NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/ai/fuzzy/137

Text File | 1993-01-21 | 2.6 KB | 53 lines

Newsgroups: comp.ai.fuzzy
Path: sparky!uunet!paladin.american.edu!howland.reston.ans.net!usc!news.service.uci.edu!gordius!til!maui!erik
From: [e-mail] (Erik Horstkotte)
Subject: Re: The interested bu lazy guy did some home
Message-ID: <[e-mail]>
Sender: [e-mail]
Nntp-Posting-Host: maui
Reply-To: [e-mail]
Organization: Togai InfraLogic, Inc.
References: <[e-mail]>
Date: Thu, 21 Jan 1993 19:14:11 GMT
Lines: 39

In article [e-mail], [e-mail] (Chris Gray) writes:
>Fuzzyists seem to think it is a feature of fuzzy logic that the rules can
>vary according to the application, and that their controllers function
>quite well even when the membership function and control functions are
>not quite right: this they call "robust", cause of much flaming from the
>classical control guys, who already assigned a meaning to this word
>(sound familiar?).

In fact, the majority of us "fuzzyists" know very well that this is not what
control theory folks call "robustness".  It appears that perhaps you've been
somewhat "poisoned" through having read Bart Kosko's book on fuzzy logic and
neural networks, where he misuses the term this way.  One thing you have to
keep in mind when reading his books, which do have some *useful* things to
say, is that Dr. Kosko has the bad habit of reinventing terminology as he
goes.  He's already somewhat infamous in the fuzzy community for discarding
all of the terminology that was built up over more than two decades about
inference methods, etc., and dealing with fuzzy systems *only* by trying to
make them look like neural networks.  This isn't to say he's a fool.  Just
a pain in the ^&*( at times, and I get tired of having to translate his
work into standard terminology!

I, for one, don't use the term to "robustness" to describe this kind of
insensitivity.  I *do* use the term for some classes of fuzzy control system
that have been empirically demonstrated to be relatively insensitive to
variations in the system being controlled - which *is* very similar to the
classical control theory definition of the term.

And yes, I do get a little annoyed at articles that make it sound like the
people in the fuzzy community are a bunch of identical sheep.  We're all
different, and just like any other group of people there are idiots and
geniuses amongst us.

Grumble mode off.
---
Erik Horstkotte, Togai InfraLogic, Inc.
The World's Source for Fuzzy Logic Solutions (The company, not me!)
[e-mail], gordius!til!erik - [phone]
[e-mail] for info, [e-mail] for fuzzy mail-server