Re: BISC Special Interest Gr

Alex Meystel (meystel@cme.nist.gov)
Sat, 28 Dec 1996 12:14:07 +0100


Reply to: RE>BISC Special Interest Group in In
Friends:

Vladik Kreinovich writes:

>last but not the least: the interest in using interval methods is being
>revived by the Zadeh's recent idea of *granularity* as a unifying theme
>for several formalisms, including fuzzy and interval methods.

One cannot say clearer! Indeed, "granularity" is a unifying theme for
many formalisms. It is also a key concept for systems of data and
knowledge representation, for architectures of intelligent systems
and a number of other things including multiresolutional methods.

I propose to use the term "granularity" in the title of the new group
explicitly: Special Interest Group on Granularity (SIGGRAN).
It should be hospitable to all research directions which employ
tessellated representations, no matter how this tessellation is
obtained. All methods of discretization, quantization, etc. will be
included. The goal of this group will be to consider all these diversified
techniques from a unified, generalized standpoint.

I think, we should commend Vladik on this important initiative!
I would be happy to participate in functioning of the group organized by
Professor Kreinovich, and volonteer to submit an article for his Web
page: "Indistinguishability, Epsilon-Net, and Minimum-complexity
Representation".

Alex Meystel

--------------------------------------
Date: 12/23/96 9:58 PM
To: Alex Meystel
From: Vladik Kreinovich
Dear Colleagues,

PROPOSAL.
We would like to suggest a new BISC Special Interest Group: a Special
Interest Group on Interval Methods in Knowledge Representations
(SIGINT).

THERE IS A GROWING INTEREST IN INTERVAL METHODS.
Interval methods are naturally related to fuzzy research:
* first, many operations with fuzzy numbers can be naturally
implemented as operations with their alpha-cuts (i.e., intervals);
not surprizingly, interval arithmetic is described in most modern
textbooks on fuzzy logics and fuzzy systems;
* second, in many cases, it is more natural to use not numbers
but intervals to describe the values of membership functions; the
resulting interval-valued fuzzy sets are very helpful in expert
systems, fuzzy control, pattern recognition, etc.
* last but not the least: the interest in using interval methods is being
revived by the Zadeh's recent idea of *granularity* as a unifying theme
for several formalisms, including fuzzy and interval methods.

There is a growing interest in interval methods:
* talks on interval methods in knowledge representation are
actively present both at conferences in Interval Computations
and at conferences on Fuzzy systems. Several relevant
workshops and special sections have been organized: e.g., during the
1993 Interval Computations conference, the 1994 NASA/NAFIPS, etc.
* new papers appear all the time in journals and conference proceedings:
the recently started special section on interval methods
in IJUFKS (Intl. Journal on Fuzziness, Uncertainty, and
Knowledge-Based Systems) already has a backlog of abstracts of
different relevant papers.
* lots of relevant papers appear in the interval journal
Reliable Computing and in the special NIFS journal specifically
devoted to interval-related intuitionistic fuzzy sets. A recently
announced IJUFKS special issue on interval methods is already filled
with high-quality papers.

THERE IS A NEED TO ORGANIZE OURSELVES: Researchers who do research in
interval methods in knowledge representation come from different
backgrounds: mainly from fuzzy systems and from interval methods.
As a result, we are
often unaware of previous work, and we spend quite some time
re-inventing the notions and results that are already well known in
other areas. People who do similar research can definitely profit from
a more active interaction. Workshops and special sections are in
order.

Journal editors would definitely appreciate having a list of people who
are interested in interval methods so that interval-related papers
could be sent to referees knowledgeable in the corresponding areas.

INITIATIVE GROUP. The idea of organizing a special interest group
re-appeared during several recent fuzzy and interval meetings. Here is
a list of people who have so far expressed interest in coordinating this
group (in alphabetic order)
* Krassimir Atanassov (Academy of Sciences, Bulgaria)
* Ladislav Kohout (Florida State University, Tallahassee)
* Vladik Kreinovich (University of Texas at El Paso)
* Weldon Lodwick (University of Colorado at Denver)
* Patricia Nava (University of Texas at El Paso)
* Hung T. Nguyen (New Mexico State University)
* I. Burhan Turksen (University of Toronto, Canada)
Our emails are in the header of this message.

VOLUNTEERS ARE WELCOME. This is by no means an exhaustive list.
Volunteers are definitely welcome. Please send your responses to Vladik
Kreinovich at vladik@cs.utep.edu.

We may want to make a web site (or at least, a section of the interval
computations website http://cs.utep.edu/interval-comp/main.html), a
mailing list, etc.