To The Editor:
I am writing concerning the December 1996 article on Fuzzy Logic under Technology Focus.
The simplifying fuzzy logic diagram does not simplify anything. Admittedly, I dropped out of graduate school after five years (mathematics) so I need some help here.
Just what is that diagram supposed to mean anyway?
Input samples from each of the ranges map to the same numerical value on the fuzzy data value ordinate scale. That appears to be nonsensical. And just what is a fuzzy data value supposed to be?
Similarly, given any fuzzy data value, there are at least five and as many as an infinite number of input samples that correspond to it.
This is more than fuzzy; it is downright murky.
I read Lotfi Zadeh`s notes on fuzzy logic in the late `60`s while I was working at IBM. At least then, the best that I can recall, his paper made some sense. This article does not. But then, I have not seen any in any popular journal that has. Except perhaps for the columns in ED that showed that fuzzy logic is nonsense for engineering use.
It certainly would be useful to explain why the maximum membership in any of the fuzzy classes is only 63 percent - if that is what the fuzzy data value is intended to represent.
Shouldn`t the total values add up to l00 or not? Why not? If something is not 100 percent absolutely hot then shouldn`t it be considered to some extent to be warm or moderate too? And if it is 100 percent in a single membership class shouldn`t the membership value be 100 percent? It would appear that the fuzzy data value is not a membership value. What is it? Inquiring minds want to know.
Hopefully, the next article will be more complete and accurate. Clearer would help too.
William B. Adams
consultant
Springfield, Va.
