EE4305 Fuzzy/Neural Systems for Intelligent Robotics

PART II: FUZZY SYSTEMS

Chapter 2: Fuzzy Sets and Crisp Sets

Topics to be Covered…

• Fuzzy sets and crisp sets

• Fuzzy operations, fuzzy relations, fuzzy compositions
• Extension principle, fuzzy numbers
• Approximate reasoning, fuzzy inference
• Multi-Rule Fuzzy Inference
• Fuzzy knowledge based control (FKBC)
• Fuzzy applications -
 Crisp Sets and Fuzzy Sets

Crisp Sets: A brief overview

What is crisp set?
Crisp set is the “set” we have learned in the past, which is one of the
most fundamental concepts in mathematics.
A set is a gathering together into a whole of definite, distinct objects of our
perception or of our thought—which are called elements of the set.
For example, the numbers 0 and 1 are distinct objects when considered
separately, but when they are considered collectively they form a single set
of size two, written {1, 0}.
 Crisp Sets and Fuzzy Sets

Crisp Sets: A brief overview

To distinguish between fuzzy sets and classical (nonfuzzy) sets, we refer to the
latter as crisp sets.
♦ Let X denotes the universe of discourse, or the universal set. This set
contains all the possible elements of concern in each particular application.
♦ The set A whose elements are a1, a2,…, an is written as A = {a1, a2,…, an}

♦ The number of elements that belong to a set A is called the cardinality of

the set and is denoted as A
♦ The family of sets consisting of all the subsets of a particular set A is
referred to as the power set of A and is indicated by P(A)
♦ What is the power set of A = {a, b, c} and the cardinality of this
power set?
P(A) = { Ø, {a}, {b}, {c}, {a, b}, {a, c}, {b, c}, A }
P ( A) = 23 = 8
→ The cardinality of the power set is P( A) = 2 A
 Crisp Sets and Fuzzy Sets

What are the methods to describe a set?

The simplest one is to list out all the members (the list method).
♦A = {a1, a2,…, an}

This method can be used only for finite sets. If a set is not finite, how
do we define a set?

A set is defined by a property satisfied by its members (the rule method).

♦A = {x|P(x)}
It is required that the property P be such that for any x, P(x) is either true or false.

For example, A = {x|x>0}.

Is there any other way to define a set?

Yes. Characteristic function!

♦ Characteristic function declares which elements of X are members of the set
and which are not.

Characteristic function χA is defined by

 1 if and only if x ∈ A
χ A ( x) = 
 0 if and only if x ∉ A

The characteristic function maps elements of X to the elements of the set {0,1}.

For each 𝑥𝑥 ∈ 𝑋𝑋, when 𝜒𝜒𝐴𝐴 𝑥𝑥 = 1, x is declared to be a member of A;

when 𝜒𝜒𝐴𝐴 𝑥𝑥 = 0, x is declared as a non-member of A.

χA(x)
A = {x|x>0} On
1

Off
0 x
 Basics of Fuzzy Sets
The characteristic function of a crisp set assigns a value of either 1 or 0 to
each individual in the universal set, thereby discriminating between
members and non-members. So the membership can only be 1 or 0.

The real-world is complex:

− Uncertainty, fuzziness inherent in our natural language
− A “Tall person?”, e.g., > 1.8 m or > 1.7 m
− “Fast”, e.g., speed > 50 km/h or speed > 80 km/h

♦ Fuzzy set theory uses linguistic variables, rather than quantitative

variables, to represent imprecise concepts:
E.g., “nearly 6-feet tall”, “very hot”, “rather small”

To deal with the human concepts which do not have sharp boundaries, the
characteristic function can be generalized such that the values assigned to the
elements fall within a specified range and indicate the membership grade of
these elements in the set in question.

Larger values denote higher degrees of set membership.

♦ How to represent the concept such as “hot ” so that machines can
communicate with humans?

♦ How strongly do you agree that a given number (Temperature)

← [0, 100] is a “hot” temperature that belongs to “Hot”?

3 Types:
(1) Threshold type

50° C

0 40° C 100° C

Hot = { x | x ≥ 40°C }
(2) Conservative type

0 20° C 50° C 60° C 100° C

Not Hot Hot

(3) Compromiser type

degree of
“agreement”

0 20° C 50° C 60° C 100° C

♦ A fuzzy set is a set containing elements that have varying degrees of
memberships in the set.
♦ Larger values of the membership denotes higher degrees of the
membership, and such a function is called a membership function.
♦ The membership function µA by which a fuzzy set A is usually defined
has the form µA : X → [0, 1]. ( Compared to χA : X → {0, 1} )

A x1 Fuzzy set A = approximately equal to 3

x2 µ A (x )
x3
..
1.0
xn 0 0.5 1
X 0.7

X
x3 A
x1
2 2.6 2.8 3 3.6 4
x2 2.8: How much close to 3?
µ A (2.8) = 0.7
♦ A possible membership function for the fuzzy set of real numbers close
to 0 can be defined as:
µA(x)
1

1
µA(x) =
1 + 10 x 2

-2 2 x

1
µA(x) =
1+ x2
♦ We may design different membership functions for the same concept.
For instance, to express the concept of “ real numbers that are close to 2”,
we may have :

Which one is better?

It turns out that many applications are not sensitive to variations of the shape
of the membership function. It is convenient to use a simple shape such as the
triangular shape.
♦ Triangular membership function: Depends on three parameters

 0 x≤a
x −a
 a≤ x≤b
∆( x; a, b, c ) =  b − a
c−x
 b≤ x≤c
c−b
 0 c≤x

a b c

The parameters a and c locate

the base of the triangle and the
parameter b locates the peak.
♦ Trapezoidal membership function: Depends on four parameters

 0 x≤a
x−a
 a≤ x≤b
b − a
Π ( x; a, b, c, d ) =  1 b ≤ x ≤ c
d − x
d − c c ≤ x ≤ d
 0 d≤x

a b c d

♦ The parameters a and d locate the base of the trapezoid and

the parameters b and c locate the shoulders.

♦ Example: Suggest a membership function for a fuzzy set of low

temperature with a universe of discourse X ∈ [0, 100].
Fuzzy sets vs. crisp sets for low, medium and high temperatures.
Representation of Fuzzy Sets

♦ Fuzzy set A for a discrete and finite universe of discourse X is

A = µA(x1)/x1 + µ A(x2)/x2 + … + µ A(xn)/xn
n
= ∑ µ A ( xi ) / xi
i =1
1
1 / 2 VS = 0.5
E.g. Integer close to 10 2

= 0.1/7 + 0.5/8 + 0.8/9 + 1/10 + 0.8/11 + 0.5/12 + 0.1/13

♦ For continuous and infinite set, the fuzzy set is: A = ∫ µ A ( x ) / x

x


 0 , 0 < x < 40
 −1
E.g. Fuzzy set Old µ A ( x ) =  −2 
 x − 40  
1 +   , 40 < x < 120

  5   
 Example: What is the fuzzy set A?

µA(x)
1

0.5

0 10 20 30 40 50 x

A = 0.5/20 + 1/30 + 0.5/40

♦ (Def) Height of a fuzzy set A is hgt(A) = the maximum value of the
membership function, i.e., supreme{µA(x)} = sup{µA(x)} = max{µA(x)}
: ‘Normal’ if hgt(A) = 1
: ‘Subnormal’ if hgt(A) < 1
♦ (Def) Support of a fuzzy set A, Supp(A) is the crisp set of all x ∈ X such
that µA(x) > 0, Supp(A) = {x ∈ X | µA(x) > 0}

♦ (Def) Core of a fuzzy set A, Core(A) is the crisp set of all x ∈ X such
that µA(x) = 1, Core(A) = {x ∈ X | µA(x) = 1}

µA(x)

Core
1
Height

0 Support x
For crisp set, since an individual x, is either a member or not a member of a set A,
we can use 𝑥𝑥 ∈ 𝐴𝐴 to denote that x is a member of A.

Can we use the same notation for fuzzy set?

No, for fuzzy set, we can only use the membership function, µ𝐴𝐴 𝑥𝑥 (or A(x)),
to indicate its membership grade.

How do we find out whether two fuzzy sets A and B are the same?

A=B if µ𝐴𝐴 𝑥𝑥 = µ𝐵𝐵 𝑥𝑥

We need to re-define all the concepts and operations associated with the crisp sets
using the membership function.
How do we find out whether a fuzzy set A is a part of fuzzy set B? In other words,
how do we define the subset?

To answer this question, let’s see how we do that for crisp sets using the
characteristic functions first.

Let A=[0,2], B=[0, 1]. Clearly B is a subset of A. Their characteristic functions are

How do we relate these two characteristic functions?

𝜒𝜒𝐵𝐵 (𝑥𝑥) ≤ 𝜒𝜒𝐴𝐴 (𝑥𝑥 )
How do we find out whether a fuzzy set A is a part of fuzzy set B?

𝐵𝐵 ⊆ 𝐴𝐴 iff µ𝐵𝐵 (𝑥𝑥) ≤ µ𝐴𝐴 (𝑥𝑥 )

For example,

𝐵𝐵 ⊆ 𝐴𝐴

How about this?

B is not a subset of A!
From this exercise, we have a systematic way (SOP) to extend the concepts
and operations from crisp sets to fuzzy sets.

First, try to find out the definition of the operations for

crisp sets in terms of the characteristic functions.

Since the characteristic functions are special cases of the membership

functions, we just extend the definition by replacing the characteristic
functions with the membership functions.

We are going to follow this procedure to define other concepts and

operations of fuzzy sets.
What is the cardinality of the crisp set? 𝐴𝐴 = � 𝜒𝜒𝐴𝐴 (𝑥𝑥 )
𝑥𝑥∈𝑋𝑋

♦ (Def) Scalar Cardinality Example

SC( A) = A = ∑µ
x∈ X
A ( x) (ages) old
5 0
♦ (Def) Relative Cardinality 10 0
A 20 0.1
SC ( A)
RC ( A) = = 30 0.2
Universal Set X 40 0.4
50 0.6
60 0.8
70 1
♦ Scalar Cardinality SC(A) 80 1

old = 0 + 0 + 0.1 + 0.2 + 0.4 + 0.6 + 0.8

+ 1 + 1 = 4.1
♦ Relative Cardinality RC(A)
4.1
old = = 0.4556
9