Fuzzy sets.

Operations with fuzzy sets

Chapter 2

Doru Todinca

Departament of Computers and Information Technology

UPT
Outline

Fuzzy sets

Properties of fuzzy sets

Operations with fuzzy sets

Properties of the operations with fuzzy sets
Outline

Fuzzy sets

Properties of fuzzy sets

Operations with fuzzy sets

Properties of the operations with fuzzy sets
Crisp (classic) sets, fuzzy sets

◮ Given an universe of discourse (crisp) X

◮ For a classic (crisp) set A ⊂ X , for each element x ∈ X , either
x ∈ A or x ∈ / A.
◮ For the set A it can be defined a characteristic function
νA : X → {0, 1}, with νA (x) = 1 iff (if and only if) x ∈ A and
νA (x) = 0 iff x ∈/A
◮ For a fuzzy set Ã, an element x ∈ X belongs to the fuzzy set
à ⊂ X in a certain degree
◮ The characteristic function of a crisp set will be extended to
the membership function of a fuzzy set, which can take values
in the real numbers interval [0, 1]
Definitions

Definition
“If X is a collection of objects” (named the universe of discourse)
“denoted generically by x, then a fuzzy set à ⊂ X is a set of
ordered pairs
à = {(x, µÃ (x))|x ∈ X }
where µÃ (x) : X → [0, 1] is called membership function or degree
of membership (also, degree of compatibility or degree of truth).
of x in A” (Zimmermann [Zim91])
If the interval of real numbers [0, 1] is replaced with the discrete
set {0, 1}, then the fuzzy set à becomes a classic (crisp) set.
Fuzzy sets. Examples of fuzzy sets

◮ Fuzzy sets can be discrete or continuous

◮ The interval [0, 1] can be extended to [0, k], where k > 0
◮ It is possible to define fuzzy sets on more complex structures
than intervals or real numbers, e.g. L-fuzzy sets, where L is a
partially ordered set (see chapter 3, Extensions of fuzzy sets)
◮ Example of discrete fuzzy set (Zimmermann [Zim91]):
◮ MF: comfortable house for a 4 person family as a function of
the number of bedrooms:
◮ The universe discourse: X = {1, 2, . . . , 10}
◮ Ã ⊂ X will be
à = {(1, 0.1), (2, 0.5), (3, 0.8), (4, 1.0), (5, 0.7), (6, 0.2)}
Examples of fuzzy sets (cnt’d)

Example of continuous
fuzzy set: real numbers 1
1/(1+(x-10)**2)

0.9
close to 10 0.8

◮ X = R (the set of 0.7

0.6

real numbers) 0.5

◮ The membership 0.4

function of the fuzzy 0.3

0.2

set à ⊂ R: 0.1

0
1 -10 -5 0 5 10 15 20

µÃ (x) =
1 + (x − 10)2 1
Figure 1: à with µÃ (x) = 1+(x−10)2
(1)
Examples of fuzzy sets (cnt’d)

Example of a continuous fuzzy

set: real numbers considerably 1

0.9
funcB(x)

larger than 11 0.8

0.7

◮ X = R (the set of real 0.6

0.5

numbers) 0.4

0.3

◮ The membership function of 0.2

the fuzzy set:B̃ ⊂ R: 0.1

0
0 5 10 15 20
( (x−11)2
2 if x ≥ 11
µB̃ (x) = 1+(x−11) Figure 2: B̃ with µB̃ (x)
0, if x < 11
(2)
Notations for fuzzy sets
1. Pairs (element, value) for discrete fuzzy sets (like in the
example with the comfortable house), respectively (generic
element, membership function) for continuous fuzzy sets: e.g.
(x, µÃ (x))
2. Solely by stating the membership function (for continuous
fuzzy sets)
3. As a “sum” for discrete fuzzy sets, respectively “integral” for
continuous fuzzy sets (this notation may create confusions !!):

n
X µÃ (xi ) µ (x1 ) µÃ (x2 ) µ (xn )
à = = à + + . . . + Ã
xi x1 x2 xn
i =1

µÃ (x)
Z
à =
x
Caution, there are neither sums nor integrals here, these are
only notations !!!
Outline

Fuzzy sets

Properties of fuzzy sets

Operations with fuzzy sets

Properties of the operations with fuzzy sets
Properties (characteristics) of fuzzy sets: normal fuzzy sets

1. Normal fuzzy sets

◮ A fuzzy set is called normal if supx µÃ (x) = 1, where sup is
the supremum of a fuzzy set
◮ The difference between the maximum and the supremum of a
set: the maximum belongs to the set, the supremum may
belong or not to that set
◮ If a fuzzy set is not normal, it can be normalized by dividing
its membership function by the supremum of the set, resulting
the normalized fuzzy set:
µÃ (x)
µÃnorm (x) =
supx µÃ (x)
Properties of fuzzy sets: support, core, boundary
2. The support of a fuzzy set
◮ The support of a fuzzy set (denoted supp) is the crisp set of
all x ∈ X for which µÃ (x) > 0
◮ In the example with the comfortable house it is the set
supp(Ã) = {1, 2, 3, 4, 5, 6, 7}
◮ Usually the elements of a fuzzy set having the degree of
membership equal to 0 are not listed
3. The (core) of a fuzzy set:
◮ is the crisp set for which µÃ (x) = 1
4. The (boundary) of a fuzzy set:
◮ is the crisp set for which 0 < µÃ (x) < 1

Exercise: represent graphically the support, the core and the

boundary for a continuous trapezoidal fuzzy set.
Properties of a fuzzy set: α-level sets
5. The α-level sets ( or α-cuts):
◮ The α-level set (where α ∈ [0, 1]) of the fuzzy set à having
the membership function µÃ (x) is the crisp set Aα for which
µÃ (x) ≥ α
◮ We can define strong α cut as the crisp set A′α for which
µÃ (x) > α
◮ In the example with the comfortable house, WHERE
à = {(1, 0.1), (2, 0.5), (3, 0.8), (4, 1.0), (5, 0.7), (6, 0.2)}, the
α-cuts of the fuzzy set à are:
◮ A0.1 = {1, 2, 3, 4, 5, 6} = supp à (the support of Ã)
◮ A0.2 = {2, 3, 4, 5, 6}
◮ A0.5 = {2, 3, 4, 5}
◮ A0.7 = {3, 4, 5}
◮ A0.8 = {3, 4}
◮ A1.0 = {4} = coreÃ
Properties of a fuzzy set: α-level sets

◮ It can be proved that for any fuzzy set Ã, it holds:

[
à = α · Aα
α

◮ Which means that, any fuzzy set can be written as the union
for all the values of α of the product between α and the
α-cuts of the fuzzy set
◮ This property is very important and it connects the fuzzy and
the crisp sets
◮ It is also very useful for proving different properties of fuzzy
sets (some properties are easier to be proved for crisp sets)
Properties of a fuzzy set: α-level sets

◮ We will illustrate this property on the example with the

comfortable house:
◮ α · Aα is the fuzzy set in which each element will hace the
membership function equal with α.
◮ 0.1 · A0.1 = {(1, 0.1), (2, 0.1), (3, 0.1), (4, 0.1), (5, 0.1), (6, 0.1)}
◮ 0.2 · A0.2 = {(2, 0.2), (3, 0.2), (4, 0.2), (5, 0.2), (6, 0.2)}
◮ ...
◮ 0.8 · A0.8 = {(3, 0.8), (4, 0.8)}
◮ 1.0 · A1.0 = {(4, 1.0)}
◮ The union of two or more fuzzy sets is defined as the
maximum between their membership function, hence
◮ 0.1 · A0.1 ∪ 0.2 · A0.2 ∪ . . . ∪ 0.8 · A0.8 ∪ 1.0 · A1.0 =
= {(1, 0.1), (2, max(0.1, 0.2)), (3, max(0.1, 0.2, . . . , 0.8)),
(4, max(0.1, . . . , 0.8, 1)), . . . (6, max(0.1, 0.2)} = Ã
Properties of fuzzy sets: convexity
6. Convexity of a fuzzy set
◮ A fuzzy set à ⊂ X is convex if and only if ∀x1 , x2 ∈ X and
∀λ ∈ [0, 1] the following relation takes place:
µÃ (λ · x1 + (1 − λ) · x2 ) ≥ min(µÃ (x1 ), µÃ (x2 ))
◮ The expression λ · x1 + (1 − λ) · x2 describes the segment
situated between the points having the abscissa x1 and x2
◮ The expression µÃ (λ · x1 + (1 − λ) · x2 ) describes the image of
this segment through the function µÃ (x)
◮ Equivalently, a fuzzy set à is convex iff all its α-level sets
are convex
◮ Which means that, if a fuzzy set is not convex, there exist
α-level sets of this fuzzy set that are not convex, i.e., there
exist segments x1α x2α which are “interrupted” (are not
continues)
Ex: Represent graphically a continuous and convex fuzzy set and a
continuous non-convex fuzzy set.
Properties of fuzzy sets: cardinality
7. Cardinality of a fuzzy set
◮ Cardinality of a finite fuzzy set à ⊂ X , denoted |Ã| is defined
as:
n
X
|Ã| = µÃ (xi )
i =1

◮ For a continuous fuzzy set à ⊂ X , its cardinality is defined:

Z
|Ã| = µÃ (x)dx
x

if the integral exist

7’ Relative cardinality of a fuzzy set
◮ Is denoted ||Ã||
|Ã|
◮ Is defined as ||Ã|| = |X | , if it exists, where X is the universe of
discourse for the set Ã
How to chose the membership functions

◮ Like in other aspects of the fuzzy sets theory, there are no

clear “recipes” for choosing the membership functions of the
fuzzy sets
◮ If we want to reduce the computations, we will prefer linear
membership functions, i.e., triangles and trapeziums
◮ There are cases when we prefer non-linear membership
functions (trigonometric, Gauss-type, etc):
◮ There exist researchers that consider that linear membership
functions do not provide the best results for some problems,
while non-linear functions perform better
◮ Sometimes the problem or the domain might need some types
of membership functions
◮ If we combine fuzzy sets theory with other methods, e.g.,
neural networks, it can be necessary to use membership
functions that are suitable for these methods.
Outline

Fuzzy sets

Properties of fuzzy sets

Operations with fuzzy sets

Properties of the operations with fuzzy sets
Operations for fuzzy sets: union, intersection, complement
◮ Given two fuzzy sets à = {(x, µÃ (x))|x ∈ X } and
B̃ = {(x, µB̃ (x))|x ∈ X } over the same universe of discourse
X , we can define operations of union, intersection and
complement. We define:
◮ the union of the fuzzy sets à si B̃ as the fuzzy set C̃ = à ∪ B̃,
given by C̃ = {(x, µC̃ (x))|x ∈ X }, where

µC̃ (x) = max(µÃ (x), µB̃ (x))

◮ the intersection of the fuzzy sets à and B̃ as the fuzzy set

D̃ = Ã ∩ B̃, given by D̃ = {(x, µD̃ (x))|x ∈ X }, where

µD̃ (x) = min(µÃ (x), µB̃ (x))

◮ the complement of à in X as the fuzzy set Ẽ = Cà X given by

Ẽ = {(x, µẼ (x))|x ∈ X }, where

µẼ (x) = 1 − µÃ (x)

Operations with fuzzy sets: inclusion, equality

◮ inclusion of fuzzy sets: given two fuzzy sets à and B̃ included

in X , the inclusion à ⊆ B̃ takes place iff µÃ (x) ≤ µB̃ (x),
(∀)x ∈ X
◮ equality of two fuzzy sets: two fuzzy sets à and B̃ included in
X are equals iff µÃ (x) = µB̃ (x), (∀)x ∈ X
◮ Equivalently, two fuzzy sets à and B̃ included in X are equals
iff à ⊆ B̃ and B̃ ⊆ Ã
Operations with fuzzy sets: examples

1. Determine the union and intersection of the fuzzy sets à =

“comfortable house for a 4 persons - family” and B̃ = “small
house”, where
à = {(1, 0.1), (2, 0.5), (3, 0.8), (4, 1.0), (5, 0.7), (6, 0.2)} and
B̃ = {(1, 1), (2, 0.8), (3, 0.4), (4, 0.1)}:
Ã∪B̃ = {(1, max(0.1, 1)), (2, max(0.5, 0.8)), (3, max(0.8, 0.4)),
(4, max(1, 0.1)), (5, max(0.7, 0)), (6, max(0.2, 0)} =
{(1, 1), (2, 0.8), (3, 0.8), (4, 1), (5, 0.7), (6, 0.2)}
à ∩ B̃ = {(1, min(0.1, 1)), (2, min(0.5, 0.8)), (3, min(0.8, 0.4)),
(4, min(1, 0.1)), (5, min(0.7, 0)), (6, min(0.2, 0)} =
{(1, 0.1), (2, 0.5), (3, 0.4), (4, 0.1), (5, 0), (6, 0)}
à ∪ B̃ can be read as “comfortable house for a 4 persons -
family or small”, and à ∩ B̃ as “comfortable house for a 4
persons - family and small”
Operations with fuzzy sets: examples (continued)

2. Determine CÃ X , where X = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}:

(“non-comfortable house for a 4 persons - family”)
CÃ X = {(1, 1 − 0.1), (2, 1 − 0.5), (3, 1 − 0.8), (4, 1 − 1), (5, 1 −
0.7), (6, 1 − 0.2), (7, 1 − 0), (8, 1 − 0), (9, 1 − 0), (10, 1 − 0)} =
{(1, 0.9), (2, 0.5), (3, 0.2), (4, 0), (5, 0.3), (6, 0.8), (7, 1), (8, 1),
(9, 1), (10, 1)}
3. Determine the union and intersection of the fuzzy sets à =
“real numbers close to 10” and B̃ = “real number
considerably larger than 11”.
◮ Analytically: C̃ = Ã ∪ B̃ si D̃ = Ã ∩ B̃, where
µC̃ (x) = max{µÃ (x), µB̃ (x)},
µD̃ (x) = min{µÃ (x), µB̃ (x)}
...
◮ Graphically: (more suited in this case), in the next slides:
Example of operations with fuzzy sets: union
1 1
1/(1+(x-10)**2) fA(x)
funcB(x)
0.9 max(fA(x), funcB(x))
0.9

0.8 0.8

0.7 0.7

0.6 0.6

0.5
0.5

0.4
0.4

0.3
0.3

0.2
0.2

0.1
0.1

0
0 0 5 10 15 20
-10 -5 0 5 10 15 20

Figure 3: à with µÃ (x) Figure 5: Ã, B̃ and à ∪ B̃

1
max(fA(x), funcB(x))
1
funcB(x)
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0 0 5 10 15 20
0 5 10

Figure 4: B̃ with µB̃ (x)

15 20
Figure 6: Ã ∪ B̃
Example of operations with fuzzy sets: intersection
1 1
1/(1+(x-10)**2) fA(x)
funcB(x)
0.9 min(fA(x), funcB(x))
0.9

0.8 0.8

0.7 0.7

0.6 0.6

0.5
0.5

0.4
0.4

0.3
0.3

0.2
0.2

0.1
0.1

0
0 0 5 10 15 20
-10 -5 0 5 10 15 20

Figure 7: µÃ (x) Figure 9: Ã, B̃ and à ∩ B̃

0.3
min(fA(x), funcB(x))
1
funcB(x)

0.9
0.25

0.8

0.7 0.2

0.6
0.15
0.5

0.4
0.1

0.3

0.2 0.05

0.1
0
0 0 5 10 15 20
0 5 10

Figure 8: µB̃ (x)

15 20
Figure 10: Ã ∩ B̃ (detail)
Example of operations with fuzzy sets: exercises

1. Determine CB̃ X , where B̃ is the fuzzy set “small house”, and

X = {1, 2, . . . , 9, 10}
2. Determine the complement of a fuzzy set that has a
continuous trapezoidal-shaped membership function
3. For this fuzzy set, determine the union and intersection
between the fuzzy set and its complement. What do you see ?
Outline

Fuzzy sets

Properties of fuzzy sets

Operations with fuzzy sets

Properties of the operations with fuzzy sets
Properties of the operations with crisp sets and fuzzy sets
For crisp sets in the universe of discourse X the following
properties are true (after [NR74]):
1. Commutativity:
A∪B = B ∪A
A∩B = B ∩A
2. Associativity:
(A ∪ B) ∪ C = A ∪ (B ∪ C )
(A ∩ B) ∩ C = A ∩ (B ∩ C )
3. Distributivity:
A ∪ (B ∩ C ) = (A ∪ B) ∩ (A ∪ C )
A ∩ (B ∪ C ) = (A ∩ B) ∪ (A ∩ C )
4. Idempotency:
A∪A = A
A∩A = A
Properties of the operations with crisp sets and fuzzy sets

5. Identity:
A∪∅=∅∪A=A
A∪X = X ∪A = X
A∩∅ = ∅∩A = ∅
A∩X = X ∩A = A
6. Transitivity: if A ⊆ B and B ⊆ C , then A ⊆ C
7. Involution: A = A, where A = CA X
8. De Morgan:
A∪B =A∩B
A∩B =A∪B
Properties of the operations with crisp sets and fuzzy sets

9. Absorption:
A ∪ (A ∩ B) = A
A ∩ (A ∪ B) = A
10. Excluded middle laws (excluded middle laws):

A∪A = X

A∩A=∅

◮ Proprieties 1–9 hold for fuzzy sets, too, but NOT the property
10.
◮ Some researchers consider this fact (non-fulfillment of the
excluded middle laws) as being the main characteristic of
fuzzy sets.
Axiomatization of the operations with fuzzy sets

◮ Bellmann and Giertz proposed a set de axioms (properties)

that should be fulfilled by the union, intersection, and
complement operations with fuzzy sets.
◮ They wanted to see if, based on a set of axioms, we can
obtain also other operations than maximum for union,
minimum for intersection, and 1 − µÃ (x) for intersection.
◮ Bellman and Giertz have shown that only the operators
maximum for union, and respectively minimum for
intersection fulfill their set of axioms
◮ However, for complement they could not obtain an unique
operator.
◮ In order to obtain an unique operator for complement, they
added the condition that the complement of 1/2 should be
1/2.
Conclusions: directions in fuzzy logic

1. The direction followed by mathematicians, who aim to:

◮ on the one side, to give a theoretical foundation to the results,
operators and formulas from fuzzy logic
◮ on the other side, try to extend other domains, mathematical
or non-mathematical, through the framework of fuzzy logic.
◮ Hence, there exists fuzzy numbers, fuzzy arithmetic, fuzzy
functions, fuzzy calculus, fuzzy probabilities, but also fuzzy
automata, fuzzy flip-flops, fuzzy codes, fuzzy reliability, etc
2. The second direction is followed by engineers, economists,
linguists, medical doctors, etc, who apply the results of fuzzy
logic in their domains of activity
◮ They must keep themselves informed on the results obtained
by mathematicians.
CV Negoiţa and DA Ralescu.
Mulţimi vagi şi aplicaţiile lor.
Editura Tehnică, 1974.
H.-J. Zimmermann.
Fuzzy Set Theory – and Its Applications, Second, Revised
Edition.
Kluwer Academic Publishers, 1991.