- Newsgroups: comp.ai.fuzzy,sci.stat.math
- From: predictr <predictr@bellatlantic.net>
- Subject: Re: Fuzziness as opposed to Probability
- Date: Fri, 03 Aug 2001 00:47:10 GMT
- Organization:
Will Dwinnell wrote:
"While I have not studied these issues obsessively, I do tend to agree
with the fuzzy critics' general complaint that too much has been made
of fuzzy logic.
...
On the other hand, people have built fuzzy systems that work, that is,
which solve the problems for which they were intended. To me, it seems
that issues like whether they could have been built using some other
formalism (be it probability or somthing else) are less important than
issues of economy and effectiveness."
Rich Ulrich responded:
"Are you saying, they're okay because they seem to work?
That's dangerous if you don't look any closer."
I agree that such uncritical thinking would be dangerous, but what I
meant by "these issues" [very top, above] was all the other,
philosophical stuff which has been made of fuzzy logic (a la Kosko, for
example). My point is that none of that theoretical hand-waving seems
important to the question of whether a real fuzzy logic system does its
job or not.
Rich Ulrich continues:
"More than a century ago, the magic elixirs -- all those products of the
"snake-oil salesman" in the U.S. -- did tend to work. For *some*
purpose. There *was* an active ingredient or two. Many of them were
100-proof drinking-alcohol (or more),
and some had opium.
As a medical potion, the dosing would slip past a lot of the "dry's";
and the retail price would be a big markup from the saloon."
Are you saying that no fuzzy systems actually solve the problems they
were intended to solve? I agree that claims of effectiveness must be
tested, but it is my assertion that at least some implemented fuzzy
systems work as advertised.
Rich Ulrich continues:
"It has been my impression that "neuro nets" and "fuzzy logic" have
been sold just as rationally as that, PROBABLY, where they do exist in
commercial products. That is: people have bought a black box, with
mystical assurances, paying several times what the price should be.
Criticism is avoided by being obscure. An extra cost of no-criticism
is no-improvement of the underlying model, whatever that was."
I cannot speak to the cost issue: that is a matter for the market. As
far as fuzzy logic and neural networks go, to me it is simply an
empirical question of whether they work or not. I think there is ample
case history indicating that they can.
Will Dwinnell
predictor@dwinnell.com