Soft Computing

EL
Introduction to Soft Computing

PT
Prof. Debasis Samanta

N
Department of Computer Science & Engineering
IIT Kharagpur

1
INTRODUCTION TO SOFT COMPUTING
• Concept of computation
• Hard computing

EL
• Soft computing

PT
• How soft computing?

N
• Hard computing vs. Soft computing
• Hybrid computing

Debasis Samanta
CSE
2
IIT Kharagpur
CONCEPT OF COMPUTATION

Antecedent Computing Consequent

𝑦 = 𝑓(𝑥)

EL
Control Action

PT
Figure: Basic of computing

N
𝑦 = 𝑓 𝑥 , 𝑓 is a mapping function.
𝑓 is also called a formal method or an algorithm to solve a problem.

Debasis Samanta
CSE
3
IIT Kharagpur
Important characteristics of computing

• Should provide precise solution.

• Control action should ne unambiguous and accurate.

EL
• Suitable for problem, which is easy to model mathematically.

PT
N
Debasis Samanta
CSE
4
IIT Kharagpur
Hard computing

• In 1996, L. A. Zade (LAZ) introduced the term hard computing.

• According to LAZ: We term a computing as Hard computing, if

EL
 Precise result is guaranteed.

PT
 Control action is unambiguous.

N
 Control action is formally defined (i.e., with mathematical model or
algorithm).

Debasis Samanta
CSE
5
IIT Kharagpur
Examples of hard computing
• Solving numerical problems (e.g., roots of polynomials,
integration, etc.).
• Searching and sorting techniques.

EL
• Solving computational geometry problems (e.g., shortest

PT
tour in a graph, finding closet pair of points given a set of

N
points, etc.).

• many more…
Debasis Samanta
CSE
6
IIT Kharagpur
Soft computing
• The term soft computing was proposed by the inventor of
fuzzy logic, Lotfi A. Zadeh. He describes it as follows.
Definition 1: Soft computing

EL
Soft computing is a collection of methodologies that aim to exploit the

PT
tolerance for imprecision and uncertainty to achieve tractability, robustness,
and low solution cost. Its principal constituents are fuzzy logic, neuro-

N
computing, and probabilistic reasoning. The role model for soft computing is
the human mind.

Debasis Samanta
CSE
7
IIT Kharagpur
Characteristics of soft computing
• It does not require any mathematical modeling of problem solving.
• It may not yield the precise solution.

EL
• Algorithms are adaptive (i.e., it can adjust to the change of dynamic
environment).

PT
• Use some biological inspired methodologies such as genetics, evolution,
Ant’s behaviors, particles swarming, human nervous system, etc.).

N
Debasis Samanta
CSE
8
IIT Kharagpur
Examples of soft computing

Soft

EL
computing
⋮

PT
N
Example: Hand written character recognition
(Artificial Neural Networks)

Debasis Samanta
CSE
9
IIT Kharagpur
Examples of soft computing

Soft Bank with

computing

EL
maximum return
⋮

PT
N
Example: Money allocation problem
(Evolutionary Computing)
Debasis Samanta
CSE
10
IIT Kharagpur
Examples of soft computing

EL
PT
NExample: Robot movement
(Fuzzy Logic)

Debasis Samanta
CSE
11
IIT Kharagpur
How soft computing?
• How a student learns from his teacher?
• Teacher asks questions and tell the answers then.
• Teacher puts questions and hints answers and asks whether the

EL
answers are correct or not.

PT
• Student thus learn a topic and store in his memory.

N
• Based on the knowledge he solves new problems.
• This is the way how human brain works.
• Based on this concept Artificial Neural Network is used to solve problems.
Debasis Samanta
CSE
12
IIT Kharagpur
How soft computing?
• How world selects the best?
• It starts with a population (random).
• Reproduces another population (next generation).

EL
• Rank the population and selects the superior individuals.

PT
• Genetic algorithm is based on this natural phenomena.

N
• Population is synonymous to solutions.
• Selection of superior solution is synonymous to exploring the optimal
solution.
Debasis Samanta
CSE
13
IIT Kharagpur
How soft computing?
• How a doctor treats his patient?
• Doctor asks the patient about suffering.
• Doctor find the symptoms of diseases.

EL
• Doctor prescribed tests and medicines.

PT
• This is exactly the way Fuzzy Logic works.

N
• Symptoms are correlated with diseases with uncertainty .
• Doctor prescribes tests/medicines fuzzily.

Debasis Samanta
CSE
14
IIT Kharagpur
Hard computing vs. Soft computing
Hard computing Soft computing
 It requires a precisely stated  It is tolerant of imprecision,
analytical model and often a lot of uncertainty, partial truth, and
computation time. approximation.

EL
 It is based on binary logic, crisp
 It is based on fuzzy logic, neural

PT
systems, numerical analysis and
nets and probabilistic reasoning.
crisp software.

N
 It has the characteristics of precision  It has the characteristics of
and categoricity. approximation and dispositionality.

Debasis Samanta
CSE
15
IIT Kharagpur
Hard computing vs. Soft computing

Hard computing Soft computing

 It is deterministic.  It incorporates stochasticity.

EL
 It can deal with ambiguous and
 It requires exact input data.
noisy data.

PT
 It is strictly sequential.  It allows parallel computations.
 It produces precise answers.  It can yield approximate answers

N
Debasis Samanta
CSE
16
IIT Kharagpur
Hybrid computing
• It is a combination of the conventional hard computing and
emerging soft computing.

EL
HC SC

PT
N
Hybrid computing

Figure: Concept of Hybrid Computing

Debasis Samanta
CSE
17
IIT Kharagpur
In this course…
• You will be able to learn
• Basic concepts of Fuzzy algebra and then how to solve problems using
Fuzzy logic.

EL
• The framework of Genetic algorithm and solving varieties of

PT
optimization problems.

N
• How to build an artificial neural network and train it with input data to
solve a number of problems, which are not possible to solve with hard
computing.

Debasis Samanta
CSE
18
IIT Kharagpur
 EL
PT
N
Debasis Samanta
CSE
19
IIT Kharagpur
 Soft Computing

EL
Introduction to Fuzzy Logic

PT
Prof. Debasis Samanta

N
Department of Computer Science and Engineering
IIT Kharagpur

1
What is Fuzzy logic?
• Fuzzy logic is a mathematical language to express something.
• This means it has grammar, syntax, semantic like a language for
communication.

EL
• There are some other mathematical languages also known
• Relational algebra (operations on sets)

PT
• Boolean algebra (operations on Boolean variables)
• Predicate algebra (operations on well formed formulae (wff), also called predicate

N
propositions)
• Fuzzy logic deals with Fuzzy set or Fuzzy algebra.

Debasis Samanta
CSE
IIT KHARAGPUR 2
What is fuzzy?

• Dictionary meaning of fuzzy is not clear, noisy, etc.

Example: Is the picture on this slide is fuzzy?

EL
• Antonym of fuzzy is crisp

PT
Example: Are the chips crisp?

N
Debasis Samanta
CSE
IIT KHARAGPUR 3
Example : Fuzzy logic vs. Crisp logic
Yes or No
Crisp answer
True or False

EL
· Milk Yes

PT
· Water
A liquid Crisp
· Coca

N
No
· Sprite

Is the liquid
colorless?

Debasis Samanta
CSE
IIT KHARAGPUR 4
Example : Fuzzy logic vs. Crisp logic

May be

May not be

EL
Fuzzy answer Absolutely

PT
Partially

N
etc.

Debasis Samanta
CSE
IIT KHARAGPUR 5
Example : Fuzzy logic vs. Crisp logic
Score

99
Extremely honest
· Ankit

· Rajesh

EL
Very honest 75
· Santosh
Fuzzy

PT
· Kabita Honest at times 55

· Salmon

N
Extremely dishonest 35
Is the person
honest?

Debasis Samanta
CSE
IIT KHARAGPUR 6
World is fuzzy!

EL
Our world better
described fuzzily

PT
N
Debasis Samanta
CSE
IIT KHARAGPUR 7
Concept of fuzzy system

Fuzzy element(s)

Fuzzy sets
I

EL
N
P Fuzzy rules
U

PT
T
Fuzzy implications
(Inferences)

N
O
U
T
Fuzzy system P
U
T

Debasis Samanta
CSE
IIT KHARAGPUR 8
Concept of fuzzy set
To understand the concept of fuzzy set it is better, if we first clear our idea of
crisp set.
X = The entire population of India.
H = All Hindu population = {ℎ1 , ℎ2 , ℎ3 , … … , ℎ𝐿 , }

EL
M = All Muslim population ={𝑚1 , 𝑚2 , 𝑚3 , … … , 𝑚𝑁 , }

PT
Universe of discourse X

N
H
Here, All are the sets of finite numbers of individuals.
M
Such a set is called crisp set.

Debasis Samanta
CSE
IIT KHARAGPUR 9
Example of fuzzy set
Let us discuss about fuzzy set.
X = All students in NPTEL.
S = All Good students.

EL
S = {(𝑠, 𝑔(𝑠)) | 𝑠 ∈ 𝑋 } and g(s) is a measurement of goodness of the
student s.

PT
N
Example:
S = 𝑅𝑎𝑗𝑎𝑡, 0.8 , 𝐾𝑎𝑏𝑖𝑡𝑎, 0.7 , 𝑆𝑎𝑙𝑚𝑎𝑛, 0.1 , 𝐴𝑛𝑘𝑖𝑡, 0.9 , etc.

Debasis Samanta
CSE
IIT KHARAGPUR 10
Fuzzy set vs. Crisp set

Crisp set Fuzzy set

 𝐹 = 𝑠, 𝜇(𝑠) |𝑠 ∈ 𝑋 and 𝜇 𝑠 is the
 𝑆 = 𝑠 𝑠 ∈ 𝑋}
degree of 𝑠.

EL
 It is a collection of elements.  It is a collection of ordered pairs.

PT
 Inclusion of an element s ∈ X into S  Inclusion of an element s ∈ X into F
is crisp, that is, has strict boundary is fuzzy, that is, if present, then with

N
yes or no. a degree of membership.

Debasis Samanta
CSE
IIT KHARAGPUR 11
Fuzzy set vs. Crisp set

Note: A crisp set is a fuzzy set, but, a fuzzy set is not necessarily a crisp set.

Example:

EL
H = (ℎ1 , 1 , (ℎ2 , 1) … … … , (ℎ𝐿 , 1)}

PT
Person = (𝑝1 , 0 , (𝑝2 , 0) … … … , (𝑝𝑁 , 0)}

N
In case of a crisp set, the elements are with extreme values of degree of
membership namely either 1 or 0.

Debasis Samanta
CSE
IIT KHARAGPUR 12
Degree of membership
How to decide the degree of memberships of elements in a fuzzy set?

City Bangalore Bombay Hyderabad Kharagpur Madras Delhi

EL
𝜇 0.95 0.90 0.80 0.01 0.65 0.75

PT
N
How the cities of comfort can be judged?

Debasis Samanta
CSE
IIT KHARAGPUR 13
Example: Course evaluation in a crisp way
𝐸𝑋 ∶ Marks ≥ 90
𝐴 ∶ 80 ≤ Marks < 90
𝐵 ∶ 70 ≤ Marks < 80

EL
𝐶 ∶ 60 ≤ Marks < 70

PT
𝐷 ∶ 50 ≤ Marks < 60

N
𝑃 ∶ 35 ≤ Marks < 50
𝐹 ∶ Marks ≤ 35

Debasis Samanta
CSE
IIT KHARAGPUR 14
Example: Course evaluation in a crisp way

F P D C B A EX
1

EL


PT
N
0
35 50 60 70 80 90 100

Debasis Samanta
CSE
IIT KHARAGPUR 15
Example: Course evaluation in a fuzzy way

F P D C B A EX
1

EL


PT
N
0
35 50 60 70 80 90 100

Debasis Samanta
CSE
IIT KHARAGPUR 16
Few examples of fuzzy set

• High Temperature
• Low Pressure
• Colour of Apple

EL
• Sweetness of Orange

PT
• Weight of Mango

N
Note: Degree of membership values lie in the range [0...1].

Debasis Samanta
CSE
IIT KHARAGPUR 17
Some basic terminologies and notations
Definition 1: Membership function (and Fuzzy set)
If 𝑋 is a universe of discourse and 𝑥 ∈ 𝑋, then a fuzzy set A in 𝑋 is
defined as a set of ordered pairs, that is
A = {(𝑥, 𝜇𝐴 𝑥 )|𝑥 ∈ 𝑋} where 𝜇𝐴 𝑥 is called the membership function

EL
for the fuzzy set A.

PT
Note: 𝜇𝐴 𝑥 map each element of X onto a membership grade (or membership
value) between 0 and 1 (both inclusive).

N
Question: How (and who) decides 𝜇𝐴 𝑥 for a fuzzy set A in X?

Debasis Samanta
CSE
IIT KHARAGPUR 18
Some basic terminologies and notations

Example:
X = All cities in India

EL
A = City of comfort
A={(New Delhi, 0.7), (Bangalore, 0.9), (Chennai, 0.8), (Hyderabad, 0.6),

PT
(Kolkata, 0.3), (Kharagpur, 0)}

N
Debasis Samanta
CSE
IIT KHARAGPUR 19
Membership function with discrete membership
values

The membership
values may be of

EL
discrete values.

µA(x)

PT
N x
A fuzzy set with discrete values of 
Debasis Samanta
CSE
IIT KHARAGPUR 20
Membership function with discrete membership
values
Either elements or their membership values (or both) also may be of discrete values.

A ={(0,0.1),(1,0.30),(2,0.78)……(10,0.1)}

EL
1.0

***
0.8

PT
0.6 Note : X = discrete value
µ

0.4 *
0.2 * *** *
*0 * How you measure happiness ??

N
2 4 6 8 10

Number of children (X)

A = “Happy family”

Debasis Samanta
CSE
IIT KHARAGPUR 21
Membership function with continuous membership
values

1.0
 B ( x)  1

EL
4
0.8  x  50 
1  
0.6  10 
µ

0.4

PT
0.2

0 50 100
B={(x,µB(x))}

N
Age (X)
Note : x = real value
B = “Middle aged” = R+

Debasis Samanta
CSE
IIT KHARAGPUR 22
Fuzzy terminologies: Support
Support: The support of a fuzzy set A is the set of all points 𝑥 ∈ 𝑋 such that 𝜇𝐴 (𝑋) > 0

Support (A)={x | µA(x) > 0}

EL
PT
µ

N
support
x

Debasis Samanta
CSE
IIT KHARAGPUR 23
Fuzzy terminologies: Core
Core: The core of a fuzzy set A is the set of all points 𝑥 in 𝑋 such that 𝜇𝐴 𝑋 = 1

core (A) = {x | µA(x) = 1}

EL
PT
1.0
µ

N
0.5

Debasis Samanta
CSE
IIT KHARAGPUR 24
Fuzzy terminologies: Normality
Normality : A fuzzy set A is a normal if its core is non-empty. In other words, we can
always find a point 𝑥 ∈ 𝑋 such that 𝜇𝐴 (𝑋) = 1

Normality (A) = FALSE

EL
1.0

PT
µ

N x

Debasis Samanta
CSE
IIT KHARAGPUR 25
Fuzzy terminologies: Crossover points
Crossover point : A crossover point of a fuzzy set A is a point 𝑥 ∈ 𝑋 at which 𝜇𝐴 (𝑋) =
0.5. That is Crossover (A) ={𝑥|𝜇𝐴 𝑥 = 0.5}

EL
1.0

PT
µ

0.5

N Crossover point
x

Debasis Samanta
CSE
IIT KHARAGPUR 26
Fuzzy terminologies: Fuzzy Singleton
Fuzzy Singleton : A fuzzy set whose support is a single point in 𝑋 with 𝜇𝐴 (𝑥) = 1 is
called a fuzzy singleton. That is |𝐴| = {𝑥|𝜇𝐴 𝑥 = 1}
singleton

EL
1.0
*

PT
* *
µ

0.5
*
* * * * *

core N x
Debasis Samanta
CSE
IIT KHARAGPUR 27
Fuzzy terminologies: 𝜶-cut and strong 𝜶-cut

𝜶-cut and strong 𝜶-cut :

 The 𝛼-cut of a fuzzy set A is a crisp set defined by
𝐴𝛼 = {𝑥|𝜇𝐴 𝑥 ≥ 𝛼}

EL
 Strong 𝛼-cut is defined similarly :

PT
𝐴′𝛼 = {𝑥|𝜇𝐴 𝑥 > 𝛼}

N
Note : Support (A) = 𝐴0’ and Core (A) = 𝐴1.

Debasis Samanta
CSE
IIT KHARAGPUR 28
Fuzzy terminologies: Bandwidth

Bandwidth :
For a fuzzy set, the bandwidth (or width) is defined as the distance between the two
unique crossover points:

EL
Bandwidth 𝐴 = |𝑥1 − 𝑥2 |

PT
where 𝜇𝐴 𝑥1 = 𝜇𝐴 𝑥2 = 0.5

N
Debasis Samanta
CSE
IIT KHARAGPUR 29
Fuzzy terminologies: Symmetry

Symmetry :
A fuzzy set A is symmetric if its membership function around a certain point 𝑥 = 𝑐,
namely 𝜇𝐴 𝑥 + 𝑐 = 𝜇𝐴 𝑥 − 𝑐 for all 𝑥 ∈ 𝑋

EL
1.0

µ T
P
N x=c
x
Debasis Samanta
CSE
IIT KHARAGPUR 30
Fuzzy terminologies: Open and Closed
A fuzzy set A is
Open left : If lim 𝑥 → −∞ 𝜇𝐴 𝑥 = 1and lim 𝑥 → +∞ 𝜇𝐴 𝑥 = 0
Open right: If lim 𝑥 → −∞ 𝜇𝐴 𝑥 = 0and lim 𝑥 → +∞ 𝜇𝐴 𝑥 = 1
Closed: If lim 𝑥 → −∞ 𝜇𝐴 𝑥 = lim 𝑥 → +∞ 𝜇𝐴 𝑥 = 0

EL
PT
Open left Open right
Closed

N
Debasis Samanta
CSE
IIT KHARAGPUR 31
Fuzzy vs. Probability
Fuzzy : When we say about certainty of a thing
Example: A patient come to the doctor and he has to diagnose so that
medicine can be prescribed.
Doctor prescribed a medicine with certainty 60% that the

EL
patient is suffering from flue. So, the disease will be cured with
certainty of 60% and uncertainty 40%. Here, in stead of flue,

PT
other diseases with some other certainties may be.

N
Probability: When we say about the chance of an event to occur
Example: India will win the T20 tournament with a chance 60% means
that out of 100 matches, India own 60 matches.

Debasis Samanta
CSE
IIT KHARAGPUR 32
Prediction vs. Forecasting
The Fuzzy vs. Probability is analogical to Prediction vs. Forecasting
Prediction : When you start guessing about things.
Forecasting : When you take the information from the past job and apply it to new job.

EL
PT
The main difference:

N
Prediction is based on the best guess from experiences.
Forecasting is based on data you have actually recorded and packed from previous job.

Debasis Samanta
CSE
IIT KHARAGPUR 33
 N
PT
34 EL
Introduction to Soft Computing

EL
Fuzzy membership functions

PT
Rof. Debasis Samanta

N
Department of Computer Science and Engineering
IIT Kharagpur

1
Fuzzy membership functions
A fuzzy set is completely characterized by its membership function (sometimes
abbreviated as 𝑀𝐹 and denoted as 𝜇 ). So, it would be important to learn how a
membership function can be expressed (mathematically or otherwise).
Example:

EL
Note: A membership function can
be on

PT
1.0 1.0

a) a discrete universe of 0.8

***
0.8

µA

µB
0.6 0.6

discourse and 0.4 * 0.4

* ****

N
0.2 0.2
*0 2 4 6 8
*
10 0 10 20 30 40 50 60

b) a continuous universe of
Number of children (X) Age (X)
discourse.
A = Fuzzy set of “Happy family” B = “Young age”

Debasis Samanta
CSE
2
IIT Kharagpur
Fuzzy membership functions
So, membership function on a discrete universe of course is trivial. However, a
membership function on a continuous universe of discourse needs a special attention.

EL

µ
µ

µ
Following figures shows x x x

PT
typical examples of < triangular > < trapezoidal > < curve >

membership functions.

N µ

µ
x x
< non-uniform > < non-uniform >

Debasis Samanta
CSE
3
IIT Kharagpur
Fuzzy MFs : Formulation and parameterization
In the following, we try to parameterize the different MFs on a continuous universe of
discourse.
Triangular MFs : A triangular MF is specified by three parameters {𝑎, 𝑏, 𝑐} and can be
formulated as follows.

EL
0 𝑖𝑓 𝑥 ≤ 𝑎
𝑥−𝑎

PT
1.0 𝑖𝑓 𝑎 ≤ 𝑥 ≤ 𝑏
𝑎 − 𝑏
triangle 𝑥; 𝑎, 𝑏, 𝑐 = 𝑐 − 𝑥
𝑖𝑓 𝑏 ≤ 𝑥 ≤ 𝑐

N
𝑐−𝑏
0 𝑖𝑓 𝑐 ≤ 𝑥
a b c

Debasis Samanta
CSE
4
IIT Kharagpur
Fuzzy MFs: Trapezoidal
A trapezoidal MF is specified by four parameters {𝑎, 𝑏, 𝑐, 𝑑} and can be defined as
follows:
0 𝑖𝑓 𝑥 ≤ 𝑎
𝑥−𝑎
𝑖𝑓 𝑎 ≤ 𝑥 ≤ 𝑏

EL
𝑏−𝑎
trapozoid 𝑥; 𝑎, 𝑏, 𝑐, 𝑑 = 1 𝑖𝑓 𝑏 ≤ 𝑥 ≤ 𝑐

PT
1.0 𝑑−𝑥
𝑖𝑓 𝑐 ≤ 𝑥 ≤ 𝑑
𝑑−𝑐

N
0 𝑖𝑓 𝑑 ≤ 𝑥

a b c d

Debasis Samanta
CSE
5
IIT Kharagpur
Fuzzy MFs: Gaussian
A Gaussian MF is specified by two parameters 𝑐, 𝜎 and can be defined as below:
1 𝑥−𝑐 2
−2 𝜎
𝑔𝑎𝑢𝑠𝑠𝑖𝑎𝑛 𝑥; 𝑐, 𝜎 = 𝑒

EL
c
0.1
0.1c 0.9c

PT
N


Debasis Samanta
CSE
6
IIT Kharagpur
Fuzzy MFs: Generalized bell
It is also called Cauchy MF. A generalized bell MF is specified by three parameters
{𝑎, 𝑏, 𝑐} and is defined as:
1
𝑏𝑒𝑙𝑙(𝑥; 𝑎, 𝑏, 𝑐) =
𝑥 − 𝑐 2𝑏
1+ 𝑎

EL
PT
b
Slope at x =
2a

N
b
Slope at y = 
x b y
2a

c-a c c+a

Debasis Samanta
CSE
7
IIT Kharagpur
Example: Generalized bell MFs

1
Example: 𝜇 𝑥 = 1+ 𝑥 2;

a = b = 1 and c = 0;

EL
1.0

PT
1

N -1 0 1

Debasis Samanta
CSE
8
IIT Kharagpur
Generalized bell MFs: Different shapes

EL
Changing a Changing b

PT
Changing a
N Changing a and b

Debasis Samanta
CSE
9
IIT Kharagpur
Fuzzy MFs: Sigmoidal MFs
Parameters: {𝑎, 𝑐} ; where 𝑐 = crossover point and 𝑎 = slope at 𝑐;

1.0

EL
1
𝑆𝑖𝑔𝑚𝑜𝑖𝑑(𝑥; 𝑎, 𝑐) = 𝑥−𝑐
− 𝑎

PT
Slope = a 1+ 𝑒
0.5

c N
Debasis Samanta
CSE
10
IIT Kharagpur
Fuzzy MFs : Example
Example : Consider the following grading system for a course.
Excellent = Marks ≤ 90
Very good = 75 ≤ Marks ≤ 90

EL
Good = 60 ≤ Marks ≤ 75

PT
Average = 50 ≤ Marks ≤ 60

N
Poor = 35 ≤ Marks ≤ 50
Bad= Marks ≤ 35

Debasis Samanta
CSE
11
IIT Kharagpur
Grading System
A fuzzy implementation will look like the following.
Bad poor Average Good Very Good Excellent

1
.8

EL
.6

PT
.4
.2
0

N
10 20 30 40 50 60 70 80 90
marks

You can decide a standard fuzzy MF for each of the fuzzy grade.
Debasis Samanta
CSE
12
IIT Kharagpur
Few More on Membership Functions

EL
PT
N
Debasis Samanta
CSE
13
IIT Kharagpur
Generation of MFs
Given a membership function of a fuzzy set representing a linguistic hedge, we can derive
many more MFs representing several other linguistic hedges using the concept of
Concentration and Dilation.
1. Concentration: 𝐴𝑘 = [𝜇𝐴 𝑥 ]𝑘 ; 𝑘 > 1

EL
2. Dilation: 𝐴𝑘 = [𝜇𝐴 𝑥 ]𝑘 ; 𝑘 < 1
Example : Age = { Young, Middle-aged, Old }

PT
Thus, corresponding to Young, we have : Not young, Very young, Not very young and so on.
Similarly, with Old we can have : Not old, Very old, Very very old, Extremely old, etc.

N
Thus, 𝜇𝐸𝑥𝑡𝑟𝑒𝑚𝑒𝑙𝑦 𝑜𝑙𝑑 𝑥 = (((𝜇𝑂𝑙𝑑 𝑥 )2 )2 )2 and so on
Or, 𝜇𝑀𝑜𝑟𝑒 𝑜𝑟 𝑙𝑒𝑠𝑠 𝑜𝑙𝑑 𝑥 = 𝐴0.5 = (𝜇𝑂𝑙𝑑 𝑥 )0.5

Debasis Samanta
CSE
14
IIT Kharagpur
Linguistic variables and values

1
𝜇𝑦𝑜𝑢𝑛𝑔 𝑥 =bell(x,20,2,0) = 𝑥
1+( )4
20
Young Old
Middle-Aged

EL
1
𝜇𝑜𝑙𝑑 𝑥 =bell(x,30,3,100) = 𝑥−100
1+( 30 )6 Very Old

PT
𝜇𝑚𝑖𝑑𝑑𝑙𝑒−𝑎𝑔𝑒𝑑 𝑥 =bell(x,30,60,50)

N
Not young=𝜇𝑦𝑜𝑢𝑛𝑔 𝑥 = 1 − 𝜇𝑦𝑜𝑢𝑛𝑔 𝑥 0 30 60 90

Young but not too young = 𝜇𝑦𝑜𝑢𝑛𝑔 𝑥 ∩ 𝜇𝑦𝑜𝑢𝑛𝑔 𝑥

Debasis Samanta
CSE
15
IIT Kharagpur
 EL
PT
N
Debasis Samanta
CSE
16
IIT Kharagpur
Introduction to Soft Computing

EL
Operations on Fuzzy sets

PT
Prof. Debasis Samanta

N
Department of Computer Science and Engineering
IIT Kharagpur

1
Basic fuzzy set operations: Union
Union (A ∪ B): 𝜇𝐴∪𝐵 𝑥 = max(𝜇𝐴 𝑥 , 𝜇𝐵 𝑥 )

Example:
𝐴 = {(𝑥1, 0.5), (𝑥2, 0.1), (𝑥3, 0.4)} and

EL
𝐵 = { 𝑥1, 0.2 , 𝑥2, 0.3 , 𝑥3, 0.5 };
𝐶 = 𝐴 ∪ 𝐵 = { 𝑥1, 0.5 , 𝑥2, 0.3 , 𝑥3, 0.5 }

PT
µA µA
µB µB

N
µ

µAUB
a p x b q c a p x b q c

Debasis Samanta
CSE
2
IIT Kharagpur
Basic fuzzy set operations: Intersection
Intersection (A ∩ B): 𝜇𝐴∩𝐵 𝑥 = min(𝜇𝐴 𝑥 , 𝜇𝐵 𝑥 )

Example:
𝐴 = {(𝑥1, 0.5), (𝑥2, 0.1), (𝑥3, 0.4)} and

EL
𝐵 = { 𝑥1, 0.2 , 𝑥2, 0.3 , 𝑥3, 0.5 };
𝐶 = 𝐴 ∩ 𝐵 = { 𝑥1, 0.2 , 𝑥2, 0.1 , 𝑥3, 0.4 }

PT
µA
µB

N
µA?B
µ

a p x b q c a p x b q c

Debasis Samanta
CSE
3
IIT Kharagpur
Basic fuzzy set operations: Complement
Complement (𝐴𝑐 ): 𝜇𝐴𝑐 𝑥 = 1 − 𝜇𝐴 𝑥

Example:
𝐴 = {(𝑥1, 0.5), (𝑥2, 0.1), (𝑥3, 0.4)}
𝐶 = 𝐴𝑐 = { 𝑥1, 0.5 , 𝑥2, 0.9 , 𝑥3, 0.6 }

EL
µA µA

PT
1.0 µA’

p x N q p x q

Debasis Samanta
CSE
4
IIT Kharagpur
Basic fuzzy set operations: Products

Algebric product or Vector product (𝑨 ∙ 𝑩):

𝜇𝐴∙𝐵 𝑥 = 𝜇𝐴 𝑥 ∙ 𝜇𝐵 𝑥

EL
Scalar product (𝜶 × 𝑨):

PT
𝜇𝛼𝐴 𝑥 = 𝛼 × 𝜇𝐴 𝑥

N
Debasis Samanta
CSE
5
IIT Kharagpur
Basic fuzzy set operations: Sum and Difference
Sum (A + B):
𝜇𝐴+𝐵 𝑥 = 𝜇𝐴 𝑥 + 𝜇𝐵 𝑥 − 𝜇𝐴 𝑥 . 𝜇𝐵 𝑥
Difference (𝐴 − 𝐵 = 𝐴 ∩ 𝐵𝐶 ) :
𝜇𝐴−𝐵 𝑥 = 𝜇𝐴∩𝐵𝐶 𝑥

EL
Disjunctive sum:
𝐴⨁𝐵 = (𝐴𝐶 ∩ 𝐵) ∪ (𝐴 ∩ 𝐵𝐶 )

PT
Bounded Sum:

N
| 𝐴 𝑥 ⨁𝐵 𝑥 |= 𝜇|𝐴(𝑥)⨁𝐵(𝑥)| = min{1, 𝜇𝐴 𝑥 + 𝜇𝐵 𝑥 }
Bounded Difference:
𝐴 𝑥 ⊖𝐵 𝑥 = 𝜇|𝐴(𝑥)⊖𝐵(𝑥)| = max{0, 𝜇𝐴 𝑥 + 𝜇𝐵 𝑥 − 1}

Debasis Samanta
CSE
6
IIT Kharagpur
Basic fuzzy set operations: Equality and Power
Equality (𝐴 = 𝐵):
𝜇𝐴 𝑥 = 𝜇𝐵 𝑥
Power of a fuzzy set 𝐴𝛼 :

EL
𝜇𝐴𝛼 𝑥 = (𝜇𝐴 𝑥 )𝛼

PT
 If α < 1, then it is called dilation

N
 If α > 1, then it is called concentration

Debasis Samanta
CSE
7
IIT Kharagpur
Basic fuzzy set operations: Cartesian product
Caretsian Product (𝐴 × 𝐵): 𝜇𝐴×𝐵 𝑥, 𝑦 = min(𝜇𝐴 𝑥 , 𝜇𝐵 𝑦 )

Example:
𝐴 𝑥 = {(𝑥1 , 0.2), (𝑥2 , 0.3), (𝑥3 , 0.5), (𝑥4 , 0.6)}

EL
𝐵 𝑦 = { 𝑦1 , 0.8 , 𝑦2 , 0.6 , 𝑦3 , 0.3 }

PT
N
A × 𝐵 = min 𝜇𝐴 𝑥 , 𝜇𝐵 𝑦 =

Debasis Samanta
CSE
8
IIT Kharagpur
Properties of fuzzy sets
Commutativity :

𝐴∩𝐵 = 𝐵∩𝐴
𝐴∪B = 𝐵∪A
Associativity :

EL
𝐴 ∪ (𝐵 ∪ 𝐶) = (𝐴 ∪ 𝐵) ∪ 𝐶

PT
𝐴 ∩ (𝐵 ∩ 𝐶) = (𝐴 ∩ 𝐵) ∩ 𝐶
Distributivity :

N
𝐴 ∪ (𝐵 ∩ 𝐶) = (𝐴 ∪ 𝐵) ∩ (𝐴 ∪ 𝐶)
𝐴 ∩ (𝐵 ∪ 𝐶) = (𝐴 ∩ 𝐵) ∪ (𝐴 ∩ 𝐶)

Debasis Samanta
CSE
9
IIT Kharagpur
Properties of fuzzy sets
Idempotence :
𝐴 ∪ 𝐴 = 𝐴
𝐴 ∩ 𝐴 =∅;
𝐴 ∪ ∅; = 𝐴
𝐴 ∩∅; = ∅;

EL
Transitivity :
If 𝐴 ⊆ 𝐵; 𝐵 ⊆ 𝐶 then 𝐴 ⊆ 𝐶

PT
Involution :
(𝐴𝑐 )𝑐 = 𝐴

N
De Morgan’s law :
(𝐴 ∩ 𝐵)𝑐 = 𝐴𝑐 ∪ 𝐵𝑐
(𝐴 ∪ 𝐵)𝑐 = 𝐴𝑐 ∩ 𝐵𝑐

Debasis Samanta
CSE
10
IIT Kharagpur
Example 1: Fuzzy Set Operations
Let A and B are two fuzzy sets defined over a universe of discourse X with
membership functions 𝜇𝐴 𝑥 and 𝜇𝐵 𝑥 , respectively. Two MFs 𝜇𝐴 𝑥 and
𝜇𝐵 𝑥 are shown graphically.

EL
PT
µA(x)

µB(x)
a1
x
a2 a3
N
a4 b1 a1=b2

x
a2=b3 a4

Debasis Samanta
CSE
11
IIT Kharagpur
Example 1: Plotting two sets on the same graph
Let’s plot the two membership functions on the same graph

EL
µB µA

PT
µ

b1
N
x
a1 a2 b4 a3 a4

Debasis Samanta
CSE
12
IIT Kharagpur
Example 1: Union and Intersection
The plots of union 𝐴 ∪ 𝐵 and intersection 𝐴 ∩ 𝐵 are shown in the following.

µB µA

EL
µ
b1 a1 a2 b4 a3 a4

PT
x

N
 A B ( x)

 A B ( x)
a2 b4
x b1 a1 a2 a3 a4
x

Debasis Samanta
CSE
13
IIT Kharagpur
Example 1: Complementation
The plots of union 𝜇𝐴 𝑥 of the fuzzy set A is shown in the following.

EL
PT
N
Debasis Samanta
CSE
14
IIT Kharagpur
Fuzzy set operations: Practice
Consider the following two fuzzy sets A and B defined over a universe of
discourse [0,5] of real numbers with their membership functions
𝑥
𝜇𝐴 𝑥 = and 𝜇𝐵 𝑥 = 2−𝑥
1+𝑥

EL
Determine the membership functions of the following and draw them

PT
graphically.
I. 𝐴 , 𝐵
II. 𝐴 ∪ 𝐵

N
III. 𝐴 ∩ 𝐵
IV. (𝐴 ∪ 𝐵)𝑐
[Hint: Use De’ Morgan law]

Debasis Samanta
CSE
15
IIT Kharagpur
Example 2: A real-life example
Two fuzzy sets A and B with membership functions 𝜇𝐴 𝑥 and 𝜇𝐵 𝑥 , respectively
defined as below.
A = Cold climate with 𝜇𝐴 𝑥 as the MF.
B = Hot climate with 𝜇𝐵 𝑥 as the M.F.

EL
µA µB
1.0

PT
Here, 𝑋 being the
universe of discourse 0.5

N
representing entire µ
range of temperatures.
-15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

Debasis Samanta
CSE
16
IIT Kharagpur
Example 2: A real-life example
What are the fuzzy sets representing the following?
1. Not cold climate
2. Not hot climate

EL
3. Extreme climate

PT
4. Pleasant climate

N
Note: Note that ”Not cold climate” ≠ ”Hot climate” and vice-versa.

Debasis Samanta
CSE
17
IIT Kharagpur
Example 2: A real-life example
Answer would be the following.
 Not cold climate
𝐴 with 1 − 𝜇𝐴 𝑥 as the MF.
 Not hot climate

EL
𝐵 with 1 − 𝜇𝐵 𝑥 as the MF.

PT
 Extreme climate

N
A ∪ 𝐵 with 𝜇𝐴∪𝐵 𝑥 = max(𝜇𝐴 𝑥 , 𝜇𝐵 𝑥 ) as the MF.
 Pleasant climate
A ∩ 𝐵 with 𝜇𝐴∩𝐵 𝑥 = min(𝜇𝐴 𝑥 , 𝜇𝐵 𝑥 ) as the MF.

Debasis Samanta
CSE
18
IIT Kharagpur
Example 2: A real-life example
The plot of the MFs of A ∪ 𝐵 and A ∩ 𝐵 are shown in the following.

EL
PT
N
Debasis Samanta
CSE
19
IIT Kharagpur
 EL
PT
N
Debasis Samanta
CSE
20
IIT Kharagpur
 Soft Computing

EL
Fuzzy Relations

PT
Prof. Debasis Samanta

N
Department of Computer Science & Engineering
IIT Kharagpur

1
Fuzzy Relations
• Crisp relations
• Operations on crisp relations

EL
• Examples on crisp relations

PT
• Fuzzy relations

N
• Operations on fuzzy relations
• Examples on fuzzy relations

Debasis Samanta
CSE
2
IIT Kharagpur
Crisp relations
• Order pairs:
Suppose, A and B are two (crisp) sets. Then Cartesian product denoted as
𝐴 × 𝐵 is a collection of order pairs, such that

EL
𝐴 × 𝐵 = {(𝑎, 𝑏)|𝑎 ∈ 𝐴 𝑎𝑛𝑑 𝑏 ∈ 𝐵}
Note :

PT
(1) 𝐴 × 𝐵 ≠ 𝐵 × 𝐴 (2) |𝐴 × 𝐵| = |𝐴| × |𝐵|

N
(3) 𝐴 × 𝐵 provides a mapping from 𝑎 ∈ 𝐴 to 𝑏 ∈ 𝐵.

A particular mapping so mentioned is called a relation.

Debasis Samanta
CSE
3
IIT Kharagpur
Crisp relations
Example:
Consider the two crisp sets 𝐴 and 𝐵 as given below.
𝐴 = {1, 2, 3, 4} 𝐵 = {3, 5, 7}.

EL
Then, 𝐴 × 𝐵 = { 1, 3 , 1, 5 , 1, 7 , 2, 3 , 2, 5 , 2, 7 , 3, 3 , (3, 5), 3, 7 ,

PT
4, 3 , 4, 5 , (4, 7)}

N
Let us define a relation as 𝑅 = {(𝑎, 𝑏)|𝑏 = 𝑎 + 1, (𝑎, 𝑏) ∈ 𝐴 × 𝐵}
Then, 𝑅 = { 2, 3 , 4, 5 } in this case.

Debasis Samanta
CSE
4
IIT Kharagpur
Crisp relations
We can represent the relation 𝑅 in a matrix form as follows.

EL
PT
N
Debasis Samanta
CSE
5
IIT Kharagpur
Operations on crisp relations
Suppose, 𝑅(𝑥, 𝑦) and 𝑆(𝑥, 𝑦) are the two relations defined over two crisp sets
𝑥 ∈ 𝐴 and 𝑦 ∈ 𝐵

• Union: 𝑅 𝑥, 𝑦 ∪ 𝑆(𝑥, 𝑦) = 𝑚𝑎𝑥(𝑅 𝑥, 𝑦 , 𝑆(𝑥, 𝑦));

EL
• Intersection: 𝑅 𝑥, 𝑦 ∩ 𝑆(𝑥, 𝑦) = 𝑚𝑖𝑛(𝑅 𝑥, 𝑦 , 𝑆(𝑥, 𝑦));

PT
• Complement: 𝑅 𝑥, 𝑦 = 1 − 𝑅(𝑥, 𝑦)

N
Debasis Samanta
CSE
6
IIT Kharagpur
Example: Operations on crisp relations
Suppose, 𝑅(𝑥, 𝑦) and 𝑆(𝑥, 𝑦) are the two relations defined over two crisp sets
𝑥 ∈ 𝐴 and 𝑦 ∈ 𝐵
0 1 0 0 1 0 0 0
0 0 1 0 0 1 0 0
𝑅= and 𝑆 =

EL
0 0 0 1 0 0 1 0
0 0 0 0 0 0 0 1

PT
Find the following
• 𝑅∪𝑆

N
• 𝑅∩𝑆
• 𝑅

Debasis Samanta
CSE
7
IIT Kharagpur
Composition of two crisp relations
Given 𝑅 is a relation on 𝑋, 𝑌 and 𝑆 is another relation on 𝑌, 𝑍. Then, 𝑅 ∘ 𝑆 is
called a composition of relation on 𝑋 and 𝑍 which is defined as follows.
𝑅 ∘ 𝑆 = { 𝑥, 𝑧 | 𝑥, 𝑦 ∈ 𝑅 𝑎𝑛𝑑 𝑦, 𝑧 ∈ 𝑆 𝑎𝑛𝑑 ∀𝑦 ∈ 𝑌}

EL
Max-Min Composition

PT
Given the two relation matrices 𝑅 and 𝑆, the max-min composition is defined as
𝑇 = 𝑅∘𝑆;

N
𝑇(𝑥, 𝑧) = 𝑚𝑎𝑥{𝑚𝑖𝑛{𝑅 𝑥, 𝑦 , 𝑆(𝑦, 𝑧) 𝑎𝑛𝑑 ∀𝑦 ∈ 𝑌}}

Debasis Samanta
CSE
8
IIT Kharagpur
Composition: Composition
Example : Given 𝑋 = 1, 3, 5 ; 𝑌 = 1, 3, 5 ; 𝑅 = 𝑥, 𝑦 𝑦 = 𝑥 + 2 ;
𝑆 = {(𝑥, 𝑦)|𝑥 < 𝑦}
Here, 𝑅 and 𝑆 is on 𝑋 × 𝑌.
Thus, we have 𝑅 = 1, 3 , 3, 5 , 𝑆 = { 1, 3 , 1, 5 , (3, 5)}

EL
1 0 1 0 1 0 1 1

PT
𝑅=3 0 0 1 𝑎𝑛𝑑 𝑆= 3 0 0 1
5 0 0 0 5 0 0 0

Using max-min composition N 1 0

𝑅∘ 𝑆=3 0
5 0
0 1
0 0
0 0
Debasis Samanta
CSE
9
IIT Kharagpur
Fuzzy relations
• Fuzzy relation is a fuzzy set defined on the Cartesian product of crisp set
𝑋1, 𝑋2, … , 𝑋𝑛
• Here, n-tuples (𝑥1, 𝑥2, … , 𝑥𝑛) may have varying degree of memberships

EL
within the relationship.

PT
• The membership values indicate the strength of the relation between the
tuples.

N
Debasis Samanta
CSE
10
IIT Kharagpur
Fuzzy relations
Example:
𝑋 = 𝑡𝑦𝑝ℎ𝑜𝑖𝑑, 𝑣𝑖𝑟𝑎𝑙, 𝑐𝑜𝑙𝑑 , 𝑌 = { 𝑟𝑢𝑛𝑛𝑖𝑛𝑔 𝑛𝑜𝑠𝑒, ℎ𝑖𝑔ℎ 𝑡𝑒𝑚𝑝, 𝑠ℎ𝑖𝑣𝑒𝑟𝑖𝑛𝑔 }
The fuzzy relation 𝑅 is defined as

EL
PT
N
𝑡𝑦𝑝ℎ𝑜𝑖𝑑 0.1 0.9 0.8
𝑅 = 𝑣𝑖𝑟𝑎𝑙 0.2 0.9 0.7
𝑐𝑜𝑙𝑑 0.9 0.4 0.6

Debasis Samanta
CSE
11
IIT Kharagpur
Fuzzy Cartesian product
Suppose
• 𝐴 is a fuzzy set on the universe of discourse 𝑋 with 𝜇𝐴(𝑥)|𝑥 ∈ 𝑋
• 𝐵 is a fuzzy set on the universe of discourse 𝑌 with 𝜇𝐵(𝑦)|𝑦 ∈ 𝑌

EL
PT
Then 𝑅 = 𝐴 × 𝐵 ⊂ 𝑋 × 𝑌 ; where 𝑅 has its membership function given
by 𝜇𝑅 𝑥, 𝑦 = 𝜇𝐴x𝐵 (𝑥, 𝑦) = 𝑚𝑖𝑛{𝜇𝐴 𝑥 , 𝜇𝐵 𝑦 }

N
Debasis Samanta
CSE
12
IIT Kharagpur
Fuzzy Cartesian product
Example :
𝐴 = {(𝑎1, 0.2), (𝑎2, 0.7), (𝑎3, 0.4)} and 𝐵 = {(𝑏1, 0.5), (𝑏2, 0.6)}

EL
PT
𝑎1 0.2 0.2
𝑅 = 𝐴 × 𝐵 = 𝑎2 0.5 0.6

N
𝑎3 0.4 0.4

Debasis Samanta
CSE
13
IIT Kharagpur
Operations on Fuzzy relations
Let 𝑅 and 𝑆 be two fuzzy relations on 𝐴 × 𝐵.
• Union: 𝜇𝑅U𝑆(𝑎, 𝑏) = 𝑚𝑎𝑥{𝜇𝑅 𝑎, 𝑏 , 𝜇𝑆(𝑎, 𝑏)}
• Intersection: 𝜇𝑅∩𝑆 (𝑎, 𝑏) = 𝑚𝑖𝑛{𝜇𝑅 𝑎, 𝑏 , 𝜇𝑆(𝑎, 𝑏)}

EL
• Complement: 𝜇𝑅 𝑎, 𝑏 = 1 − 𝜇𝑅 (𝑎, 𝑏)

PT
• Composition: 𝑇 = 𝑅∘𝑆

N
𝜇𝑅∘𝑆 = 𝑚𝑎𝑥𝑦∈𝑌 {𝑚𝑖𝑛 𝜇𝑅 𝑥, 𝑦 , 𝜇𝑆 𝑦, 𝑧 }

Debasis Samanta
CSE
14
IIT Kharagpur
Operations on Fuzzy relations: Example
Example : 𝑋 = (𝑥1 , 𝑥2 , 𝑥3 ), 𝑌 = 𝑦1 , 𝑦2 , 𝑍 = 𝑧1 , 𝑧2 , 𝑧3 ,

𝑥1 0.5 0.1 𝑦1 0.6 0.4 0.7

𝑅 = 𝑥2 0.2 0.9 𝑎𝑛𝑑 𝑆= 𝑦
2 0.5 0.8 0.9
𝑥3 0.8

EL
0.6

PT
𝑥1 0.5 0.4 0.5
𝑅 ∘ 𝑆 = 𝑥2 0.5 0.8 0.9
𝑥3 0.6

N
0.6 0.7

𝜇𝑅∘𝑆 𝑥1, 𝑦1 = 𝑚𝑎𝑥 𝑚𝑖𝑛 𝜇𝑅 𝑥1, 𝑦1 , 𝜇𝑆 𝑦1, 𝑧1 ) , 𝑚𝑖𝑛(𝜇𝑅 𝑥1, 𝑦2 , 𝜇𝑆 𝑦2, 𝑧1

= 𝑚𝑎𝑥 𝑚𝑖𝑛 0.5, 0.6 , 𝑚𝑖𝑛 0.1, 0.5 = 𝑚𝑎𝑥 0.5, 0.1 = 0.5 𝑎𝑛𝑑 𝑠𝑜 𝑜𝑛.

Debasis Samanta
CSE
15
IIT Kharagpur
Fuzzy relation : An example
Consider the following two sets 𝑃 and 𝐷, which represent a set of paddy plants and a set
of plant diseases. More precisely
𝑃 = {𝑃1 , 𝑃2 , 𝑃3 , 𝑃4 } a set of four varieties of paddy plants
𝐷 = {𝐷1 , 𝐷2 , 𝐷3 , 𝐷4 } of the four various diseases affecting the plants.
In addition to these, also consider another set 𝑆 = {𝑆1 , 𝑆2 , 𝑆3 , 𝑆4 } be the common

EL
symptoms of the diseases.
Let, 𝑅 be a relation on 𝑃 × 𝐷, representing which plant is susceptible to which diseases,

PT
which is stated as

N
𝑃1 0.6 0.6 0.9 0.8
𝑃 0.1 0.2 0.9 0.8
𝑅= 2
𝑃3 0.9 0.3 0.4 0.8
𝑃4 0.9 0.8 0.4 0.2
Debasis Samanta
CSE
16
IIT Kharagpur
Fuzzy relation : An example
Also, consider 𝑇 be the another relation on 𝐷 × 𝑆, which is given by

𝐷1 0.1 0.2 0.7 0.9

𝐷 1.0 1.0 1.4 0.6
𝑆= 2
𝐷3 0.0 0.0 0.5 0.9

EL
𝐷4 0.9 1.0 0.8 0.2

PT
Obtain the association of plants with the different symptoms of the disease using
max-min composition.

N
Hint: Find 𝑅 ∘ 𝑇, and verify that
𝑃1 0.8 0.8 0.8 0.9
𝑃 0.8 0.8 0.8 0.9
𝑅∘𝑆 = 2
𝑃3 0.8 0.8 0.8 0.9
𝑃4 0.8 0.8 0.7 0.9
Debasis Samanta
CSE
17
IIT Kharagpur
Fuzzy relation : Another example
Let, 𝑅 = 𝑥 is relevant to 𝑦
and 𝑆 = 𝑦 is relevant to 𝑧
be two fuzzy relations defined on 𝑋 × 𝑌 and 𝑌 × 𝑍, respectively, where 𝑋 =
{1,2,3}, 𝑌 = {𝛼, 𝛽, 𝛾, 𝛿} and 𝑍 = {𝑎, 𝑏}. Assume that 𝑅 and 𝑆 can be

EL
expressed with the following relation matrices :

PT
𝛼 0.9 0.1
1 0.1 0.3 0.5 0.7

N
𝛽 0.2 0.3
𝑅 = 2 0.4 0.2 0.8 0.9 𝑎𝑛𝑑 𝑆 = 𝛾
0.5 0.6
3 0.6 0.8 0.3 0.2 𝛿 0.7 0.2

Debasis Samanta
CSE
18
IIT Kharagpur
Fuzzy relation : Another example
Now, we want to find 𝑅 ∘ 𝑆, which can be interpreted as a derived fuzzy relation
𝑥 is relevant to 𝑧.
Suppose, we are only interested in the degree of relevance between 2 ∈ 𝑋 and
𝑎 ∈ 𝑍. Then, using max-min composition,

EL
𝜇𝑅∘𝑆 (2, 𝑎) = max{ 0.4 ∧ 0.9 , 0.2 ∧ 0.2 , 0.8 ∧ 0.5 , 0.9 ∧ 0.7 }
= max{0.4, 0.2, 0.5, 0.7} = 0.7

PT
R s

N
1

 a
2

 b
3

Debasis Samanta
CSE
19
IIT Kharagpur
2D Membership functions : Binary fuzzy relations
(Binary) fuzzy relations are fuzzy sets 𝐴 × 𝐵 which map each element in 𝐴 × 𝐵
to a membership grade between 0 and 1 (both inclusive).
Note that a membership function of a binary fuzzy relation can be depicted with
a 3D plot.

EL
 ( x, y )

PT
N
Important: Binary fuzzy relations are fuzzy sets with two dimensional MFs and
so on.

Debasis Samanta
CSE
20
IIT Kharagpur
2D membership function : An example
+
Let, 𝑋 = 𝑅 = 𝑦 (the positive real line) and
𝑅 = 𝑋 × 𝑌 = "𝑦 is much greater than 𝑥“
The membership function of 𝜇𝑅 𝑥, 𝑦 is defined as
(𝑦 − 𝑥)

EL
𝑖𝑓 𝑦 > 𝑥
𝜇𝑅 𝑥, 𝑦 = 4
0 𝑖𝑓 𝑦 ≤ 𝑥

PT
Suppose, 𝑋 = {3,4,5} and 𝑌 = {3,4,5,6,7}, then

N
3 0
𝑅=4 0
5 0
0.25
0
0
0.5
0.25
0
0.75
0.5
0.25
1.0
0.75
0.5

Debasis Samanta
CSE
21
IIT Kharagpur
Example:
How you can derive the following?
If x is A or y is B then z is C;
Given that
• R1: If x is A then z is 𝐶 [𝑅1 ∈ 𝐴 × 𝐶]

EL
• R2: If y is B then z is 𝐶 [𝑅2 ∈ 𝐵 × 𝐶]

PT
Hint:
 You have given two relations R1 and R2.

N
 Then, the required can be derived using the union operation of R1
and R2

Debasis Samanta
CSE
22
IIT Kharagpur
 EL
PT
N
Debasis Samanta
CSE
23
IIT Kharagpur