BISC: BISC NEWS Letter for the Week of July 16

From: masoud nikravesh (
Date: Mon Jul 16 2001 - 12:50:30 MET DST

  • Next message: J Lawry: "PhD studentships at Bristol"

    Berkeley Initiative in Soft Computing (BISC)

    Dear Colleagues,

    To ensure your messages to our mailing list will be distributed
    efficiently please forward your messages or announcements with regards
    to CFP (call for paper), conference, workshop, book announcement, etc.

    1. Title
    2. A short message in order of 2-3 sentence
    3. Contact person's name and Email (address, tel, etc.)
    4. URL for more detailed information.
    5. Please avoid long messages if would be possible

    Please note, we do post the messages weekly, except special BISC
    announcements and viewpoints.

    Best Regards,
    Masoud Nikravesh
    UC Berkeley
    BISC program


    Call for papers for the special issue of the
    International Journal of Applied Mathematics and Computer Science
    on Computing with Words (CW).
            Computing with words is an emerging methodology in which the
    objets of computation are words or propositions in a natural language.
    Basically, computing with words serves to harness the high expressive
    power of natural languages to provide solutions to problems which lie
    beyond the reach of conventional computational techniques. In so doing,
    computing with words opens the door to a wide-ranging enlargement of
    the role of natural languages in the basic and applied sciences, and
    especially in information processing, decision and control.

         Computing with words draws upon many fields which in one way or
    another relate to reasoning and computation in the context of natural
    languages. Among such fields are fuzzy logic, granular computting,
    semantics of natural languages and computational linguistics.

       Papers which relate to computing with words directly or indirectly
    are invited for submission. The deadline for submission is October 15,
    The notification of acceptance/rejection will be sent by December 15,
    2001.The Special Issue is scheduled to appear in
    To facilitate the editorial process the interested authors are requested
    to send a short e-mail message to the editors with a
    tentative title of their intended contribution.
    Guest Editors:
    Danuta Rutkowska, E-mail:
    Janusz Kacprzyk, E-mail:
    Lotfi A. Zadeh, E-mail:

    You are kindly invited to participate with presentation at the session
    paradigms for understanding society" of the XVth World Congress of
    Sociology (Brisbane, Australia, 7-13 July, 2002).
    More information about the session you can find at

    The deadline for a 500-1000 word abstract (sent by E-mail to me) is 31
    August 2001.

    Kind regards,


    Dear Colleague

    I would like to invite you to attend the
    IEEE International Conference on Software Maintenance, 2001, and
    associated workshops: IEEE SCAM, IEEE WESS, IEEE WSE, TABOO.
    FLORENCE, ITALY, 6-10 November 2001

    Sponsored by IEEE
    Supported bt the: EC-IST, University of Florence, O-Groupi, IBM Italy
    in collaboration with: TABOO, AICA, AIIA, ERCIM, UNINFO, CESVIT, ...

    ICSM is the major international conference in the field of software and
    systems maintenance, evolution, and management.

    Theme: Systems and Software Evolution in the era of the Internet
    kEYWORDS: software evolution, embedded suystems, program analysis,
    reengineering, managment, maintenance, lyfe cycle, Internet and
    systems, Multimedia systems, User interface evolution, Commercial
    (COTS), Program comprehension, Formal methods, Empirical studies,
    Testing and regression testing, Measurement of software, METRICS,etc.

    Outstanding Keynotes such as:
    Prof. David Lorge Parnas and Prof. Dieter Rombach. Kent Beck
    110 technical presentations, 4 workshops,
    Industrial papers and experiences, reseach papers and award, tutorials,
    tool expositions, dissertation forum and award, workshops, panels,
    and other exciting activities have been planned.

    Please forward the following to anybody who you think may be interested.

    The discount for the advanced registration fee will be active for few

    Apologies for multiple receptions.
    If you would like to be removed from our list please send an email to with REMOVE in the subject.

    Paolo Nesi
    (ICSM2001 General Chair)


    We apologize if you receive multiple copies of this message.

    Dear Colleagues,

    We are organizing an exciting event: HIS'2001: International Workshop on

    Hybrid Intelligent Systems in conjunction with The 14th Australian Joint

    Conference on Artificial Intelligence (AI'01).


    (Technically co-sponsored by The World Federation of Soft Computing)

    HIS'01 is an International Workshop that brings together researchers,
    developers, practitioners, and users of neural networks, fuzzy inference

    systems, evolutionary algorithms and conventional techniques. The aim of

    HIS'01 is to serve as a forum to present current and future research
    work as
    well as to exchange research ideas in this field.

    HIS'01 invites authors to submit their original and unpublished work
    demonstrate current research using hybrid computing techniques and their

    applications in science, technology, business and commercial.

    Topics of interest include but not limited to:

    Applications / techniques using the following, but not limited to:

     * Machine learning techniques (supervised/unsupervised/
       reinforcement learning)

     * Neural network and evolutionary Algorithms

     * Neural networks and fuzzy inference systems

     * Fuzzy clustering algorithms optimized using evolutionary

     * Evolutionary computation (genetic algorithms, genetic
       programming ,evolution strategies, grammatical evolution etc.)

     * Hybrid optimization techniques (simulated annealing, tabu search,
       GRASP etc.)

     * Hybrid computing using neural networks - fuzzy systems -
       evolutionary algorithms

     * Hybrid of soft computing and hard computing techniques

     * Models using inductive logic programming, logic synthesis,
       grammatical inference, case-based reasoning etc.

     * Other intelligent techniques ( support vector machines,
       rough sets, Bayesian networks, probabilistic reasoning,
       minimum message length etc)


    In soft computing, intelligent control theory, and in "data mining,"
    is a "simple" basic question which has been revisited again and again by

    many people:

    How can any system (brain or software...) learn to approximate a
    mapping from a vector of inputs X to a vector of outputs Y, when given a

    database (fixed or real-time) of examples of X and Y?

    (One example: Shankar Shastri of Berkeley and his student Claire Tomlin
    have done excellent work in "hybrid control" -- which ends up requiring
    nonlinear function approximator in the insides of the design. In fact,
    is all quite close
    to what we have done with approximate dynamic programming or
    different words, different spins, but the same underlying mathemnatics.)

    In neural networks, we call this the "supervised learning task." In
    logic, Jim Bezdek
    has called it "system identification." But in any case... one cannot
    systems capable of
    brain-like decision capability without a subsystem that can perform that

    task (among others).
    Thus I would argue that no model of learning in the brain could capture
    higher abilities
    of the brain, UNLESS it had enough richness to be able to handle this
    simple task.


    Here is my concern: in the last few years, there has been a certain
    of drifting apart
    between the computational neuroscience world and the world of
    Many people believe this is just fine... but what if the consequence is
    that the neuroscience side
    is dominated by models which cannot possibly explain the basic learning
    capabilities of the system they
    are studying? Perhaps there is a great need for some new mathematical
    results which would explain
    more clearly what the problem is, and encourage more interest in the
    of model which can solve it.

    (By the way, if anyone is interested... research on these lines would
    well as one of the many topics of
    great interest to what we fund in computational intelligence...)

    Moe precisely:

    Most people on this list probably know already that many types of ANN
    fuzzy system are "universal approximators,"
    that they can learn any smooth mapping from X to Y.

    Some of you may know about the very important results of Andrew Barron
    (statistician at Yale), related to some results
    of Sontag of Rutgers: he proved, in effect, that some universal
    approximators are a lot better than others.
    There are lots of simple "smoothed lookup table" approximators which
    fine for low dimensions...
    but the number of parameters or hidden units required grows
    with the number of inputs. But for multilayer perceptrons (MLPs)
    the growth in complexity is only polynomial. This is an incredibly
    important result. It says that MLPs may be viable
    for large (brain-like) induction problems, while those others are not.
    usual theorems for fuzzy logic approximation
    and RBF approximation(and fuzzy ARTmap) are all based on some kind of
    linear basis function argument, or someting very close to it,
    which would imply an exponential growth in terms.

    Now: DO THOSE RESULTS show that MLPS trained with backprop can perform
    basic task of approximating at least smooth
    nonlinear functions, and scaling up, while fuzzy and Hebbian systems
    cannot? (If virtually all models now used in
    computational neureoscience are of the Hebbian variety, continuous
    or discrete (spiking), this is serious...)

    Not quite.

    For example... when I think about Elastic Fuzzy Logic (ELF, first
    by me in the IIzuka 1992 proceedings, pretty much equivalent to
    parts of some of the later designs of Yager and Fukuda)....

    it is clear that feedback to redefine the membership functions and so on

    can achieve a lot better, more parsimonious
    approximation ability than mere preset lookup tables! I would conjecture

    that ELF
    could also achieve Barron-like capability. And Cooper (of Nestor) has
    played similar games
    of tuning hidden units... long ago...

    BUT: to achieve all this, one needs a FEEDBACK to train/select those


    One may define a general concept of H-locality (Hebb-locality), similar
    some of the rules Grossberg has
    discussed, which prohibit both backpropagation and other similar types
    feedback. The conjecture
    is that ANY system made of simple units, whose learning must obey
    H-locality, can never acheive
    Barron-like polynomial scaling in approximating smooth functions. (
    Technical point: **IF** one allows sneaky re-use
    of weights, as in some of the multiplexing schemes Hecht-Nielsen has
    about, one can actually
    implement backpropagation itself in networks which obey some of the
    Grossberg rules -- but is this a plausible way to model
    biological systems? Or do the biologists implicitly add further rules
    rule out this kind of multiplexing?)

    In any case, to prove or disprove this would be of ENORMOUS scientific

    Proving it... would eliminate the main reason for not using GENERALIZED
    backpropagation more in neuroscience models...
    Disproving it would presumably lead to new Hebbian-style learning models

    that could actually work on the difficult
    kinds of tasks which engineers need to address.

    More capable neuroscience models would of course make it much more
    plausible for engineers to seriously consider trying to use/miomic such
    in addressing difficult computational tasks.

    Either way, the resolution of this issue will be important to bridging
    deep divide between "machine learning" (in the largest
    sense) and studies of learning in the brain.

    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>

    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
    (2) To unsubscribe from this mailing list, send a message body of
    (3) To reach the human who maintains the list, send mail to
    (4) WWW access and other information on Fuzzy Sets and Logic see
    (5) WWW archive:

    This archive was generated by hypermail 2b30 : Mon Jul 16 2001 - 13:14:41 MET DST