Sec. 2.5. More on Fuzzy Union, Intersection, and Complement* 37
Definition 2.26 T-norm
A T-norm operator [3] is a two-place function T(.,-) satisfying
T(0,0) =0, T(a,1)=T(1,a)=a (boundary)
T(a,b) < T(c,d) ifa0.
The first row of Figure 2.17 shows the surface plot of these T-conorm operators.
The second row demonstrates the corresponding two-dimensional MFs when a =
pa(z) = trapezoid(z;3,8, 12,17) and b = up(z) = trapezoid(y;3,8,12, 17); these
MF5 are the Cartesian coproduct of A and B using these four T-conorm operators.
It can also be verified that
Smaz (a,b) < Sap(a,b) < Stp(a,b) < Sap(a, b)- (2.49)
a
Note that these essential requirements for T-norm and T-conorm operators can-
not uniquely determine the classical fuzzy intersection and union—namely, the min
and max operators. Stronger restrictions have to be taken into consideration to
pinpoint the min and max operators. For a detailed treatment of this subject,
see [1].40 Fuzzy Sets Ch. 2
Theorem 2.1 Generalized DeMorgan’s law
‘T-norms T(-,) and T-conorms S(-,-) are duals which support the generalization of
DeMorgan’s law:
T(a,b) = N(S(N(a), N(b)));
S(a,b) = N(T(N(a), N(b))),
where N(-) is a fuzzy complement operator. If we use # and } for T-norm and
‘T-conorm operators, respectively, then the preceding equations can be rewritten as
a%b = N(N(a)4N(b)),
a+b = N(N(a)#N(b)).
(2.50)
(2.51)
a
Thus for a given T-norm operator, we can always find a corresponding T-conorm
operator through the generalized DeMorgan’s law, and vice versa. (In fact, the four
T-norm and T-conorm operators in Examples 2.12 and 2.13, respectively, are dual
in the sense of the generalized DeMorgan’s law. The reader is encouraged to verify
this.)
2.5.3 Parameterized T-norm and T-conorm*
Several parameterized T-norms and dual T-conorms have been proposed in the
past, such as those of Yager [9], Dubois and Prade [4], Schweizer and Sklar [7], and
Sugeno 8]. For instance, Schweizer and Sklar’s T-norm operator can be expressed
as
Tss(a, 6, p) = (max{0, (a~? + b-? ~ 1) }-*
Sss(a,b,p) = 1 — [max{0, ((1—a)-? + (1-8)? — I)}-*
It is observed that
(2.52)
limy-s0 T'ss(a,b, p) = ab,
limp sao Ts (a,b,p) = min(a,8), Me
which correspond to two of the more commonly used T-norms for the fuzzy AND
operation.
To give a general idea of how the parameter p affects the T-norm and T-conorm
operators, Figure 2.18(a) shows typical membership functions of fuzzy sets A and
B; Figure 2.18(b) and Figure 2.18(c) are Tss(a, 6, p) and Sss(a, b, p), respectively,
with p = co (solid line), 1 (dashed line), 0 (dotted line) and —1 (dash-dotted line).
Note that the bell-shaped membership functions of A and B in Figure 2.18(a) are
defined as follows:
pa(z) = bell(x; —5,2,7.5) = — (2.54)
up(2) = bell@35,1,5) = aaa (2.55)
For completeness, other types of parameterized T-norms are given next.