Fuzzy Sets, Logic and Systems & Applications

Lecture 01

1 Introduction
1.1 What are Fuzzy Systems?
• Definition: Fuzzy systems are computational frameworks based on fuzzy logic, a
form of multivalued logic designed to handle ambiguity and partial truths.

• Motivation: Traditional binary logic (true/false) is too rigid for uncertain or im-
precise information. Fuzzy systems emulate human reasoning, where statements
can be “partly true” instead of strictly true or false.

1.2 Relevance in Artificial Intelligence (AI)

• Role in AI: Fuzzy logic is foundational in Computational Intelligence (CI) for
tackling complex, uncertain problems.

• Key Benefit: Allows intuitive and flexible modeling of real-world problems with
linguistic or subjective terms (e.g., “hot,” “cold,” “moderate”).

2 Historical Context of Artificial Intelligence

2.1 Early Foundations
• 1936-1937: Alan Turing proposed the Universal Turing Machine.

• 1940s: Warren McCulloch and Walter Pitts developed threshold logic, an early
model of neural networks.

2.2 Mid-Century Milestones

• 1950: The Turing Test was proposed to assess machine intelligence.

• 1955: John McCarthy coined the term “Artificial Intelligence.”

• 1957: The Perceptron model was introduced as an early neural network capable of
learning.

1
2.3 Emergence of Fuzzy Logic
• 1965: Lotfi A. Zadeh introduced Fuzzy Logic to mathematically handle linguistic
and ambiguous variables.

3 Overview of Computational Intelligence (CI)

3.1 Definition of CI
• CI refers to nature-inspired methodologies, such as fuzzy systems, neural networks,
and evolutionary algorithms, for solving complex and uncertain problems.

3.2 Key Agents in CI

1. Fuzzy Systems: Approximate reasoning and uncertainty modeling.

2. Neural Networks: Learning and pattern recognition.

3. Evolutionary Algorithms: Optimization using genetic principles.

4. Hybrid Systems: Combining multiple approaches (e.g., Neuro-Fuzzy systems).

4 Fuzzy Logic: Basics and Concepts

4.1 Multivalent Logic
• Traditional (Binary) Logic: Statements are strictly “true” or “false.”

• Fuzzy (Multivalent) Logic: Statements can have degrees of truth in the range
[0, 1].

4.2 Fuzzy Sets

• A fuzzy set allows elements to have varying degrees of membership, represented by
a membership function µA (x) ∈ [0, 1].

4.3 Membership Functions

• Common types: Triangular, Trapezoidal, Gaussian, and Bell-shaped.

• Choice of membership function affects system smoothness and performance.

5 Fuzzy Logic vs. Boolean Logic

5.1 Boolean Logic
• Binary and rigid: Statements are either “true” (1) or “false” (0).

2
5.2 Fuzzy Logic
• Flexible and nuanced: Allows “partial truths.”

• Example: A glass can be “half-full” or “mostly empty,” representing intermediate

degrees of truth.

6 Applications of Fuzzy Logic

1. Industrial Control Systems: Process automation in factories, robotics.

2. Consumer Electronics: Air conditioners, washing machines.

3. Automotive: Anti-lock braking systems, engine control.

4. Medical Diagnostics: Handling imprecise symptoms to aid decision-making.

5. Finance: Risk analysis and forecasting.

7 Key Takeaways
• Fuzzy Logic bridges binary logic and real-world complexity by allowing partial
truths.

• It is foundational to Computational Intelligence and AI, enabling robust solutions

for uncertainty and vagueness.

• Widely used in applications ranging from consumer products to industrial systems

and healthcare.

Quote by Lotfi A. Zadeh

“As the complexity of a system increases, it becomes more difficult and even-
tually impossible to make a precise statement about its behavior.”

3
 Fuzzy Sets, Logic and Systems & Applications
Lecture 02

1 Introduction to Fuzzy Systems in Applications

• Fuzzy Systems are effective in solving real-world problems where data is ambigu-
ous, imprecise, or incomplete.

• These systems are commonly applied in various domains, such as control systems,
decision-making, and medical diagnostics.

2 Real-Life Applications of Fuzzy Systems

2.1 Home Appliances
• Air Conditioners: Adjust cooling based on fuzzy temperature ranges.

• Washing Machines: Modify washing cycles based on fuzzy inputs like load size
and dirt level.

2.2 Automotive Systems

• Anti-lock Braking Systems (ABS): Use fuzzy logic to enhance braking efficiency
under varying road conditions.

• Cruise Control: Maintains optimal speed by adjusting acceleration and braking

fuzzily.

2.3 Healthcare
• Disease Diagnosis: Fuzzy systems incorporate imprecise symptom descriptions
to make diagnostic decisions.

• Treatment Recommendations: Evaluate uncertain medical parameters to pro-

pose suitable treatments.

2.4 Industrial Systems

• Process Control: Manage variables like temperature, pressure, and flow rate using
fuzzy logic.

• Robotics: Motion planning and control in dynamic, uncertain environments.

1
2.5 Finance and Risk Analysis
• Forecasting: Handle ambiguous market indicators to predict trends.

• Risk Assessment: Incorporate subjective data to calculate risk levels.

3 Why Fuzzy Logic Works Well

• Handles Uncertainty: Naturally accommodates ambiguity and vagueness in
data.

• Human-Like Reasoning: Mimics human decision-making with soft boundaries

instead of strict thresholds.

• Improved System Efficiency: Addresses nonlinearities and complexities effec-

tively.

4 Key Characteristics of Fuzzy Systems in Applica-

tions
• Flexibility: Can adapt to various application needs.

• Robustness: Provides reliable outputs despite uncertainties in inputs.

• Scalability: Easily extends to multi-variable systems.

5 Key Takeaways
• Fuzzy systems are widely applied in domains where traditional systems struggle
due to ambiguous, nonlinear, or uncertain conditions.

• They are especially effective in consumer electronics, automotive systems, health-

care, and finance.

• The ability to handle uncertainty and mimic human reasoning makes fuzzy systems
indispensable in modern technology.

2
 Fuzzy Sets, Logic and Systems & Applications
Lecture 03

1 Introduction to MATLAB Fuzzy Logic Toolbox

• MATLAB provides a Fuzzy Logic Toolbox for designing and simulating fuzzy
systems.

• This toolbox is a powerful tool for modeling systems with uncertainties and non-
linear behavior.

2 Features of MATLAB Fuzzy Logic Toolbox

• Defining Inputs and Outputs:

– Specify variables for fuzzy systems, such as temperature, speed, or pressure.

• Creating Membership Functions:

– Assign membership functions to inputs and outputs.

– Supported shapes include Triangular, Trapezoidal, Gaussian, and Bell-shaped
functions.

• Building Fuzzy Rules:

– Logical IF-THEN rules link inputs to outputs using fuzzy logic operators.

• Simulating Fuzzy Systems:

– Test and visualize the behavior of fuzzy systems under various input condi-
tions.

3 Steps to Create a Fuzzy System in MATLAB

1. Define the Input and Output Variables:

• Example: Input variables could be “temperature” and “humidity,” while the

output variable might be “fan speed.”

2. Create Membership Functions for Each Variable:

• Use triangular, trapezoidal, or other shapes to define the degree of membership.

1
 3. Define Fuzzy Rules:

• Example: “IF temperature is high AND humidity is low, THEN fan speed is
high.”

4. Simulate the Fuzzy System:

• Visualize the behavior of the system for different input scenarios.

4 Applications in MATLAB
• Control Systems:

– Example: Designing a fuzzy controller for temperature regulation in an HVAC

system.

• Decision-Making Systems:

– Example: Optimizing resource allocation based on fuzzy logic.

• Optimization and Simulation:

– Example: Tuning system parameters using fuzzy inference.

5 Key Takeaways
• The MATLAB Fuzzy Logic Toolbox simplifies the design and implementation of
fuzzy systems.

• It provides tools for defining variables, creating membership functions, building

rules, and simulating fuzzy systems.

• Applications span various domains, including control systems, decision-making, and

optimization tasks.

2
 Fuzzy Sets, Logic and Systems & Applications
Lecture 04

1 Advanced Features of MATLAB Fuzzy Logic Tool-

box
1.1 Overview
The MATLAB Fuzzy Logic Toolbox provides advanced tools to design, analyze, and
simulate fuzzy systems. These tools enable users to create and refine fuzzy inference
systems for complex real-world applications.

1.2 Key Features

• Rule Editor:

– Allows the creation, modification, and deletion of fuzzy rules.

– Supports logical operators (AND, OR) for complex rule definitions.

• Membership Function Editor:

– Enables customization of membership function shapes.

– Supports a wide range of functions, including Gaussian, triangular, trape-
zoidal, and bell-shaped.

• Simulation Environment:

– Provides real-time visualization of system outputs.

– Allows testing of the fuzzy system with various input scenarios.

2 Steps for Building and Refining Fuzzy Systems

2.1 Step-by-Step Process
1. Define Inputs and Outputs:

• Example: Inputs could be “temperature” and “pressure,” and the output could
be “flow rate.”

2. Create Membership Functions:

1
 • Define membership functions for each input and output variable.
• Choose appropriate shapes based on system requirements.

3. Develop Fuzzy Rules:

• Use IF-THEN logic to define rules linking inputs to outputs.

• Example: “IF temperature is high AND pressure is low, THEN flow rate is
medium.”

4. Simulate and Test the System:

• Run simulations to observe system performance under different input condi-

tions.
• Adjust rules or membership functions if necessary.

5. Optimize the System:

• Use optimization tools or fine-tune membership functions for better results.

3 Applications of Fuzzy Systems in MATLAB

• Dynamic Systems:

– Simulate real-time control scenarios, such as temperature regulation.

• Case Studies:

– Traffic light control for smoother vehicle flow.

– Water heating systems adjusting temperature based on fuzzy inputs.

• Decision-Making Systems:

– Multi-criteria decision analysis (e.g., evaluating risks or optimizing resource

allocation).

4 Key Takeaways
• The MATLAB Fuzzy Logic Toolbox provides an intuitive interface for creating,
testing, and refining fuzzy systems.

• Advanced features like the Rule Editor and Membership Function Editor make it
easy to customize and optimize systems.

• Applications span dynamic control, decision-making, and real-time simulations,

highlighting the versatility of fuzzy logic in engineering and science.

2
 Fuzzy Sets, Logic and Systems & Applications
Lecture 05

1 Introduction to Membership Functions

• Definition: Membership functions define the degree to which an element belongs
to a fuzzy set.

• Role in Fuzzy Systems:

– Represents uncertainty and fuzziness in a system.

– Serves as the basis for fuzzification and defuzzification processes.

2 Key Properties of Membership Functions

• Range: Values lie between 0 and 1, indicating the degree of membership.

• Shape: Determines how an input maps to membership levels.

• Smooth Transitions: Enables gradual changes in membership, crucial for real-

world systems.

3 Types of Membership Functions

• Triangular Membership Function:

– Simple and computationally efficient.

– Defined by three parameters: a, b, c, where a and c are the base points, and b
is the peak.
– Mathematical expression:


 0, x<a
 x−a , a ≤ x ≤ b

b−a
µ(x) = c−x (1)
 , b≤x≤c
 c−b


0, x>c

• Trapezoidal Membership Function:

– Defined by four parameters: a, b, c, d.

1
 – Flat top allows for a wider range of full membership.
– Mathematical expression:


 0, x<a

x−a
 b−a , a≤x≤b



µ(x) = 1, b≤x≤c (2)
 d−x
, c≤x≤d




 d−c
0, x>d

• Gaussian Membership Function:

– Smooth and continuous, ideal for natural systems.

– Defined by a mean (c) and a standard deviation (σ).
– Mathematical expression:

(x − c)2
 
µ(x) = exp − (3)
2σ 2

• Bell-Shaped Membership Function:

– Similar to Gaussian but allows for more control over the steepness.
– Mathematical expression:
1
µ(x) = (4)
x−c 2b
1+ a

where a determines the width, b controls the slope, and c is the center.

• Piecewise Membership Function:

– Combines multiple types for more complex systems.

– Example: A combination of triangular and trapezoidal functions for specialized
scenarios.

4 Designing Membership Functions

• System Requirements:

– Understand the system’s inputs and outputs.

– Define ranges and boundaries based on domain knowledge.

• Optimization:

– Use data-driven techniques to refine membership function parameters.

– Tools like MATLAB’s Membership Function Editor help visualize and optimize
shapes.

2
5 Applications of Membership Functions
• Control Systems:

– Example: Adjusting fan speed based on temperature inputs.

• Decision-Making:

– Example: Grading systems where performance levels are not binary.

• Image Processing:

– Example: Edge detection using fuzzy rules.

6 Key Takeaways
• Membership functions are the backbone of fuzzy systems, enabling the representa-
tion of uncertainty and vagueness.

• Different types (e.g., triangular, trapezoidal, Gaussian) are suited to different ap-
plications.

• Proper design and optimization of membership functions are critical for achieving
accurate and efficient fuzzy systems.

3
 Fuzzy Sets, Logic and Systems & Applications
Week 2: Lectures 06 - 10

Membership Functions
There are some commonly employed membership functions (MFs) in fuzzy theory, as
listed below.

1. Triangular Membership Function:

A triangular membership can be defined by three parameters [a, b, c] with (a < b <
c). These parameters determine the coordinates of three corners of a triangular
membership function. The membership function is defined as follows:



 0 x≤a



 x−a a≤x≤b


b−a
triangle(x; a, b, c) =
c−x
b≤x≤c




 c−b



0 c≤x

An alternate expression for the triangular function using min and max can be
written as:
   
x−a c−x
triangle(x; a, b, c) = max min , ,0
b−a c−b

2. Trapezoidal Membership Function:

A trapezoidal MF is specified by four parameters [a, b, c, d] with a < b ≤ c < d as
follows:


0 x≤a



x−a
a≤x≤b



 b−a



trapezoid(x; a, b, c, d) = 1 b≤x≤c



d−x




 d−c
c≤x≤d



d≤x

0

1
 An alternate expression for the trapezoidal function using min and max can be
written as:
   
x−a d−x
trapezoid(x; a, b, c, d) = max min , 1, ,0
b−a d−c
3. Gaussian Membership Function:
A Gaussian MF is specified by two parameters [c, σ] as:

1 x − c !2
−
gaussian(x; c, σ) = e 2 σ

where c represents the MF’s center and σ determines the MF’s width.
4. Generalized Bell-shaped Membership Function:
A generalized bell-shaped MF (or bell MF) is specified by three parameters [a, b, c]
as:

1
bell(x; a, b, c) =
x−c 2b
1+ a

Here:

• a defines the width of the membership function i.e., larger value creates a
wider membership function.
• b defines the shape of the curve on either side of the central plateau i.e., larger
value creates a steeper transition.
• c defines the center of the membership function.

5. Sigmoidal Membership Function:

The sigmoidal membership function is given by:

1
Sig(x; a, c) =
1 + e−a(x−c)
Here,

• The magnitude of a controls the width of the transition area, and

• c defines the center of the transition area.

6. Left-right (L-R) Membership Function:

A left-right MF or L-R MF is specified by three parameters [α, β, c] as:

c−x


FL

α
x≤c
LR(x; c, α, β) =  
x−c
FR

β
x≥c

where FL (.) and FR (.) are monotonically decreasing functions, defined on [0, ∞)
with FL (0) = FR (0) = 1 and limx→∞ FL (.) = limx→∞ FR (.) = 0.

2
 • c defines the point where the value of the membership function is 1.
• α controls the width of the left region for x ≤ c.
• β controls the width of the right region for x ≥ c.

7. Pi-shaped (π) Membership Function:

There are two π membership functions. The first π membership function is defined
with two parameters [a, b]. The function has a membership value of 1 at point a
and a membership value of 0.5 at a − b and a + b, respectively. Unlike the function,
π function decreases towards zero asymptotically as we move away from point a.
The first π membership function is defined as:

1
π1 (x; a, b) =
x−a 2


1+ b

The another four-variable π membership function with four parameters is defined

as:

lw

 lp+lw−x
x < lp



π2 (x; lw, lp, rp, rw) = 1 lp ≤ x ≤ rp



 rw

x > rp
x−rp+rw

Parameters lw and rw define the feet of the membership function, and lp and rp
define its shoulders, respectively. This membership function was used to define
the fuzzy set “Good Stopping Accuracy” in the Sendai Subway Train Control
System.

8. Open Right Membership Function:

The open right membership function is defined by two parameters [α, β] as follows:



 0 x<α


x−α
OpenR(x; α, β) = β−α
α≤x≤β




1 x>β

9. Open Left Membership Function:

The open left membership function is defined by two parameters [α, β] as follows:



1 x<α


β−x
OpenL(x; α, β) = β−α
α≤x≤β




0 x>β

3
 10. S-shaped Membership Function:
A S-shaped membership function is defined by two parameters [a, b] as follows:



 0 x≤a


x−a 2

a+b


2 a≤x≤


b−a 2
f (x; a, b) =
x−b 2 a+b


1−2 ≤x≤b




 b−a 2



1 x≥b

The parameter a defines the feet of the membership function, and b defines its
shoulder.

Nomenclatures used in Fuzzy Set Theory

There are some commonly used nomenclatures in fuzzy set theory as listed below.

1. Support:
The support of a fuzzy set A is the set of all points x in the universe of discourse
X for any associated membership function such that:

µA (x) > 0 (1)

It can be represented mathematically as:

Support(A) = {x | µA (x) > 0} (2)

where x belongs to the universe of discourse X. The support of A consists of all

values of x where the membership function is greater than zero.
Additionally, this condition can be expressed as:

µA (x) > 0, ∀x ∈ (x1 , x2 ) in the universe of discourse X (3)

Outside this range:

µA (x) = 0, ∀x ∈
/ (x1 , x2 ) (4)

2. Core:
The core of fuzzy set A is the set of all points x in the universe of discourse X such
that:

µA (x) = 1 (5)

It can be represented mathematically as:

4
 Core(A) = {x | µA (x) = 1} (6)

For the given fuzzy set A with universe of discourse X, we can express this as:

µA (x) = 1, ∀x ∈ [x1 , x2 ] (7)

3. Crossover Points:
The crossover points of a fuzzy set A are the points x in the universe of discourse
X at which:

µA (x) = 0.5 (8)

It can be represented mathematically as:

Crossover(A) = {x | µA (x) = 0.5} (9)

For the given fuzzy set A with universe of discourse X, the crossover points are:

Crossover(A) = {x1 , x2 } (10)

4. Height of a Fuzzy Set:

• The height of a fuzzy set A is the maximum value of the membership function.
It is represented as:

hgt(A) = max{µA (x)} (11)

If hgt(A) < 1, then the fuzzy set A is called a Subnormal fuzzy set.
The hgt(A) is viewed as the degree of validity or credibility of information expressed
by the fuzzy set A.
For example, given fuzzy sets A, B, and C:

hgt(A) = 1, hgt(B) = 0.40, hgt(C) = 0.75 (12)

5. Normality of a Fuzzy Set:

A fuzzy set A is normal if its core is non-empty.
In other words, we can always find a point x in the universe of discourse X such
that:

µA (x) = 1 (13)

The core of the fuzzy set A is given by:

Core(A) = {x2 } (14)

Since the core of A is non-empty, the fuzzy set A is a normal fuzzy set.

5
6. Subnormal Fuzzy Sets:
A fuzzy set A is called normal if there exists at least one element x in the universe
of discourse X such that:

µA (x) = 1 (15)

A fuzzy set that does not satisfy this condition, meaning:

max µA (x) < 1 (16)

is called a subnormal fuzzy set.

A normal fuzzy set reaches a membership degree of 1 for at least one element in X.
A subnormal fuzzy set** never reaches a membership value of 1. The height of a
fuzzy set A is given by:

hgt(A) = max{µA (x)} (17)

If hgt(A) < 1, then A is a subnormal fuzzy set.

Subnormal fuzzy sets are useful in cases where uncertainty does not fully support
the membership of any element to the set with full confidence.
Normalization of a Fuzzy Set:
The fuzzy set A′ is the normalized version of a fuzzy set A, which is represented by:

A′ = Norm(A) (18)

The normalization is carried out as follows:

X µA (x)
A′ = Norm(A) = /x (19)
X
hgt(A)

where:

• A′ is the normalized fuzzy set.

• hgt(A) is the height of the fuzzy set A, defined as max{µA (x)}.
• The normalization ensures that the membership values are scaled in such a
way that the highest membership degree reaches 1.

7. Fuzzy Singleton:
A fuzzy set whose support is a single point in the universe of discourse X with
µA (x) = 1 is called a Fuzzy Singleton. In other words, we can say that:

“A fuzzy singleton has a core with only one element in its set.”

6
 Mathematically, it is represented as:

Support(A) = {x | µA (x) > 0} = {x1 } (20)

Core(A) = {x | µA (x) = 1} = {x1 } (21)

This means that a fuzzy singleton has exactly one element where the membership
function is equal to 1.

8. α-Cut of a Fuzzy Set:

The α-Cut or α-level set of a fuzzy set A is a crisp set defined as:

Aα = {x | µA (x) ≥ α} (22)

where:

• Aα represents the subset of elements in X where the membership function

µA (x) is at least α.
• The α-Cut transforms a fuzzy set into a crisp set by including only those
elements with membership values above the threshold α.
• This concept is useful in fuzzy logic applications, such as decision-making and
approximate reasoning.

9. Strong α-Cut of a Fuzzy Set:

The strong α-Cut or strong α-level set of a fuzzy set A is a crisp set defined as:

A′α = {x | µA (x) > α} (23)

where:

• A′α represents the subset of elements in X where the membership function

µA (x) is strictly greater than α.
• Unlike the standard α-Cut, which includes elements where µA (x) ≥ α, the
strong α-Cut excludes elements where µA (x) = α.
• This distinction is important in fuzzy logic applications where strict inclusion
criteria are required.

α-Cut and Strong α-Cut of a Fuzzy Set:

Using the notation for a level set, we can represent an α-Cut and Strong α-Cut for
a fuzzy set A as:

Aα = {x ∈ X | A(x) ≥ α} A′α = {x ∈ X | A(x) > α}

Core(A) = {x | µA (x) = 1} Support(A) = {x | µA (x) > 0}

Additionally, the relationships between the core and support of a fuzzy set can be
written as:

7
 Core(A) = A1 Support(A) = A′0

where:

• Aα represents the set of elements with membership values greater than or equal
to α.
• A′α represents the set of elements with membership values strictly greater than
α.
• Core(A) contains elements where membership is exactly 1.
• Support(A) contains elements where membership is positive.

10. Convexity of a Fuzzy Set:

A fuzzy set A is convex if and only if for any x1 , x2 ∈ X and any λ ∈ [0, 1], the
following condition holds:

µA (λx1 + (1 − λ)x2 ) ≥ min[µA (x1 ), µA (x2 )] (24)

where:

• Alternatively, A is convex if all its α-level sets are convex.

• A convex fuzzy set is described by a membership function whose values are:
– Strictly monotonically increasing.
– Strictly monotonically decreasing.
– Strictly monotonically increasing and then strictly monotonically decreas-
ing with increasing values for elements in the universe of discourse.
• For any elements x1 , x2 , x3 in a fuzzy set A with x1 < x2 < x3 , the condition
for convexity is defined as:

µA (x2 ) ≥ min[µA (x1 ), µA (x3 )] (25)

Properties of Convex Fuzzy Sets:

• A special property of convex fuzzy sets is that the intersection of two convex
fuzzy sets is also convex.
• If fuzzy sets A and B are convex, then the fuzzy set A ∩ B is also convex.

Comparison with Convexity of Crisp Sets and Functions:

• A crisp set C in Rn is convex if and only if for any two points x1 ∈ C and
x2 ∈ C, their convex combination λx1 +(1−λ)x2 is still in C, where 0 ≤ λ ≤ 1.
• The definition of convexity of a fuzzy set is not as strict as the common def-
inition of convexity for functions. For comparison, the definition of convexity
of a function f (x) is given as:

f (λx1 + (1 − λ)x2 ) ≥ λf (x1 ) + (1 − λ)f (x2 ) (26)

8
11. Cardinality of a Fuzzy Set:
In a crisp set, the cardinality of a set is a measure of the ”number of elements
of the set”. For example, the set:

A = {5, 10, 15, 20}

contains 4 elements, and therefore A has a cardinality of 4.

In contrast, for a continuous fuzzy set, the universe of discourse contains an infinite
number of elements.
Thus, the Cardinality of a Continuous Fuzzy Set is Infinite.

12. Fuzzy Number:

• A fuzzy number A is a fuzzy set in the real number space (R) that satisfies
the condition for normality and convexity.
• Most (non-composite) fuzzy sets used in the literature satisfy the condition for
normality and convexity, so fuzzy numbers are the most basic type of fuzzy
sets.

13. Bandwidth of Normal and Convex Fuzzy Sets:

For a normal and convex fuzzy set A, the bandwidth or width is defined as the
distance between the two unique crossover points.

Width(A) = |x1 − x2 |

Here,
µA (x1 ) = µA (x2 ) = 0.5

14. Fuzzy Symmetry:

A fuzzy set A is symmetric if its membership function is symmetric around a certain
point x = c, i.e., it satisfies the condition given below for the universe of discourse
X:
µA (c + x) = µA (c − x) ∀x ∈ X

15. Open Left/Open Right Fuzzy Set:

• A fuzzy set A is open left if

lim µA (x) = 1 and lim µA (x) = 0

x→−∞ x→+∞

• A fuzzy set B is open right if

lim µB (x) = 0 and lim µB (x) = 1

x→−∞ x→+∞

9
Set Theoretic Operations on Fuzzy Sets
• Set Theory Notations:
For two sets A and B, which consist of the collection of some elements in the
universe of discourse X, the set theory notations are defined as follows:

x∈X → x belongs to X
x∈A → x belongs to A
x∈
/A → x does not belong to A
A⊂B → A is fully contained in B
A⊆B → A is contained in B and equivalent to B
A=B → A is equivalent to B, i.e., A ⊆ B and B ⊆ A
ϕ → Null/empty set i.e. the set contains no element

• Equality of Fuzzy Sets:

Let A and B be two fuzzy sets. They are equal if and only if the following conditions
are satisfied:

µA (x) = µB (x), ∀x ∈ X
Here, X is the universe of discourse, and µA (x) and µB (x) are membership degrees
for an element x in the fuzzy sets A and B, respectively.
• Containment (Subset) of a Fuzzy Set:
Fuzzy set B contains another fuzzy set A if and only if the following condition
satisfies:

µA (x) ≤ µB (x), ∀x ∈ X
It can be denoted by:

A⊆B
• Complement of a Classical Set:
The complement of a set A, denoted by Ā, represents all elements in the universe
of discourse X that do not belong to set A.
It can be represented as:

Ā = {x | x ∈
/ A and x ∈ X}
• Complement of a Fuzzy Set:
The complement of a fuzzy set A is a fuzzy set Ā in the universe of discourse X
whose membership function is defined as:

µĀ (x) = 1 − µA (x) ∀x ∈ X

The complement of a fuzzy set should contain all elements along with their mem-
bership values from the universe of discourse.

10
 Fuzzy Sets, Logic and Systems & Applications
Week 3: Lectures 11 - 15

Set Theoretic Operations on Fuzzy Sets

• Set Theory Notations:
For two sets A and B, which consist of the collection of some elements in the
universe of discourse X, the set theory notations are defined as follows:

x∈X → x belongs to X
x∈A → x belongs to A
x∈
/A → x does not belong to A
A⊂B → A is fully contained in B
A⊆B → A is contained in B and equivalent to B
A=B → A is equivalent to B, i.e., A ⊆ B and B ⊆ A
ϕ → Null/empty set i.e. the set contains no element

• Equality of Fuzzy Sets:

Let A and B be two fuzzy sets. They are equal if and only if the following conditions
are satisfied:

µA (x) = µB (x), ∀x ∈ X

Here, X is the universe of discourse, and µA (x) and µB (x) are membership degrees
for an element x in the fuzzy sets A and B, respectively.

• Containment (Subset) of a Fuzzy Set:

Fuzzy set B contains another fuzzy set A if and only if the following condition
satisfies:

µA (x) ≤ µB (x), ∀x ∈ X

It can be denoted by:

A⊆B

1
• Complement of a Classical Set:
The complement of a set A, denoted by Ā, represents all elements in the universe
of discourse X that do not belong to set A.
It can be represented as:

Ā = {x | x ∈
/ A and x ∈ X}

• Complement of a Fuzzy Set:

The complement of a fuzzy set A is a fuzzy set Ā in the universe of discourse X
whose membership function is defined as:

µĀ (x) = 1 − µA (x) ∀x ∈ X

The complement of a fuzzy set should contain all elements along with their mem-
bership values from the universe of discourse.

• Union of Classical Sets:

The union of two classical sets represents all elements in the universe of discourse
X that belong to either the set A or the set B, or both. It is denoted by A ∪ B.
It can be represented as:

A ∪ B = {x | x ∈ A or x ∈ B}

• Union of Fuzzy Sets:

The union of two fuzzy sets A and B in the universe of discourse X is defined as:

µA∪B (x) = max[µA (x), µB (x)] ∀x ∈ X

• Intersection of Classical Sets:

The intersection of two sets A and B represents all elements in the universe of
discourse X that simultaneously belong to both sets A and B. It is denoted by
A ∩ B.
It can be represented as:

A ∩ B = {x | x ∈ A and x ∈ B}

• Intersection of Fuzzy Sets:

The intersection of two fuzzy sets A and B represents all elements for which the
corresponding membership values can be computed as:

µA∩B (x) = min[µA (x), µB (x)] ∀x ∈ X

2
 • Difference of Classical Sets:
The difference of a set A with respect to B is defined as the collection of all elements
in the universe of discourse X that belong to set A but do not belong to B. It is
denoted by A|B.
It can be represented as:

A|B = {x | x ∈ A and x ∈
/ B}, B|A = {x | x ∈
/ A and x ∈ B}

• Difference of Fuzzy Sets:

For the given fuzzy sets A and B with the membership function values given as
µA (x) and µB (x), respectively, in the universe of discourse X, the fuzzy difference
is given as:

µA|B (x) = min[µA (x), µB̄ (x)] ∀x ∈ X

µB|A (x) = min[µB (x), µĀ (x)] ∀x ∈ X

Properties of Fuzzy Sets

Table 1: Properties of Classical and Fuzzy Sets

Property CLASSICAL SETS FUZZY SETS
Law of Contradiction A ∩ Ā = ϕ A ∩ Ā ̸= ϕ
Law of Excluded Middle A ∪ Ā = X A ∪ Ā ̸= X
Idempotency A ∩ A = A, A ∪ A = A A ∩ A = A, A ∪ A = A
Involution Ā¯ = A Ā¯ = A
Commutativity A ∩ B = B ∩ A, A ∪ B = B ∪ A A ∩ B = B ∩ A, A ∪ B = B ∪ A
Associativity (A ∪ B) ∪ C = A ∪ (B ∪ C) (A ∪ B) ∪ C = A ∪ (B ∪ C)
(A ∩ B) ∩ C = A ∩ (B ∩ C) (A ∩ B) ∩ C = A ∩ (B ∩ C)
Distributivity A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C) A ∪ (B ∩ C) = (A ∪ B) ∩ (A ∪ C)
A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C) A ∩ (B ∪ C) = (A ∩ B) ∪ (A ∩ C)
Absorption A ∪ (A ∩ B) = A A ∪ (A ∩ B) = A
A ∩ (A ∪ B) = A A ∩ (A ∪ B) = A
Absorption of Complement A ∪ (Ā ∩ B) = A ∪ B A ∪ (Ā ∩ B) ̸= A ∪ B
A ∩ (Ā ∪ B) = A ∩ B A ∩ (Ā ∪ B) ̸= A ∩ B
DeMorgan’s Laws A ∪ B = Ā ∩ B̄ A ∪ B = Ā ∩ B̄
A ∩ B = Ā ∪ B̄ A ∩ B = Ā ∪ B̄

If we compare the properties of crisp sets with fuzzy sets, we observe that the following
three properties are not similar:

• Law of Contradiction

• Law of Excluded Middle

• Absorption of Complement

3
 Fuzzy Sets, Logic and Systems & Applications
Week 4: Lectures 16 - 20

Distance between fuzzy sets

Within the universe of discourse X, the distance d(A, B) between two fuzzy sets A and
B can be defined by the extension principle. The distance d(A, B) is given as:
X µd(A,B) (δ)
d(A, B) = → For discrete fuzzy sets A and B
X
δ

µd(A,B) (δ)
Z
d(A, B) = dX → For continuous fuzzy sets A and B
X δ
where,

min µA (xA ), µB (xB ) ∀δ ∈ R+



µd(A,B) (δ) = max
δ=d(xA ,xB )

and δ is the difference between corresponding generic values of two fuzzy sets for a generic
variable.
Please note that xA and xB are the generic variables of the fuzzy sets A and B, respec-
tively, ∀xA , xB ∈ X.
Example 1: Find the distance d(A, B) between fuzzy sets A and B given below with
the universe of discourse X = {1, 2, 3, 4}.

A = {(1, 0.5), (2, 1), (3, 0.3)}

B = {(2, 0.4), (3, 0.4), (4, 1)}

Solution: For calculating the fuzzy distance d(A, B) between fuzzy sets A and B, the
membership function values µd(A,B) (δ) for δ ∈ R+ have to be obtained.

Fuzzy set A Fuzzy set B

A = {xA , µA (xA )} B = {xB , µB (xB )}

∀xA , xB ∈ X

Distance d(A, B) = {δ, µd(A,B) (δ)|∀δ ∈ R+ }

1
Arithmetic Operations on Fuzzy Numbers
There are four types of arithmetic operations that can be performed on fuzzy sets, pro-
vided fuzzy sets are qualified for fuzzy numbers. These operations are:

• Addition of fuzzy numbers

• Subtraction of fuzzy numbers

• Multiplication of fuzzy numbers

• Division of fuzzy numbers

Addition of Fuzzy Numbers

Let A and B be two fuzzy numbers with the universe of discourse X. If we perform the
addition, it results in a new fuzzy number C as,

C =A+B
The new fuzzy number C is defined as,
X µC (xC )
C= → For discrete
X
xC

µC (xC )
Z
C= dX → For continuous
X xC
The membership function values of a fuzzy number C are

µC (xC ) = µA+B (xC )

2
 µC (xC ) = µA+B (xC ) = max µA (xA ) ∧ µB (xB )
 
xA ,xB

where xC = xA + xB ; ∀xA , xB , xC ∈ X.
X µC (xC )
C=
X
xC
X µC (xC ) X µA+B (xC )
C= =
X
xC X
xC
 
X µC (xC ) X µA+B (xC ) X maxxA ,xB µA (xA ) ∧ µB (xB )
C= = =
X
xC X
xC X
xC

3
 Fuzzy Sets, Logic and Systems & Applications
Week 5: Lectures 21 - 25

Arithmetic Operations on Fuzzy Numbers

There are four types of arithmetic operations that can be performed on fuzzy sets, pro-
vided fuzzy sets are qualified for fuzzy numbers. These operations are:

• Addition of fuzzy numbers

• Subtraction of fuzzy numbers

• Multiplication of fuzzy numbers

• Division of fuzzy numbers

Addition of Fuzzy Numbers

Let A and B be two fuzzy numbers with the universe of discourse X. If we perform the
addition, it results in a new fuzzy number C as,

C =A+B (1)
The new fuzzy number C is defined as,
X µC (xC )
C= → For discrete (2)
X
xC

µC (xC )
Z
C= dX → For continuous (3)
X xC
The membership function values of a fuzzy number C are

µC (xC ) = µA+B (xC ) (4)

µC (xC ) = µA+B (xC ) = max µA (xA ) ∧ µB (xB )

 
(5)
xA ,xB

where xC = xA + xB ; ∀xA , xB , xC ∈ X.
X µC (xC )
C= (6)
X
xC

1
 X µC (xC ) X µA+B (xC )
C= = (7)
X
xC X
xC
 
X µC (xC ) X µA+B (xC ) X maxxA ,xB µA (xA ) ∧ µB (xB )
C= = = (8)
X
xC X
xC X
xC

Subtraction of Fuzzy Numbers

Let A and B be two fuzzy numbers with the universe of discourse X. The subtraction of
fuzzy numbers results in a new fuzzy number C given by,

C =A−B (9)
The membership function values of the fuzzy number C are defined as,

µC (xC ) = µA−B (xC ) (10)

µC (xC ) = max µA (xA ) ∧ µB (xB )

 
(11)
xA ,xB

where xC = xA − xB ; ∀xA , xB , xC ∈ X.

Multiplication of Fuzzy Numbers

Multiplication of fuzzy numbers results in a new fuzzy number C as,

C =A×B (12)
The membership function values of C are given by,

µC (xC ) = µA×B (xC ) (13)

µC (xC ) = max µA (xA ) ∧ µB (xB )

 
(14)
xA ,xB

where xC = xA × xB ; ∀xA , xB , xC ∈ X.

Division of Fuzzy Numbers

Division of fuzzy numbers results in a new fuzzy number C as,
A
C= (15)
B
The membership function values of C are given by,

µC (xC ) = µA/B (xC ) (16)

µC (xC ) = max µA (xA ) ∧ µB (xB )

 
(17)
xA ,xB

2
 xA
where xC = xB
; ∀xA , xB , xC ∈ X and xB ̸= 0.
Note: Addition, Subtraction, Multiplication, and Division of fuzzy numbers always result
in a fuzzy number provided the universe of discourse is sufficiently large.

Complement of Fuzzy Sets

Let c : [0, 1] → [0, 1] be a mapping function that transforms the membership function of
fuzzy set A into the membership function of the complement of fuzzy set A, i.e.,

c[µA (x)] = µĀ (x) (18)

For the basic fuzzy complement,
c[µA (x)] = 1 − µA (x). (19)
Axioms for Fuzzy Complement:
Axiom c1: c(0) = 1 and c(1) = 0 (boundary condition).
Axiom c2: ∀µ(x1 ), µ(x2 ) ∈ [0, 1], if µ(x1 ) ≤ µ(x2 ), then c(µ(x1 )) ≥ c(µ(x2 )) (non-
increasing condition).
Here µ(x1 ) and µ(x2 ) denote the membership values.
Any function c : [0, 1] → [0, 1] that satisfies Axioms c1 and c2 is called a fuzzy comple-
ment.

Sugeno’s Class of Complement

Sugeno’s class of complement is defined by
1 − µ(x)
cλ (µ(x)) = (20)
1 + λµ(x)
where, λ ∈ (−1, ∞).
For each value of the parameter λ, we obtain the complement of a fuzzy set. It is easier
to check that Sugeno’s class of complement satisfies Axioms c1 and c2 as stated in the
previous slide.
Note: If λ = 0 for Sugeno’s class of complement, it becomes the basic fuzzy complement,
i.e.,
cλ (µ(x)) = 1 − µ(x) (21)

Yager’s Class of Complement

Yager’s class of complement is defined by
cw (µA (x)) = (1 − µA (x)w )1/w (22)
where w ∈ (0, ∞).
For each value of the parameter w, we obtain the complement of a fuzzy set. It is easier
to check that Yager’s class of complement satisfies Axioms c1 and c2 as stated earlier.
Note: If w = 1 in Yager’s class of complement, it becomes the basic fuzzy complement,
i.e.,
cw (µA (x)) = 1 − µA (x) (23)

3
T-norm or S-conorm
Let T : [0, 1] × [0, 1] → [0, 1] be a mapping function that transforms the membership
functions of fuzzy sets A and B into the membership function of the T-norm of fuzzy
sets A and B with the universe of discourse X. It can be defined as:

T [µA (x), µB (x)] = µA∩B (x), ∀x ∈ X (24)

where, µA (x) and µB (x) denote the membership function values for fuzzy sets A and B,
respectively.

• ∧ is the T-norm operator.

• T-norm is also known as ”S-conorm”.

Axioms for T-norm

For the function T to be qualified as a fuzzy intersection, it must satisfy at least the
following four requirements:

Axiom T1: Boundary Condition

T [0, 0] = 0 (25)
T [µA (x), 1] = T [1, µA (x)] = µA (x) (26)

Axiom T2: Commutativity

T [µA (x), µB (x)] = T [µB (x), µA (x)] (27)

Axiom T3: Non-Decreasing

If µA (x) ≤ µB (x) and µC (x) ≤ µD (x) then T [µA (x), µC (x)] ≤ T [µB (x), µD (x)] (28)

Axiom T4: Associativity

T [T [µA (x), µB (x)], µC (x)] = T [µA (x), T [µB (x), µC (x)]] (29)
Any function T : [0, 1] × [0, 1] → [0, 1] that satisfies Axioms T1 to T4 is called a T-norm.

T-norm Operators
There are four commonly used T-norm operators:

Minimum
Tmin (µA (x), µB (x)) = min(µA (x), µB (x)) = µA (x) ∧ µB (x) (30)

Algebraic Product
Tap (µA (x), µB (x)) = µA (x) × µB (x) (31)

4
Bounded Product
Tbp (µA (x), µB (x)) = max(0, µA (x) + µB (x) − 1) (32)

Drastic Product

µA (x), if µB (x) = 1

Tdp (µA (x), µB (x)) = µB (x), if µA (x) = 1 (33)

0, if µA (x), µB (x) < 1


S-norm or T-conorm
Let S : [0, 1] × [0, 1] → [0, 1] be a mapping function that transforms the membership
functions of fuzzy sets A and B into the membership function of the S-norm of fuzzy sets
A and B with the universe of discourse X. It can be defined as:

S[µA (x), µB (x)] = µA∪B (x), ∀x ∈ X (34)

where, µA (x) and µB (x) denote the membership function values for fuzzy sets A and B,
respectively.

Axioms for S-norm

For the function S to be qualified as a fuzzy union, it must satisfy at least the following
four requirements:

Axiom S1: Boundary Condition

S[1, 1] = 1 (35)
S[µA (x), 0] = S[0, µA (x)] = µA (x) (36)

Axiom S2: Commutativity

S[µA (x), µB (x)] = S[µB (x), µA (x)] (37)

Axiom S3: Non-Decreasing

If µA (x) ≤ µB (x) and µC (x) ≤ µD (x) then S[µA (x), µC (x)] ≤ S[µB (x), µD (x)] (38)

Axiom S4: Associativity

S[S[µA (x), µB (x)], µC (x)] = S[µA (x), S[µB (x), µC (x)]] (39)
Any function S : [0, 1] × [0, 1] → [0, 1] that satisfies Axioms S1 to S4 is called an S-norm.

S-norm Operators
There are four commonly used S-norm operators:

5
Maximum
Smax (µA (x), µB (x)) = max(µA (x), µB (x)) = µA (x) ∨ µB (x) (40)

Algebraic Sum
Sas (µA (x), µB (x)) = µA (x) + µB (x) − (µA (x) × µB (x)) (41)

Bounded Sum
Sbs (µA (x), µB (x)) = min(1, µA (x) + µB (x)) (42)

Drastic Sum

µA (x), if µB (x) = 0

Sds (µA (x), µB (x)) = µB (x), if µA (x) = 0 (43)

1, if µA (x), µB (x) > 0


6
 Fuzzy Sets, Logic and Systems & Applications
Week 6: Lectures 26 - 30

Parameterized T-norm or S-conorm

Dombi’s Class of T-norm
Let A and B be two fuzzy sets with the universe of discourse X. The Dombi’s class of
T-norm is defined as follows:
1
Tλ (µA (x), µB (x)) = (1)
 λ  λ  λ1
1 1
1+ µA (x)
− 1 + µB (x) − 1

where, λ ∈ (0, ∞) and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

Dubois-Prade’s Class of T-norm

Let A and B be two fuzzy sets with the universe of discourse X. The Dubois-Prade’s
class of T-norm is defined as follows:

µA (x) × µB (x)
Tα (µA (x), µB (x)) = (2)
max(µA (x), µB (x), α)
where, α ∈ [0, 1] and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

Yager’s Class of T-norm

Let A and B be two fuzzy sets with the universe of discourse X. The Yager’s class of
T-norm is defined as follows:
h 1
i
Tw (µA (x), µB (x)) = 1 − min 1, ((1 − µA (x))w + (1 − µB (x))w ) w (3)

where, w ∈ (0, ∞) and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

1
Parameterized S-norm or T-conorm
Dombi’s Class of S-norm
Let A and B be two fuzzy sets with the universe of discourse X. The Dombi’s class of
S-norm is defined as follows:
1
Sλ (µA (x), µB (x)) = (4)
 −λ  −λ − λ1
1 1
1+ µA (x)
−1 + µB (x) − 1

where, λ ∈ (0, ∞) and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

Dubois-Prade’s Class of S-norm

Let A and B be two fuzzy sets with the universe of discourse X. The Dubois-Prade’s
class of S-norm is defined as follows:

µA (x) + µB (x) − µA (x) × µB (x) − min(µA (x), µB (x), (1 − α))

Sα (µA (x), µB (x)) = (5)
max((1 − µA (x)), (1 − µB (x)), α)

where, α ∈ [0, 1] and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

Yager’s Class of S-norm

Let A and B be two fuzzy sets with the universe of discourse X. The Yager’s class of
S-norm is defined as follows:
h 1
i
Sw (µA (x), µB (x)) = min 1, (µA (x)w + µB (x)w ) w (6)

where, w ∈ (0, ∞) and µA (x) and µB (x) denote the membership function values for fuzzy
sets A and B, respectively.

2
Fuzzy Relation
Cartesian Product of Crisp Sets
Let A and B be two arbitrary crisp sets with the universe of discourse X and Y , respec-
tively.
The Cartesian product of A and B is the crisp set of all ordered pairs (x, y) such that
x ∈ X and y ∈ Y . It is denoted by A × B and can be defined as:

A × B = {(x, y)|x ∈ X and y ∈ Y } (7)

Order of Cartesian Product

The order in which A and B appears is important i.e.,
• if A ̸= B, then A × B ̸= B × A.
Example, we have: A = {0, 1}; B = {a, b, c}
Then, the Cartesian product of crisp sets A and B will be

A × B = {(0, a), (0, b), (0, c), (1, a), (1, b), (1, c)} (8)

and the Cartesian product of crisp sets B and A will be

B × A = {(a, 0), (b, 0), (c, 0), (a, 1), (b, 1), (c, 1)} (9)

Hence,
A × B ̸= B × A (10)

Cartesian Product of Multiple Crisp Sets

In general, the Cartesian product of arbitrary n crisp sets A1 , A2 , A3 , . . . , An is the
crisp set of all n-tuples (x1 , x2 , x3 , . . . , xn ) such that xi ∈ Xi for i ∈ {1, 2, 3, . . . }. It can
be written as:

A1 × A2 × A3 . . . An = {(x1 , x2 , . . . , xn )|x1 ∈ X1 , x2 ∈ X2 , . . . , xn ∈ Xn } (11)

A relation among crisp sets A1 , A2 , A3 , . . . , An is a subset of the Cartesian product A1 ×
A2 × A3 . . . An . This means that we can use Q(A1 , A2 , A3 , . . . , An ) to denote a relation
among elements of these sets given below.

Q(A1 , A2 , A3 , . . . , An ) ⊂ A1 × A2 × A3 . . . An (12)
A relation set between crisp sets A and B is the subset of Cartesian product A × B.

Cartesian Product of Fuzzy Sets

Let A and B be two fuzzy sets on the universe of discourse X and Y , respectively. The
Cartesian product between fuzzy sets A and B will be represented by A × B.
The resulting fuzzy set can be written as:

A × B = {((x, y), µA×B (x, y)) | ∀(x, y) ∈ X × Y } (13)

3
The membership function values of the resulting fuzzy set can be defined as:

µA×B (x, y) = min(µA (x), µB (y)) (14)

Fuzzy Relation: Definition

Let us consider two fuzzy sets A and B with the universe of discourse X and Y , respec-
tively.
Fuzzy relation between two fuzzy sets A and B maps the elements of one universe of
discourse x ∈ X to another universe of discourse y ∈ Y through the Cartesian product of
respective universes of discourse, i.e., (x, y) ∈ X × Y , where (x, y) is the set of all ordered
pairs.
However, the “strength” of the relation between ordered pairs (x, y) is measured with a
membership function expressing various “degrees” of strength of the relation on the unit
interval [0, 1].

Fuzzy Relation: Mathematical Representation

A fuzzy relation Q is a mapping from the Cartesian space X × Y to the interval [0, 1],
where the strength of the mapping is expressed by the membership function of the relation
for ordered pairs from the two universes of discourse, or µQ (x, y).
It can be represented as:

Q(x, y) = {((x, y), µQ (x, y)) | ∀(x, y) ∈ X × Y } (15)

Fuzzy Relation for Multiple Fuzzy Sets

A fuzzy relation for n-fuzzy sets can be defined using the Cartesian product of fuzzy sets,
i.e., A1 × A2 × A3 . . . An .
A fuzzy relation Q in A1 × A2 × A3 . . . An is defined as:

Q = {((x1 , x2 , x3 , . . . , xn ), µQ (x1 , x2 , x3 , . . . , xn )) | (x1 , x2 , x3 , . . . , xn ) ∈ X1 ×X2 ×X3 . . . Xn }

(16)
where µQ : [0, 1].

4
Operations on Crisp and Fuzzy Relation
Union of Crisp Relations
Let R and S be the crisp relations defined on the space X ×Y . Then, the union is defined
by:

T =R∪S (17)
T is said to be the union of R and S:
If
∀(x, y) ∈ R or ∀(x, y) ∈ S (18)
Then
∀(x, y) ∈ T | ∀(x, y) ∈ X × Y (19)

Union of Fuzzy Relations

Let R and S be the fuzzy relations defined on the space X × Y . Then the union of R
and S is defined as:

T = {(x, y), µT (x, y) | ∀(x, y) ∈ X × Y } (20)

where
µT (x, y) = µR∪S (x, y) = max(µR (x, y), µS (x, y)) (21)

Intersection of Crisp Relations

Let R and S be the crisp relations defined on the space X × Y . Then, the intersection is
defined by:

T =R∩S (22)
T is said to be the intersection of R and S:
If
∀(x, y) ∈ R and ∀(x, y) ∈ S (23)
Then
∀(x, y) ∈ T | ∀(x, y) ∈ X × Y (24)

Intersection of Fuzzy Relations

Let R and S be the fuzzy relations defined on the space X × Y . Then, the intersection
is defined by:

T = {(x, y), µT (x, y) | ∀(x, y) ∈ X × Y } (25)

where
µT (x, y) = µR∩S (x, y) = min(µR (x, y), µS (x, y)) (26)

5
Complement of Crisp Relation
Let R be the crisp relation defined on the space X × Y . Then, the complement of relation
R is defined by:
If ∀(x, y) ∈
/ R, then ∀(x, y) ∈ R̄; i.e.,

R̄ = {(x, y) | ∀(x, y) ∈
/ R} (27)
where ∀(x, y) ∈ X × Y .

Complement of Fuzzy Relation

Let R be the fuzzy relation defined on the space X ×Y . Then, the complement of relation
R is defined by:

R̄ = {((x, y), µR̄ (x, y))} (28)

where

µR̄ (x, y) = 1 − µR (x, y) ∀(x, y) ∈ X × Y (29)

Containment of Crisp Relation

Let R and S be the crisp relations defined on the space X × Y . Then, the containment
is defined by:

T =R⊂S (30)
R is contained in S.
If
∀(x, y) ∈ R and ∀(x, y) ∈ S (31)
Then
R(x, y) ≤ S(x, y) | ∀(x, y) ∈ X × Y (32)

Containment of Fuzzy Relation

Let R and S be the fuzzy relations defined on the space X × Y . Then, the containment
is defined by:

T =R⊂S (33)
R is contained in S.
If
∀(x, y) ∈ R and ∀(x, y) ∈ S (34)
Then
µR (x, y) ≤ µS (x, y) | ∀(x, y) ∈ X × Y (35)

6
 Fuzzy Sets, Logic and Systems & Applications
Week 7: Lectures 31 - 35

Projection of Fuzzy Relation Set

Let us consider a fuzzy relation R ⊆ A × B between fuzzy sets A and B with the universe
of discourse X and Y , respectively. The fuzzy relation R can be defined as,

R = A × B = {((x, y), µR (x, y)) | ∀(x, y) ∈ X × Y } (1)

The projection of relation R onto A is denoted by RA , and the projection of relation R
onto B is denoted by RB . The membership values of the projection of fuzzy relation R
can be found as follows:

Projection of R on A:
µRA (x) = max µR (x, y) (2)
y

Projection of R on B:
µRB (y) = max µR (x, y) (3)
x

The fuzzy relation R(x, y) and its projections can also be represented as integrals:
Z
R(x, y) = µR (x, y) d(x, y) (4)
X×Y
Z
RA (x) = µRA (x) dx (5)
X
Z
RB (y) = µRB (y) dy (6)
Y
The projection of a fuzzy relation can be extended to n-dimension. Let relation R be
defined among fuzzy sets A1 , A2 , A3 , . . . , An in the space X1 × X2 × X3 . . . Xn . Projecting
this relation to X1 × X2 × X3 · · · × Xk gives a projected relation as,

RA1 ×A2 ×A3 ···×Ak (7)

and the membership values of the projected relation can be defined as,

µRA1 ×A2 ×A3 ···×Ak (x1 , x2 , x3 , . . . , xk ) = max µR (x1 , x2 , x3 . . . , xn ) (8)

X1 ,X2 ,X3 ...,Xm

∀(x1 , x2 , x3 , . . . , xn ) ∈ X1 × X2 × X3 · · · × Xn (9)

1
where, k + m = n.
Here, X1 , X2 , X3 . . . , Xm represent the omitted dimensions, and X1 , X2 , X3 . . . , Xk is the
remaining dimensions. It can be concluded as,

{X1 , X2 , X3 . . . , Xn } = {X1 , X2 , X3 . . . , Xk } ∪ {X1 , X2 , X3 . . . , Xm } (10)

Cylindrical Extension of one-dimensional Fuzzy Set

If A is a fuzzy set with the universe of discourse X, then its cylindrical extension in space
X × Y is a fuzzy set C(A) defined by,

For continuous case:

Z
C(A) = µC(A) (x, y)/(x, y) (11)
X×Y
Z
A= µA (x)/x (12)
X

For discrete case:

X
C(A) = µC(A) (x, y)/(x, y) (13)
X×Y
X
A= µA (x)/x (14)
X

where A is referred to as a base set.

Here, Y is the universe of discourse for the second dimension of the cylindrical extension.
The membership function values of the fuzzy set C(A) are found by:

µC(A) (x, y) = µA (x), ∀(x, y) ∈ X × Y (15)

Cylindrical Extension of a Fuzzy Relation

Let us consider two fuzzy sets A and B with the universe of discourse X and Y , respec-
tively. A fuzzy relation R defined in the space X × Y can be extended to obtain a new
fuzzy set C(R), which can be represented as,


C(R) = (x, y, z), µC(R) (x, y, z) | ∀(x, y, z) ∈ X × Y × Z (16)
The membership values of C(R) can be found as,

µC(R) (x, y, z) = µR (x, y), ∀(x, y, z) ∈ X × Y × Z (17)

where Z is the universe of discourse for the third dimension of the cylindrical extension.

2
Properties of Crisp Sets and Fuzzy Sets
If we compare the properties of crisp sets with fuzzy sets, we observe that the following
three properties do not hold for fuzzy sets:

• Law of Contradiction

• Law of Excluded Middle

• Absorption of Complement

Figure 1: Properties of Crisp and Fuzzy Sets

Properties of Crisp Relations and Fuzzy Relations

Similarly, if we compare the properties of crisp relations with fuzzy relations, the following
three properties do not hold for fuzzy relations:

• Law of Contradiction

• Law of Excluded Middle

• Absorption of Complement

3
Figure 2: Properties of Crisp and Fuzzy Relations

4
Extension Principle
The extension principle is a basic concept of fuzzy set theory that provides a general
procedure for transforming a fuzzy set from one universe of discourse to another universe
of discourse, provided we have point-to-point mapping of a function f (.) to a mapping
between known.
This procedure generalizes a common point-to-point mapping of a function f (.) to a
mapping between fuzzy sets. More specifically, suppose that f is a function from X to
Y and A is a fuzzy set with the universe of discourse X defined as,

µA (x1 ) µA (x2 ) µA (xn )

A(x) = + + ··· + (18)
x1 x2 xn
Then, the extension principle states that the image of fuzzy set A under the mapping
f (.) can be expressed as a fuzzy set B as,

µA (x1 ) µA (x2 ) µA (xn )

B(y) = + + ··· + (19)
y1 y2 yn
where yi = f (xi ) or xi = f −1 (yi ); ∀i = 1, . . . , n.
If f (.) is a many-to-one mapping then there exist x1 , x2 ∈ X, x1 ̸= x2 , such that

f (x1 ) = f (x2 ) = y ∗ , y∗ ∈ Y (20)

In this case, the membership value of fuzzy set B at y = y ∗ will be:

µB (y ∗ ) = max(µA (x1 ), µA (x2 )) (21)

More generally, we have

µB (y) = max
−1
µA (x) (22)
x∈f (y)

where f −1 (y) denotes the set of all points in the universe of discourse x ∈ X such that
f (x) = y.
This is called the extension principle.
Suppose that function f is a mapping from n-dimensional Cartesian product space X1 ×
X2 × · · · × Xn to a one-dimensional universe of discourse Y such that y = f (x1 , . . . , xn ),
and suppose A1 , . . . , An are n fuzzy sets with the universe of discourse X1 , . . . , Xn , re-
spectively.
Then, the extension principle asserts that the membership values of fuzzy set B induced
by the mapping f are defined by,
(
max(xi1 ,xi2 ,...,xin )=f −1 (y) [mini µAi (xi )] , if f −1 (y) ̸= ∅
µB (y) = (23)
0, if f −1 (y) = ∅

5
 Fuzzy Sets, Logic and Systems & Applications
Week 8: Lectures 36 - 40

Composition of Fuzzy Relations

Fuzzy relations in different product spaces can be combined through a composition op-
eration. There are mainly two types of composition operations for fuzzy relations:

• Max-min Composition

• Max-product Composition

These compositions help determine the relationship between elements of different fuzzy
sets by defining new membership values based on specific operations.

Max-min Composition
Let R1 (x, y) and R2 (y, z) be two fuzzy relations defined on the product spaces X × Y
and Y × Z, respectively. The max-min composition of R1 and R2 results in a fuzzy
set defined as:

R1 ◦ R2 = {(x, z), µR1 ◦R2 (x, z) | ∀(x, z) ∈ X × Z} (1)

The membership values of R1 ◦ R2 are defined as:

µR1 ◦R2 (x, z) = max min [µR1 (x, y), µR2 (y, z)] |∀(x, y) ∈ X × Y and ∀(y, z) ∈ Y × Z (2)
y

Alternatively, using logical operators:

_
µR1 ◦R2 (x, z) = [µR1 (x, y) ∧ µR2 (y, z)] (3)
y

where:

• ∧ represents the min operator.

• ∨ represents the max operator.

1
Interpretation
• If R1 (x, y) and R2 (y, z) are represented as fuzzy relation matrices, their composition
R1 ◦ R2 is similar to matrix multiplication but replaces multiplication (×) and
addition (+) with the min and max operators, respectively.

• Due to this similarity, the max-min composition is also called the max-min prod-
uct.

• This is the best-known composition proposed by Prof. L. A. Zadeh.

Max-product Composition
Similar to max-min composition, let R1 (x, y) and R2 (y, z) be two fuzzy relations defined
on X × Y and Y × Z, respectively. The max-product composition of R1 and R2
results in a fuzzy set:

R1 ◦ R2 = {(x, z), µR1 ◦R2 (x, z) | ∀(x, z) ∈ X × Z} (4)

The membership values of R1 ◦ R2 are given by:

µR1 ◦R2 (x, z) = max [µR1 (x, y) × µR2 (y, z)] |∀(x, y) ∈ X × Y and ∀(y, z) ∈ Y × Z (5)
y

Alternatively, using logical notation:

_
µR1 ◦R2 (x, z) = [µR1 (x, y) · µR2 (y, z)] (6)
y

where ∨ is the max operator.

2
Properties of Composition of Fuzzy Relations
The composition of fuzzy relations holds the following properties:

Table 1: Properties of Fuzzy Relation Composition

PROPERTIES COMPOSITION OF FUZZY RELATIONS
Associativity (R ◦ S1 ) ◦ T = R ◦ (S1 ◦ T )
Distributivity over Union R ◦ (S1 ∪ S2 ) = (R ◦ S1 ) ∪ (R ◦ S2 )
Weak Distributivity over Intersection R ◦ (S1 ∩ S2 ) ⊆ (R ◦ S1 ) ∩ (R ◦ S2 )
Monotonicity S1 ⊆ S2 ⇒ R ◦ S1 ⊆ R ◦ S2

Here, R, S1 , S2 , and T are fuzzy relations defined on the spaces X × Y, Y × Z, Y × Z, and

Z × W , respectively.

3
Fuzzy Tolerance Relations
Let R be a fuzzy relation defined in the space X × X such that R ⊆ X × X, then R will
be a fuzzy tolerance relation if it satisfies the following two properties.
1. Reflexivity:
µR (xi , xi ) = 1.0, ∀xi ∈ X. (7)
2. Symmetry:
µR (xi , xj ) = µR (xj , xi ), ∀xi , xj ∈ X. (8)
where, µR (xi , xi ), µR (xi , xj ), and µR (xj , xi ) are the membership values of the elements
(xi , xi ), (xi , xj ), and (xj , xi ), respectively, for the fuzzy relation R.

Fuzzy Equivalence Relations

Let R be a fuzzy relation defined in the space X × X such that R ⊆ X × X. It is known
as a fuzzy equivalence relation if all three of the following properties are satisfied:
1. Reflexivity:
µR (xi , xi ) = 1.0, ∀xi ∈ X. (9)
2. Symmetry:
µR (xi , xj ) = µR (xj , xi ), ∀xi , xj ∈ X. (10)
3. Transitivity:

µR (xi , xj ) = λ1 and µR (xj , xk ) = λ2 ⇒ µR (xi , xk ) = λ, ∀xi , xj , xk ∈ X. (11)

where, λ ≥ min[λ1 , λ2 ].
Here, µR (xi , xj ), µR (xj , xk ), and µR (xi , xk ) are the membership values of the elements
(xi , xj ), (xj , xk ), and (xi , xk ), respectively, for the fuzzy relation R.
A fuzzy tolerance relation R defined in the space X × X, that has properties of reflexivity
and symmetry, can be reformed into a fuzzy equivalence relation by at most (n − 1)
compositions with itself, where n is the cardinal number of the set defining R. It is given
below.

Rn−1 = R ◦ R ◦ R · · · ◦ R = R′ . (12)

4
Some more properties
(i) Anti-reflexivity
Let R be a fuzzy relation defined in the space X × X such that R ⊆ X × X, then R will
satisfy the anti-reflexivity property if

µR (xi , xi ) = 0, ∀xi ∈ X. (13)

where, µR (xi , xi ) is the membership value of the element (xi , xi ) for the fuzzy relation R.
(ii) Anti-symmetry
Let R be a fuzzy relation defined in the space X × X such that R ⊆ X × X, then R will
satisfy the anti-symmetry property if

if µR (xi , xj ) > 0, then µR (xj , xi ) = 0, ∀xi , xj ∈ X, xi ̸= xj . (14)

where, µR (xi , xj ) and µR (xj , xi ) are the membership values of the elements (xi , xj ) and
(xj , xi ), respectively, for the fuzzy relation R.

5
 Fuzzy Sets, Logic and Systems & Applications
Week 9: Lectures 41 - 45

Linguistic Hedges
A linguistic variable is characterized by a quintuple as,

(x, T (x), X, G, M ) (1)

Here,

• x is the name of the linguistic variable.

• T (x) is the term set of x, i.e., the set of its linguistic values or linguistic terms.

• X is the universe of discourse.

• G is a syntactic rule which generates the term in T (x).

• M is a semantic rule which associates with each linguistic value x, i.e., M (x), where
M (x) denotes a fuzzy set with the universe of discourse X.

For a linguistic variable “age”,

• Linguistic variable (x): “age” is a linguistic variable.

• Term set (T (x)): If “age” is interpreted as a linguistic variable, then its term set
T (age) could be as follows:

T (age) = {young, middle aged, old} (2)

The “age is young” denotes the assignment of the linguistic value “young” to the
linguistic variable “age”.

• Universe of discourse (X): In the term set T (age), each term can be character-
ized by a fuzzy set of the universe of discourse X = [0, 100].

• Syntactic rule (G): The syntactic rule refers to the way the linguistic values in
the term set T (age) are generated. In the figure below, “young”, “middle aged”,
etc., are the syntactic labels.

• Semantic rule (M ): The semantic rule specifies the procedure for computing the
meaning of any linguistic value through specified membership functions.

1
For linguistic variables, we use words as values of linguistic variables. In the case of
linguistics, we often use more than one word to describe a variable.
For example:
If “the intensity of light” is a linguistic variable, then its linguistic values might be “very
bright,” “slightly dim,” and “more or less bright,” etc.
So, the values of a linguistic variable may be a composite term and can be classified into
three groups:
1. Primary Terms
2. Linguistic Hedges
3. Negation / Complement and Connectives

Primary Terms
• For instance, if “age” is a linguistic variable, then its primary terms could be young,
middle aged, old, etc.
• Each primary term can be defined by a fuzzy set.
• For example, young can be expressed by a fuzzy set, and middle aged and old can
also be expressed by other fuzzy sets.

Linguistic Hedges
• In linguistics, fundamental atomic terms are often modified with adjectives (nouns)
or adverbs (verbs) such as very, low, slight, more or less, fairly, slightly, almost,
barely, mostly, roughly, approximately, etc.
• These modifiers are known as linguistic hedges, i.e., the singular meaning of an
atomic term is modified or hedged from its original interpretation.
Let us consider a fuzzy set A with the universe of discourse X defined as
Z
For continuous A = µA (x)/x (3)
x∈X
X
For discrete A = µA (x)/x (4)
x∈X

Membership Functions for Linguistic Hedges

i. The membership values for the linguistic term “Very or Too” is defined as,
µV ery A (x) = µT oo A (x) = [µA (x)]2 (5)

ii. The membership values for the linguistic term “More or Less” is defined as,
1
µM ore or Less A (x) = [µA (x)] 2 (6)

iii. The membership values for the linguistic term “Extremely Or Very Very Very” is
defined as,
µExtremely A (x) = µV ery V ery V ery A (x) = [µA (x)]8 (7)

2
Linguistic Hedges and Negation/ Complement and Connectives
In the case of linguistic terms, we can interpret the negation operator NOT and the
connectives AND and OR as defined below:

Z
N OT (A) = ¬A = [1 − µA (x)]/x (1)
x∈X
Z
A AND B = A ∩ B = [µA (x) ∧ µB (x)]/x (2)
x∈X
Z
A OR B = A ∪ B = [µA (x) ∨ µB (x)]/x (3)
x∈X

where, A and B are the linguistic values with the membership values µA (·) and µB (·).

3
Concentration
Let A be a linguistic value characterized by a fuzzy set with the membership value µA (·).
Then A(k) is interpreted as the concentration of the original linguistic value or a fuzzy
set A and is expressed as:
Z
(k)
A = [µA (x)]k /x (For continuous) (1)
X
X
A(k) = [µA (x)]k /x (For discrete) (2)
X

Normally, the concentration is defined as:

Z
(2)
CON (A) = A = [µA (x)]2 /x (For continuous) (3)
X
X
CON (A) = A(2) = [µA (x)]2 /x (For discrete) (4)
X

Dilation
Let A be a linguistic value characterized by a fuzzy set with the membership value µA (·).
Then A(k) is interpreted as the dilation of the original linguistic value or a fuzzy set A
and is expressed as:
Z
(k)
A = [µA (x)]k /x (For continuous) (5)
X
X
A(k) = [µA (x)]k /x (For discrete) (6)
X

Normally, the dilation is defined as:

Z
(0.5)
DIL(A) = A = [µA (x)]0.5 /x (For continuous) (7)
X
X
DIL(A) = A(0.5) = [µA (x)]0.5 /x (For discrete) (8)
X

Composite Linguistic Term

Composite linguistic terms can be formed from one or more combinations of primary
terms, logical connectives, and linguistic hedges. Some examples of composite linguistic
terms are as follows:
• not very young and not very old
• young but not too young
• middle aged or old
• young and old

4
 Fuzzy Sets, Logic and Systems & Applications
Week 10: Lectures 46 - 50

Contrast Intensification
Let A be a linguistic value characterized by a fuzzy set with the membership value µA (·)
as:

Z
A= µA (x)/x (For continuous) (1)
X
X
A= µA (x)/x (For discrete) (2)
X

Then, the contrast intensification on a linguistic value A is defined by:

(
2A(2) , for 0 ≤ µA (x) ≤ 0.5, ∀x ∈ X
IN T (A) = (2)
(3)
¬2(¬A) , for 0.5 ≤ µA (x) ≤ 1, ∀x ∈ X
The contrast intensifier IN T increases the values of µA (x), which are above 0.5, and
diminishes those which are below this point. Thus, contrast intensification has the effect
of reducing the fuzziness of linguistic value A. The inverse operator of the contrast
intensifier is contrast diminisher DIM .

Orthogonality
A term set T (x) = t1 (x), . . . , tn (x) of a linguistic variable x on the universe of discourse
X is orthogonal if it satisfies the following property:
n
X
µti (x) = 1, ∀x ∈ X (1)
i=1

where, the fuzzy sets ti (x) are convex and normal fuzzy sets defined on X and these fuzzy
sets make up the term set T (x).

1
Fuzzy If-Then Rule
A fuzzy if-then rule, also known as a fuzzy rule, fuzzy implication, or fuzzy conditional
statement, assumes the form:

IF x is A THEN y is B (1)
where, A and B are linguistic values characterized by fuzzy sets with the universe of
discourse X and Y , respectively.
Often, ”x is A” is called the antecedent or premise, while ”y is B” is called the
consequence or conclusion.

Examples
• If the temperature is very cold, then stop the air conditioner.

• If the road is good, then driving is smooth.

• If the mango is yellow, then the mango is sweet.

• If speed is high, then apply the brake a little.

Fuzzy Rule Interpretation

A fuzzy if-then rule ”IF x is A THEN y is B” can also be abbreviated as:

A→B (2)
The above expression describes a relation between two linguistic values, A and B. There
are two ways to interpret a fuzzy if-then rule:

1. A coupled with B

2. A entails B

This suggests that a fuzzy if-then rule can be defined as a fuzzy relation R on the space
X × Y , as:

For continuous:
Z
R=A→B =A×B = µR (x, y)/(x, y), ∀x, y ∈ X × Y (3)
X×Y

For discrete:
X
R=A→B =A×B = µR (x, y)/(x, y), ∀x, y ∈ X × Y (4)
X×Y

2
Fuzzy Rule Interpretation as A coupled with B
If A → B is interpreted as A coupled with B, then it can be interpreted by a fuzzy relation
R as follows:

For continuous:
Z
R=A→B =A×B = T [µA (x), µB (y)]/(x, y), ∀x, y ∈ X × Y
X×Y

For discrete:
X
R=A→B =A×B = T [µA (x), µB (y)]/(x, y), ∀x, y ∈ X × Y
X×Y

where, A → B represents the fuzzy relation R and T is the T-norm operator.

Hence, there are four different fuzzy relations that can be defined using four commonly
used T-norm operators as follows:

(a) A coupled with B using minimum T-norm operator

(b) A coupled with B using algebraic product T-norm operator

(c) A coupled with B using bounded product T-norm operator

(d) A coupled with B using drastic product T-norm operator

(a) A coupled with B using minimum T-norm operator

For continuous:
Z
Rmin = A → B = A × B = (µA (x) ∧ µB (y))/(x, y), ∀x, y ∈ X × Y
X×Y

For discrete:
X
Rmin = A → B = A × B = (µA (x) ∧ µB (y))/(x, y), ∀x, y ∈ X × Y
X×Y

(b) A coupled with B using algebraic product T-norm operator

For continuous:
Z
Rap = A → B = A × B = (µA (x) × µB (y))/(x, y), ∀x, y ∈ X × Y
X×Y

For discrete:
X
Rap = A → B = A × B = (µA (x) × µB (y))/(x, y), ∀x, y ∈ X × Y
X×Y

3
(c) A coupled with B using bounded product T-norm operator
For continuous:
Z
Rbp = A → B = A × B = (0 ∨ (µA (x) + µB (y) − 1))/(x, y), ∀x, y ∈ X × Y
X×Y

For discrete:
X
Rbp = A → B = A × B = (0 ∨ (µA (x) + µB (y) − 1))/(x, y), ∀x, y ∈ X × Y
X×Y

(d) A coupled with B using drastic product T-norm operator

For continuous:
Z
Rdp = A → B = A × B = µRdp (x, y)/(x, y), ∀x, y ∈ X × Y
X×Y

For discrete:
X
Rdp = A → B = A × B = µRdp (x, y)/(x, y), ∀x, y ∈ X × Y
X×Y

where, 
µA (x) if µB (y) = 1

µRdp (x, y) = µB (y) if µA (x) = 1

0 otherwise


4
Fuzzy Rule Interpretation as A entails B
If A → B is interpreted as A extbfentails B, then it can be defined by the following four
different forms:

(a) Material implication:

Rmi = A → B = ¬A ∪ B (5)

(b) Propositional calculus:

Rpc = A → B = ¬A ∪ (A ∩ B) (6)

(c) Extended propositional calculus:

Repc = A → B = (¬A ∩ ¬B) ∪ B (7)

(d) Generalization of modus ponens:

Rgmp = A → B = A ≲ B (8)

where, A → B represents the fuzzy relation R.

(a) Material Implication

For continuous:
Z
Rmi = A → B = ¬A ∪ B = (1 ∧ (1 − µA (x) + µB (y))) /(x, y), ∀x, y ∈ X × Y (9)
X×Y

For discrete:
X
Rmi = A → B = ¬A ∪ B = (1 ∧ (1 − µA (x) + µB (y))) /(x, y), ∀x, y ∈ X × Y (10)
X×Y

Zadeh’s arithmetic rule, which follows ¬A ∪ B using the bounded sum operator for
union.

(b) Propositional Calculus

For continuous:
Z
Rpc = A → B = ¬A∪(A∩B) = ((1 − µA (x)) ∨ (µA (x) ∧ µB (y))) /(x, y), ∀x, y ∈ X×Y
X×Y
(11)
For discrete:
X
Rpc = A → B = ¬A∪(A∩B) = ((1 − µA (x)) ∨ (µA (x) ∧ µB (y))) /(x, y), ∀x, y ∈ X×Y
X×Y
(12)
Zadeh’s max-min rule, which follows ¬A ∪ (A ∩ B) using min for ∩ and max for ∪.

5
(c) Extended Propositional Calculus
For continuous:
Z
Repc = A → B = (¬A∩¬B)∪B = (((1 − µA (x)) ∧ (1 − µB (y))) ∨ µB (x)) /(x, y), ∀x, y ∈ X×Y
X×Y
(13)
For discrete:
X
Repc = A → B = (¬A∩¬B)∪B = (((1 − µA (x)) ∧ (1 − µB (y))) ∨ µB (x)) /(x, y), ∀x, y ∈ X×Y
X×Y
(14)
Boolean fuzzy implication, using max for ∪.

(d) Generalization of Modus Ponens

For continuous:
Z
Rgmp = A → B = A ≲ B = µRgmp (x, y)/(x, y), ∀x, y ∈ X × Y (15)
X×Y

For discrete:
X
Rgmp = A → B = A ≲ B = µRgmp (x, y)/(x, y), ∀x, y ∈ X × Y (16)
X×Y

where,
(
1, if µA (x) ≤ µB (y)
µRgmp (x, y) = µA (x) ≲ µB (y) = (17)
µB (y)/µA (x), if µA (x) > µB (y)

Goguen’s fuzzy implication, which follows A ≲ B using the algebraic product for the
T-norm operator.

6
Fuzzy Inference System
• Fuzzy inference is the process of mapping from a given input to an output using
fuzzy logic. It involves all the pieces discussed in the previous sections: membership
functions, fuzzy logic operators, and if-then rules.

What is a Fuzzy Inference System (FIS)?

• A nonlinear mapping that derives its output based on fuzzy reasoning and a set of
fuzzy if-then rules. The domain and range of the mapping could be fuzzy sets or
points in multidimensional spaces.

• Fuzzy inference systems have been successfully applied in various fields such as
automatic control, data classification, decision analysis, expert systems, computer
vision, etc.

• Because of its multidisciplinary nature, a fuzzy inference system is also known as:

– Fuzzy-rule based system

– Fuzzy expert system
– Fuzzy model
– Fuzzy associate memory
– Fuzzy logic controller
– Fuzzy system

Fuzzy Reasoning Process

The steps of fuzzy reasoning (inference operations upon fuzzy IF–THEN rules) performed
by FISs are:
1. Compare the input variables with the membership functions on the antecedent part
to obtain the membership values of each linguistic label. (This step is often called
fuzzification.)

2. Combine (usually multiplication or min) the membership values on the premise

part to get firing strength (degree of fulfillment) of each rule.

3. Generate the qualified consequents (either fuzzy or crisp) or each rule depending
on the firing strength.

4. Aggregate the qualified consequents to produce a crisp output. (This step is called
defuzzification.)

Fuzzy Knowledge Base

The rule base is referred to as the knowledge base.
• A rule base contains a number of fuzzy IF-THEN rules;

7
 • A database that defines the membership functions of the fuzzy sets used in the
fuzzy rules.

Fuzzifier
Converts the crisp input to a linguistic variable using the membership functions stored
in the fuzzy knowledge base.

Defuzzifier
It converts the fuzzy output of the inference engine to crisp using membership functions
analogous to the ones used by the fuzzifier.
Five commonly used defuzzification methods are as follows:

• Centroid of area (COA)

• Bisector of area (BOA)

• Mean of maximum (MOM)

• Smallest of maximum (SOM)

• Largest of maximum (LOM)

Fuzzy Reasoning
Fuzzy reasoning is also known as approximate reasoning. It is an inference procedure
that derives conclusions from a set of fuzzy if-then rules and known facts.

Compositional Rule of Inference

This idea is proposed by Prof. Zadeh. A similar concept has been used for the max-min
composition of fuzzy relation sets. Moreover, the extension principle is actually a special
case of the compositional rule of inference.

Computational Aspects of Fuzzy Reasoning

Now, we will discuss the computational aspects of fuzzy reasoning using the following:

(i) Single Rule with Single Antecedent

(ii) Single Rule with Multiple Antecedents

(iii) Multiple Rules with Multiple Antecedents

8
 Fuzzy Sets, Logic and Systems & Applications
Week 11: Lectures 51 - 55

Mamdani Fuzzy Model

In 1975, Prof. E. H. Mamdani built one of the first fuzzy systems to control a steam
engine and boiler combination using a set of linguistic control rules obtained from
human operators.

Steps in Mamdani Fuzzy Inference System

The common steps to compute the output of a fuzzy inference system are as follows:

• Determining the set of fuzzy rules.

• Fuzzifying the inputs using input membership functions.

• Combining the fuzzified inputs according to the fuzzy rules to establish a rule
strength.

• Finding the consequence of the rule by combining the rule strength and output
membership function.

• Combining the consequences to get the overall consequent i.e., aggregation.

• Defuzzifying the output distribution (this step is required only if a crisp output
is needed).

Fuzzy Rule Format

Rule: IF x is A THEN y is B

Mamdani Fuzzy Inference Process

The Mamdani fuzzy inference process consists of the following four steps:

1. Fuzzification of the input variables:

• Fuzzification is the process of converting a crisp quantity into a fuzzy quantity.

2. Rule evaluation:

1
 • The next step is to take the fuzzified inputs and apply the antecedents of the
fuzzy rules.
• If a given fuzzy rule has multiple antecedents, the fuzzy operator (AND or
OR) is used to obtain a single number representing the result of the antecedent
evaluation. This number (the truth value) is then applied to the consequent
membership function.

3. Aggregation of the outputs of fuzzy rules:

• Aggregation is the process of unification of the outputs of all the rules. The
input of the aggregation process is the list of clipped or scaled consequent
membership functions, and the output is one fuzzy set for each output variable.

4. Defuzzification:

• The input for the defuzzification process is the aggregated output fuzzy sets.
• If only crisp values are needed, a defuzzifier is used to convert a fuzzy set to a
crisp value.

Mamdani Fuzzy Model Compositions

The Mamdani Fuzzy Model uses Max-Min and Max-Product compositions for Fuzzy
and Crisp inputs:

• Single Rule with Single Antecedent

• Single Rule with Multiple Antecedents

• Multiple Rules with Multiple Antecedents

2
Defuzzification
Defuzzification is the process of converting a fuzzy output set into a crisp value. Several
defuzzification methods exist, including:
• Centroid of Area (COA)

• Bisector of Area (BOA)

• Mean of Maximum (MOM)

• Smallest of Maximum (SOM)

• Largest of Maximum (LOM)

Centroid of Area (COA)

R
µA (z)z dz
zCOA = Rz (1)
µ (z) dz
z A
where µA (z) is the aggregated output membership function. This is the most widely
adopted defuzzification strategy.

Bisector of Area (BOA)

Z zBOA Z β
µA (z)dz = µA (z)dz (2)
α zBOA

where, α = min{z|z ∈ Z} and β = max{z|z ∈ Z}. The vertical line z = zBOA partitions
the output region between z = α and z = β into two regions with the same area.

Mean of Maximum (MOM)

The mean of maximum is calculated as:
R
′ zdz
zM OM = Rz (3)
z′
dz

where, z ′ = {z|µA (z) = µ∗ }. If µA (z) has a single maximum at z = z ∗ , then zM OM = z ∗ .

Smallest of Maximum (SOM)

zSOM is the minimum value of z for which the membership function has its maximum value.
(4)

Largest of Maximum (LOM)

zLOM is the maximum value of z for which the membership function has its maximum value.
(5)

3
 Fuzzy Sets, Logic and Systems & Applications
Week 12: Lectures 56 - 60

Larsen Fuzzy Model

If a fuzzy rule base with n rules for input membership functions Ai and Bi with the
universe of discourse X and Y , respectively, and the output membership function Ci
with the universe of discourse Z is defined as:
IF x is Ai AND/OR y is Bi THEN z is Ci
where i = 1, 2, 3, . . . , n and fuzzy sets Ai , Bi , and Ci are expressed as:
Z Z Z
Ai = µAi (x)/x; Bi = µBi (y)/y; Ci = µCi (z)/z (1)
x∈X y∈Y z∈Z

Firing Strengths
The firing strength of the ith -rule is defined as:
wi = µAi (x) ∧ µBi (y) ⇒ For max-min composition (2)
wi = µAi (x) × µBi (y) ⇒ For max-product composition (3)

Fuzzy Output Membership Function

n
_
µ (z) =
C′ (wi × µCi (z)) (4)
i=1

Final Output
Z
′
C = µC ′ (z)/z (5)
z∈Z
The fuzzy output C ′ can be further defuzzified to obtain a crisp value.

Larsen Fuzzy Model Compositions

The Larsen Fuzzy Model uses Max-Min Composition and Max-Product Compo-
sition for Fuzzy and Crisp Inputs:
• Single Rule with Single Antecedent
• Single Rule with Multiple Antecedents
• Multiple Rules with Multiple Antecedents

1
Tsukamoto Fuzzy Model
In the Tsukamoto fuzzy model, the consequent of each fuzzy if-then rule is represented
by a fuzzy set with a monotonically (increasing/decreasing) membership function.
As a result, the inferred output of each rule is a crisp value corresponding to the firing
strength w of that rule. The overall output is taken as the weighted average of the output
of each rule.
Tsukamoto fuzzy model avoids the time-consuming process of defuzzification.
If a fuzzy rule base with n rules for input membership functions Ai and Bi with the
universe of discourse X and Y , respectively, and the output membership function Ci
with the universe of discourse Z is defined as:

IF x is Ai AND/OR y is Bi THEN z is Ci

where i = 1, 2, 3, . . . , n and fuzzy sets Ai , Bi , and Ci are expressed as:

Z Z Z
Ai = µAi (x)/x; Bi = µBi (y)/y; Ci = µCi (z)/z (6)
x∈X y∈Y z∈Z

The firing strength of the ith -rule is defined by,

wi = µAi (x) ∧ µBi (y) ⇒ µCi (zi ) = wi (7)

The overall output is taken as the weighted average of each rule’s output as follows:
Pn
∗ i=1 wi × zi
z = P n (8)
i=1 wi

where zi is the output of each rule induced by the firing strength wi and the output
membership function Ci .
Since the reasoning mechanism of the Tsukamoto fuzzy model doesn’t strictly follow the
compositional rule of inference, the output is always crisp, even when the inputs are fuzzy.

Takagi-Sugeno-Kang (TSK) Fuzzy Model

The TSK fuzzy model was proposed by Takagi, Sugeno, and Kang in 1985 to develop
a systematic approach for generating fuzzy rules from a given input-output dataset.
A typical fuzzy rule in a TSK fuzzy model has the form:

IF x is A AND/OR y is B THEN z = f (x, y)

where A and B are fuzzy sets in the antecedent, while z = f (x, y) is a crisp function in
the consequent.
Usually, f (x, y) is a polynomial, but it can be any function as long as it can appropriately
describe the output of the model within the fuzzy region specified by the antecedent of
the fuzzy rule.

• TSK fuzzy model takes comparatively minimum computation time.

• TSK fuzzy model takes only crisp values as inputs.

2
Since each rule has a crisp output, the overall output is obtained via weighted average,
thus avoiding the time-consuming process of defuzzification required in a Mamdani fuzzy
model.
Classification of TSK Models:

• When f (x, y) is a first-order polynomial, the resulting fuzzy inference system is

called a First-Order TSK Fuzzy Model.

• When f (·) is a constant, we have a Zero-Order TSK Fuzzy Model.

The Zero-Order TSK Fuzzy Model is a special case of the Mamdani fuzzy model in
which the consequent of each rule is specified by a fuzzy singleton (or a pre-defuzzified
consequent) and a special case of the Tsukamoto fuzzy model, in which the consequent
of each rule is specified by a membership function of a step function centered at the
constant.
If a fuzzy rule base with n rules for input membership functions Ai and Bi with the
universe of discourse X and Y , respectively, is defined as:

IF x is Ai AND/OR y is Bi THEN zi = f (x, y)

where i = 1, 2, 3, . . . , n and fuzzy sets Ai and Bi are expressed as:

Z Z
Ai = µAi (x)/x; Bi = µBi (y)/y (9)
x∈X y∈Y

The firing strength of the ith -rule is defined by,

wi = µAi (x) ∧ µBi (y) (10)

The overall output is taken as the weighted average of each rule’s output as follows:
Pn
∗ i=1 wi × zi
z = P n (11)
i=1 wi

where zi is a polynomial in the input variables x and y.