Re: Need proofs of approximation power of the 1st-order Sugeno systems

From: Bruno Di Stefano (nuptek@sympatico.ca)
Date: Sun Sep 02 2001 - 13:43:08 MET DST

  • Next message: Ajith Abraham: "HIS 2001 - Final Call for Papers"

    This is a multi-part message in MIME format.
    --------------98BE539152D15D40313440DC
    Content-Type: text/plain; charset=us-ascii
    Content-Transfer-Encoding: 7bit

    I am a bit busy and I cannot get too much into details. However,
    I can contribute a bit of references to the Fuzzy Approximation
    Theorem. You may already know some or all. They are just background
    material, but they contain many references.

    You can look at: http://sipi.usc.edu/~kosko/
    There you will be able to download:
    - B. Kosko, S. Mitaim, "Adaptive Joint Fuzzy Sets for Function
    Approximation",
      Proceedings of the 1997 International Conference on Neural Networks
      (ICNN-97), pp. 537-542, June 1997
    - B. Kosko, S. Mitaim, "What is the Best Shape for a Fuzzy Set in
    Function
      Approximation?", Proceedings of the 5th IEEE International Conference
      on Fuzzy Systems (FUZZ-96), pp. 1237-1243, September 1996
    Both of them contain information related to FAT.
    Other papers listed in http://sipi.usc.edu/~kosko/ should
    be of interest to you.

    I also suggest that you look at:
    - B. Kosko, "Neural Networks and Fuzzy Systems: A Dynamical Systems
      Approach to Machine Intelligence", Englewood Cliffs, NJ: Prentice
      Hall, 1992, ISBN 0 13 611 435 0
    - Bart Kosko, "Fuzzy Engineering, 1/e", 1996 Prentice Hall,
      (ISBN 0-13-124991-6)
    - Bart Kosko, "Fuzzy Systems as Universal Approximators",
      Proceedings of the First IEEE Conference on Fuzzy Systems
      ZZ-92), pages 1153-62, San Diego, March 1992
    - C.H. Chen, "Fuzzy Logic and Neural Network Handbook"
      McGraw-Hill Series on Computer Engineering. New York, NY:
      McGraw-Hill, 1996. ISBN 0 07 011189 8
      (Chapter 9 of this book is authored by B. Kosko and
      has as title "Additive fuzzy systems: from function
      approximation to learning").

    As a last resource, my (growing) web pages of links:
    http://www3.sympatico.ca/nuptek/NuptekFuzzyLogicLinks.html
    (A frame of: http://www3.sympatico.ca/nuptek/)
    http://www.ecf.utoronto.ca/~bruno/fzlnk.html
    (A frame of: http://www.ecf.utoronto.ca/~bruno/)

    I remember just now that "Fuzzy Logic Toolbox
    For Use with MATLAB","User's Guide", Version 2 or more recent,
    describes how to create a fuzzy inference system from data,
    using ANFIS (adaptive neuro-fuzzy inference
    system). The system is a Sugeno system. The method is
    described from page 2-86 to page 2-119. While this will
    not help you with the proof, if you have MATLAB, it will
    help you to "play" with the concept and to come up with
    some idea for the proof. There is also a list of
    references at the end of the chapter.

    If you tell us (all people in this newsgroup) something more
    in detail about what you are doing, somebody may come up with
    some better idea.

    Good luck.

    Best regards

    Bruno Di Stefano

    Vitaliy Kolodyazhniy wrote:
    >
    > Dear all,
    >
    > I am a 2-nd year PhD student. Currently, I am summarizing all my research
    > and working on my thesis. The title of the thesis is "Control of stochastic
    > systems using fuzzy-neural models". I have developed a number of adaptive
    > control schemes based on the 1st-order Sugeno models, and now I need
    > mathematical proofs of approximation power of such models. As it seems to
    > me, most existing proofs deal with the 0-th order Sugeno-type or alike fuzzy
    > systems, among them are the results obtained by L.-X. Wang, B. Kosko, V.
    > Kreinovich, and many others. Intuitively, it is quite clear that the
    > 1st-order Sugeno systems possess much better approximation capability than
    > the 0th-order ones, but so far I haven't found any mathematical work on this
    > issue.
    >
    > Another important thing (in close connection with the first one) is
    > estimation of the approximation accuracy with a given number of clusters
    > (rules) with given parameters.
    >
    > Could you give me any links to downloadable papers? Though I would
    > appreciate any help.
    >
    > Thanks in advance.
    >
    > Vitaliy Kolodyazhniy
    > kolodyazhniy@ieee.org
    >
    > ############################################################################
    > 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

    -- Bruno Di Stefano ----------------------------------------
    Business: nuptek@sympatico.ca www3.sympatico.ca/nuptek
    UofT: bruno@ecf.toronto.edu www.ecf.toronto.edu/~bruno
    IEEE: b.distefano@ieee.org
    Private: brunodistefano@hotmail.com
    ____________________________________________________________
    --------------98BE539152D15D40313440DC
    Content-Type: text/x-vcard; charset=us-ascii;
     name="nuptek.vcf"
    Content-Transfer-Encoding: 7bit
    Content-Description: Card for Bruno Di Stefano
    Content-Disposition: attachment;
     filename="nuptek.vcf"

    begin:vcard
    n:Di Stefano;Bruno
    tel;fax:416 925 7288
    tel;work:416 925 7231
    x-mozilla-html:FALSE
    url:http://www3.sympatico.ca/nuptek
    org:Nuptek Systems Ltd
    version:2.1
    email;internet:nuptek@sympatico.ca
    title:Bruno Di Stefano
    adr;quoted-printable:;;15 McMurrich Street=0D=0Asuite 1002;Toronto;Ontario;M5R 3M6;Canada
    note;quoted-printable:Nuptek Systems Ltd =3D=3D> Electronics, Computer, and Software Engineering /// Real-Time Embedded Systems /// Fuzzy Logic /// Algorithms Development & Simulation =0D=0A
    x-mozilla-cpt:;22368
    fn:Bruno Di Stefano c/o Nuptek Systems Ltd.
    end:vcard

    --------------98BE539152D15D40313440DC--

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



    This archive was generated by hypermail 2b30 : Sun Sep 02 2001 - 13:59:20 MET DST