CONCEPTS AND TECHNOLOGIES

91
CONCEPTS AND TECHNOLOGIES

92
CONCEPTS AND TECHNOLOGIES

93
CONCEPTS AND TECHNOLOGIES

94
CONCEPTS AND TECHNOLOGIES

FUZZY SETS
The dichotomous nature of the characteristic function was extended to define the membership
function which is the origin of fuzzy sets Membership function is gradual In a fuzzy set context we
say that the membership of an element is graded Instead of the values for a characteristic function
which are 0 or 1, the values for a membership function can assume any value lying in the interval
[0,1]

UNCERTAINTY BASED MODELS AND ROUGH SET

One of the earliest models of uncertainty is Fuzzy set introduced by L. A. Zadeh in 1965 It introduced
the concept of graded membership of elements in a set instead of dichotomous membership in crisp
sets introduced by G. Cantor in 1820 It is more realistic and has better modelling power Given a
universal set U, a fuzzy subset X of U is defined through a membership function associated with X,
denoted by μX defined as μX:U → [0,1] such that for any x ∈ U, μX(x) =α, where α ∈ [0,1].

MEMBERSHIP FUNCTION
We shall denote the membership function of a fuzzy set A
defined over a universal set U by , which is defined as such that for each ,
Each fuzzy set is completely and uniquely defined by one particular membership function.
• The fuzzy membership functions are not only dependent upon the concept but also upon the context.

95
CONCEPTS AND TECHNOLOGIES

Example

REPRESENTATION OF FUZZY SETS

There are several different ways in which a fuzzy set can be represented. A fuzzy set is denoted by an
ordered set of pairs, the first element of which denotes the element and the second the degree of
membership. Let X be a set. Then a fuzzy set A in X is denoted by the collection of pairs. Example
Let U = {1,2,3,4,5,6}. Then A = {(1, 0.6),(2, 0.3),(4, 0.7),(5,1)} is a fuzzy set on U. Note that the
elements 3 and 6 have membership values 0. A fuzzy set A can be represented as:

The ‘+’ , ‘/’ and symbols have not been used with their usual meanings, these are only
representational.
For example, the fuzzy set example above can be represented as A= + + + .6 1 .3 2 .7 4 1 7.

CHARACTERISTICS OF FUZZY MEMBERSHIP FUNCTIONS

The core of a membership function for some fuzzy set A is defined as that region of the universe that
is characterized by complete and full membership in the set A.
That is, the core comprises of those elements x of the universe such that = 1.

96
CONCEPTS AND TECHNOLOGIES

The support of a membership function for some fuzzy set A is defined as that region of the universe
that is characterized by nonzero membership in the set A.
That is, the support comprises of those elements x of the universe such that >0.
The boundaries of a membership function for some fuzzy set A are defined as that region of the
universe containing elements that have a nonzero membership but not complete membership.
That is, the boundaries comprise those elements x of the universe such that 0 < < 1.
A normal fuzzy set is one whose membership function has at least one element x in the universe
whose membership value is unity.
In fuzzy sets, where one and only one element has a membership equal to one, the element is typically
referred to as the prototype of the set, or the prototypical element.
A convex fuzzy set is described by a membership function whose membership values are
• strictly monotonically increasing,
• strictly monotonically decreasing,
• whose membership values are strictly monotonically increasing then strictly monotonically
decreasing with increasing values for elements in the universe.

if for any elements x, y, and z in a fuzzy set A, the relation x < y < z implies that
then A is said to be a convex fuzzy set.
A special property of two convex fuzzy sets, say A and B, is that the intersection of these two convex
fuzzy sets is also a convex fuzzy set.
That is, for A and B, which are both convex, A ∩B is also convex.
The crossover points of a membership function are defined as the elements in the universe for which a
particular fuzzy set A has values equal to 0.5, that is, for which = 0.5.
The height of a fuzzy set A is the maximum value of the membership function, that is, hgt(A) = max{
ua(x) }.
If the hgt(A) < 1, the fuzzy set is said to be subnormal.
The hgt(A) may be viewed as the degree of validity or credibility of information expressed by A.
The most common forms of membership functions are those that are normal and convex.
However, many operations on fuzzy sets, hence operations on membership functions, result in fuzzy
sets that are subnormal and non-convex.

97
CONCEPTS AND TECHNOLOGIES

Examples

BASIC OPERATIONS ON FUZZY SETS

Union of Sets: For any two sets A and B we define their union as:

This can be rewritten using characteristic functions as:

Extending this union of two fuzzy sets is defined as:

The intersection of two sets A and B is defined as:

Using characteristic functions this can be rewritten as:

Extending this intersection of two fuzzy sets is defined as:

For any two sets A and B the complement of B in A is defined as:

Using characteristic functions it can be rewritten as:

Extending this intersection of two fuzzy sets is defined as:

The complement of a set A in a universal set U is given by:

Using the characteristic function it can be rewritten as:

Extending this the complement of a fuzzy set is defined as:

For any two sets A and B we say that:

Using characteristic functions it can be rewritten as:

Extending this, for any two fuzzy sets A and B we say that:

So, we say that A = B iff:

98
CONCEPTS AND TECHNOLOGIES

Examples

PROPERTIES OF FUZZY SETS

RESULT

99
CONCEPTS AND TECHNOLOGIES

PROPERTIES OF FUZZY SETS

Examples

100
CONCEPTS AND TECHNOLOGIES

CLASSICAL RELATIONS
Let A and B be two sets. Then a relation from A to B is a subset of the Cartesian product A x B A
binary relation on a set A is a subset of A x A Universal relation on A is A x A Identity relation on A
is id A x x {( , )|x A} =  In general a n-ary relation on A is a subset of ... n A AAA =   − n times
Since relations are sets their union, intersection and complements are defined just as for sets.

RELATION IN TERMS OF CHARACTERISTIC FUNCTION

The characteristic function is used to assign values of relationship R as the mapping of the Cartesian
space X × Y to the binary values of (0, 1)

Let R and S as two separate relations on the Cartesian universe X × Y. Then

COMPOSITION OF RELATIONS
There are two common forms of the composition operation • one is called the max–min composition •
the other the max–product composition • Let R be a relation from X to Y and S be a relation from Y to
Z and we define T = R ◦ S, • (max-min)  T ( , ) x z =   ( y Y  R x y   ( , ) ( , )) R • (max-
product)  T ( , ) x z =   ( y Y  R x y   y z ( , ) ( , ))

101
CONCEPTS AND TECHNOLOGIES

FUZZY RELATIONS

PROPERTIES OF FUZZY RELATIONS

The properties of commutativity, associativity, distributivity, involution, and idempotency all hold for
fuzzy relations De Morgan’s principles hold for fuzzy relations.

FUZZY CARTESIAN PRODUCT

Let A be a fuzzy set on universe X and B be a fuzzy set on universe Y, then the Cartesian product
between fuzzy sets A and B will result in a fuzzy relation R, which is contained within the full
Cartesian product space,

102
CONCEPTS AND TECHNOLOGIES

EXAMPLE OF FUZZY CARTESIAN PRODUCT

Let A be a fuzzy set (vector) having four elements Then A can be represented as a column vector of
size 4 × 1 Let B be a fuzzy set (vector) having five elements Then B can be represented as a row
vector size of 1 × 5 The resulting fuzzy relation R will be represented by a matrix of size 4 × 5, that is,
R will have four rows and five columns

COMPOSITION OF FUZZY RELATIONS

Composition of fuzzy relations can be defined just as it is for crisp (binary) relations Suppose R is a
fuzzy relation on the Cartesian space X × Y S is a fuzzy relation on Y × Z Then the composition𝑆𝑜𝑅 =
T is a fuzzy relation on X × Z Then fuzzy max–min composition is defined in terms of the set-
theoretic notation and manner:

103
CONCEPTS AND TECHNOLOGIES

CRISP EQUIVALENCE RELATION

A relation R on a universe X can also be thought of as a relation from X to X A relation R is an
equivalence relation if it has the following three properties: reflexivity symmetry transitivity

POSITIVE EXAMPLES
Parallelism among straight lines in a plane Similarity among triangles in a plane Congruence among
triangles in a plane Over Z, x R y iff x – y is even “Works in the same building as” among workers in
a company “Reads in the same class as” among students in a college NEGATIVE EXAMPLES: “x
R y iff x is a brother of y” among people in a town “x R y iff x is a friend of y” among people in a
country.

TOLERANCE RELATION
A relation R on a set X is said to be a tolerance (proximity) relation iff it is only reflexive and
symmetric but not transitive • Example: • X={1,2,3} and R={(1,1), (2,2), (3,3), (1,2), (2,1)} • NOTE:
If Ris a tolerance relation on X of cardinality n then

FUZZY EQUIVALENCE RELATION

A fuzzy relation R on a single universe X is a fuzzy equivalence relation if all three of the following
properties hold

104
CONCEPTS AND TECHNOLOGIES

FUZZY PROXIMITY RELATION

A fuzzy relation R on a single universe X is a fuzzy proximity relation iff it is fuzzy reflexive and
fuzzy symmetric NOTE: If R is a fuzzy tolerance relation on X of cardinality n then

EXAMPLE

105
CONCEPTS AND TECHNOLOGIES

VALUE ASSIGNMENTS TO FUZZY RELATIONS

The most definitive way for determining value assignments for relations is actually a family of
procedures termed similarity methods introduced by Zadeh in 1971

We shall consider two such methods coming under this broad category
1. Cosine Amplitude
2. Max-Min method

MAX-MIN METHOD
Another popular method, which is computationally simpler than the cosine amplitude method, is
known as the max– min method. In this case the matrix
elements rip

EXAMPLE
• Suppose there was an earth quack in India affecting 5 regions
• Suppose the buildings in the area can be categorized into one of the three categories; no damage,
medium damage and serious damage.
• Suppose the damage ratio in each region is given
by

106
CONCEPTS AND TECHNOLOGIES

• We shall use the cosine formula to derive the membership values between the 5 regions. Here n=5
and
m=3
• So,

• For example,

• Here,

• For example,

FUZZIFICATION

Fuzzification is the process of making a crisp set fuzzy

We do this by simply recognizing that many of the quantities that we consider to be crisp and
deterministic are actually not deterministic at all; they carry considerable uncertainty
If the form of uncertainty happens to arise because of imprecision, ambiguity, or vagueness, then
the variable is probably fuzzy and can be represented by a membership function.
The representation of imprecise data as fuzzy sets is a useful but not mandatory step when those
data are used in fuzzy systems

Example 1:
Suppose in the reading of a voltage, we say it is low voltage
We need not measure it precisely
When we measure it precisely it may be 0.3
That is the membership value of the current voltage in the fuzzy set representing low voltage is
0.3
The membership function for “low” may be given by

107
CONCEPTS AND TECHNOLOGIES

EXAMPLES OF FUZZIFICATION
Example
We can say that somebody is young
He may be 37 years
But when we measure it precisely we say he is 0.92 young

Example
In finding the height of a person, we may say that he is tall.
Actually, his height may be 5 feet 10 inches.
So, taking the height of tall persons, we may say he is 0.95 tall.

DEFUZZIFICATION

PROPERTIES OF LAMBDA CUTS

SOME PROOFS

PROOF OF PROPERTY-4

108
CONCEPTS AND TECHNOLOGIES

λ-

CUTS FOR FUZZY RELATIONS

PROPERTIES OF λ-CUTS OF FUZZY RLATIONS

DEFUZZIFICATION TO SCALARS

There may be situations where the output of a fuzzy process needs to be a single scalar quantity as
opposed to a fuzzy set
Defuzzification is the conversion of a fuzzy quantity to a precise quantity, just as fuzzification is the
conversion of a precise quantity to a fuzzy quantity
The output of a fuzzy process can be the logical union of two or more fuzzy membership functions
defined on the universe of discourse of the output variable

SOME TYPES OF MEMBERSHIP

FUNCTIONS

109
CONCEPTS AND TECHNOLOGIES

Let the two fuzzy outputs be given by and . Let these be given by a trapezoidal subnormal function
and normal triangular function respectively
Their union will be a non-uniform figure (a composition of the two figures)
The number of output functions may be more than 2
The problem is how to express such an output as a single scalar

COMPOSITION OF TWO MEMBERSHIP FUNCTIONS

Composition of a subnormal trapezoidal fuzzy membership function and a normal triangular fuzzy
membership function

110
CONCEPTS AND TECHNOLOGIES

111
CONCEPTS AND TECHNOLOGIES

112
CONCEPTS AND TECHNOLOGIES

113
CONCEPTS AND TECHNOLOGIES

114
CONCEPTS AND TECHNOLOGIES

115
CONCEPTS AND TECHNOLOGIES

116
CONCEPTS AND TECHNOLOGIES

DEFUZZIFICATION
The process of associating non-fuzzy sets with fuzzy sets
Two types exist
Defuzzification to Crisp sets
Defuzzification to scalars
Under defuzzification to crisp sets we have

𝜆−cutsets, 0 ≤ 𝜆 ≤
For any fuzzy set A, it is denoted by Aλ

117
CONCEPTS AND TECHNOLOGIES

DEFUZZIFICATION TO SCALARS
In certain cases a fuzzy process needs to be a single scalar quantity as opposed to a fuzzy set
As mentioned defuzzification is the process of conversion of a fuzzy quantity to a precise quantity
The output of a fuzzy process can be the logical union of two or more fuzzy membership functions
defined on the universe of discourse of the output variable.

METHODS OF CONVERSION TO SCALARS

• There are several methods
• We select only SEVEN of these under two categories
• Group of FOUR and group of THREE
• Group of FOUR

➢ Max membership principle

➢ Centroid method

➢ Weighted average method

➢ Mean-max membership

DIFFERENT APPROACHES TO EXPRESS THE OUTPUT AS A SCALAR

• Max-membership principle:
• Also known as the height method, this scheme is limited to peaked output functions. This method is
given by the algebraic expression

• where z∗ is the defuzzified value

(1) MAX MEMBERSHIP PRINCIPLE

• Also known as height method
• Limitation: Applicable only to peaked output functions
• The output is the point z* (defuzzified value)

118
CONCEPTS AND TECHNOLOGIES

(2) CENTROID METHOD

• Also called as Centre of area or centre of gravity
• Most prevalent and physically appealing of all the defuzzification methods
• DevelopedbySugeno 1985and Lee 1990
• Givenbythealgebraic expression

119
CONCEPTS AND TECHNOLOGIES

• Arailroad company intends to lay a new rail line in a particular part of a county.
• The whole area through which the new line is passing must be purchasedfor right-of-way
onsiderations
• It is surveyed in 3 stretches and the data are collected for analysis
• The surveyed data for the road are given by three fuzzy sets B1, B2 and B3
• For the railroad to purchase the land, it must have an assessment of the amount of land to be
purchased
• Thethreesurveys onright-of-way width are ambiguous
• However, because some of the land along the proposed railway route is already public domain, will
not need to be purchased
• Thethreefuzzysets are shown in thefigures next

120
CONCEPTS AND TECHNOLOGIES

We want to aggregate these three survey results to find the single most nearly representative
right-of-way width to allowthe railroad to make itsinitial estimate of the right-of-way
purchasing cost.
The union of the three surveys is given by

The line OA is joining the two points (0, 0) and (1.5, 1) Its equation is i.e. y = (2/3)x = 0.67x

• We shall use the centroid method to find z* • z*using the centroid method is given by

121
CONCEPTS AND TECHNOLOGIES

(3) WEIGHTED AVERAGE METHOD:

• This method is most frequently used in fuzzy application since it is one of the more
computationally efficient methods
• It is usually restricted to symmetrical output membership functions
• The output z* is given by

 Where is the centroid of each symmetric membership function 𝑧

• Weighted average method: The weighted average method is the most frequently used in
fuzzy applications since it is one of the more computationally efficient methods.

EXAMPLE
• The two central values are 45 and 75 respectively
• The weights are, the membership values and are 0.7 and 1 respectively
• So,

(4) MEAN MAX MEMBERSHIP METHOD

• Alsocalled as middle of maxima
• Itis closely related to Maximum membership principle
• The difference being the location of the maximum membership can be a plateau rather than
a single point
• Introduced by Sugeno 1985and also Lee 1990

122
CONCEPTS AND TECHNOLOGIES

EXAMPLE-1

(MEAN MAX MEMBERSHIP METHOD)

• In the mean max membership method for defuzzification z* is given by (a + b)/2, where a
and b are the minimum and the maximumvalues wherethe maximummembership occurs
• Thismethod(also called middle-of-maxima)
• Is closely related to the first method (Max membership principle),
• Except that the locations of the maximum membership can be non-unique (i.e., the
maximum membership can be a plateau rather than a single point).
• This method is given by the expression (Sugeno, 1985; Lee, 1990)

(5)CENTRE OF SUMS METHOD

• This process involves the algebraic sum of individual output fuzzy sets, instead of their
union
• Advantages:
• This is faster than many defuzzification methods that are currently in use
• Thisis not restricted to symmetric membership functions
• Drawbacks:
Theintersecting areas are added twice
The method also involves finding the centroids of the individual membership functions

• The defuzzified value z* is given by

• Here 𝑧 is the distance to the centroid of each of the respective membership functions
• Thismethodis similar to the weighted average method
• Herethe weights are the areas of the respective membership functions
• In the weighted average method the weights are the individual membership values

123
CONCEPTS AND TECHNOLOGIES

EXAMPLE-2 AND CENTRE OF SUM METHOD

• There are 3 graphs
• Here we use the area of the trapezium as
• ½(sum of the two parallel sides)x(distance between the parallel sides)
• The three centroids are 2.5, 5 and 6.5 respectively
• So, the computations for the graphs in the numerator of the formula are:
• (2.5)x(0.5)x(0.3)x(3+5); (5)x(0.5)x(0.5)x(2+4) and
(6.5)x(0.5)x1x(3+1) respectively
• In the denominator we will have the same factors without the centroid values (2.5 in the
first, 5 in the second and 6.5 in the third)

(6) CENTER OF LARGEST AREA METHOD

• If the output fuzzy set has at least two convex sub regions then the centre of gravity of the
convex fuzzy sub region with the largest area is used to obtain the defuzzified value z* of
the output
• The formulais:

 Here 𝐶𝑚 is the convex sub region that has the largest area making up𝑪k
• This condition applies in the case when the overall output
is non-convex
• When is convex, z* is the same quantity as determined by the centroid method or the
centre of largest area method as then there will be only one convex region

124
CONCEPTS AND TECHNOLOGIES

CENTRE OF LARGEST AREA METHOD

• Asthe whole area is convex, according to the centre of largest area method, the output is
same as that of the centroid method.
• So,z*=2.5

(7)FIRST (OR LAST) OF MAXIMA

• This method uses the overall output or union of all individual output fuzzy sets 𝑪𝒌 to
determine the smallest value (or the largest value) of the domain with maximized membership
degree in 𝐶𝑘
• Theformulafor theoutput is determined as
• Thelargest height in the union (hgt(𝐶𝑘)) is determined

125
CONCEPTS AND TECHNOLOGIES

COMPUTATIONS WITH EXAMPLE-2

• The maximum value occurs in the last trapezium and it is a plateau
• So,thefirst max = 6
• Thelast max =7

SUMMARY
• Q. Of the seven defuzzification methods presented, which is the best?
• Ans.It is context or problem dependent
• Five Criteria of Hellendoorn and Thomas (1993):
• Continuity: A small change in the input of a fuzzy process should not produce a large
change in the output
• Disambiguity: The defuzzification method should always result in a unique value for z*;
i.e. there should not be any ambiguity in the defuzzified value
• Plausibility: z* should lie approximately in the middle of the support region of 𝐶𝑘 and have
a high degree of membership in 𝐶k

 Computational Simplicity: The more time consuming a method is, the less value it should
have in a computation system
• Weighting Method: Which weighs he output fuzzy sets. It compares the output values and
finds the difference. This criteria is not easy.
• NOTE: Other methods available are intended to seem as superior to the simple methods
presented here.

126
CONCEPTS AND TECHNOLOGIES

127