P.Werbos, Elastic fuzzy logic: a better fit to neurocontrol and true

intelligence J. of Intelligent and Fuzzy Systems , Vol.1, 365-377,
1993.
Reprinted and updated in M.Gupta, ed, Intelligent Control, IEEE Press,
New
York, 1995. Equivalent systems were later described by other authors. A
U.S.
patent has been granted on this, but
is not intended to constrain honest university research. I am not aware
of any
chips implementing this, but there are places
in Japan where they may well exist.

Best of luck,

Paul W.

If you ever want to remove yourself from this mailing list,
you can send mail to <Majordomo@EECS.Berkeley.EDU> with the following
command in the body of your email message:
unsubscribe bisc-group
or from another account,
unsubscribe bisc-group <your_email_adress>
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Do NOT send unsubscribe requests to bisc-group@cs.berkeley.edu
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

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