>In different articles I have seen multi-valued logic and fuzzy-logic used interchangeably.
>Is this true, or is multi-valued logic a subset of fuzzy-logic?
>Can anyone provide an answer or refer me to a source?
>My email address is:
>mark@infosys.agrenv.mcgill.ca
>Thanks
Fuzzy logic is one of an infinite number of multi-valued logics, ranging
from Zadeh's fuzzy logic (a AND b = min(a, b), a OR b = max(a, b) to the
Lukasiewicz logic ( a AND b = max(0, 1 - (a + b)), a OR b = min(1, a + b)).
There is an article by Enrique Ruspini in Computer Magazine in (I
believe) 1980 which discusses this.
The Zadeh fuzzy logic has several advantages over the others. (1) There
is a lot of experience with it; there is a lot of math behind it; and
(very important to me) you can write complex rule antecedents without
having multiple ANDs drift off to zero, and multiple ORs which don't
drift off to one.
It is a big mistake to concentrate on fuzzy logic without at the same
time using fuzzy sets and fuzzy numbers, perhaps of even greater
importance than fuzzy logic. Fuzzy sets permit us to handle ambiguities
and contradictions with ease; this is of great importance in emulating
human reasoning. Fuzzy numbers give us an exceptionally easy way to
handle uncertain numbers. And the use of "hedges" permits describing a
fuzzy number as (for example) "roughly 2".
Of course, fuzzy control methods are here to stay. But fuzzy control is a
very small subset of what is now being called "fuzzy reasoning", which is
much more interesting from an AI standpoint.
------------------------------