Fuzzy Logic System

Membership function with discrete membership

values

The membership values may be of discrete values.


A

notation if universe is discrete[Zadeh]

A = μA(x1)/x1 + μA(x2)/x2 + . . . + μA(xi)/xi + . .

..

20 / 69
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)}

1.0
0.8
Note : X = discrete value
0.6
µ

0.4

0.2
How you measure happiness ??
0 2 4 6 8 10

Number of children (X)

A = “Happy family”

21 / 69
Membership function with continuous
membership values

1.0
0.8
 B (x) = 1
 x −50 
4
1+  
0.6  10 

0.4

0.2

0 50 100
B

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

If the universe is continuous

22 / 69
Fuzzy terminologies: Support
• Support: The support of a fuzzy set A is the set of all points x ∈ X
such that µA (x ) > 0, - is a crisp subset of the universe
• Leave off all elements that have a zero membership grade

23 / 69
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}

1.0
µ

0.5

24 / 69
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.

1.0

25 / 69
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}.

26 / 69
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}.

27 / 69
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
}

Strong α-cut is defined similarly :

Aα ’ = {x | µA (x) > α
}
Note : Support(A) = A0’ and Core(A) = A1.

28 / 69
Fuzzy terminologies: Convexity
Convexity : A fuzzy set A is convex if and only if for any x1 and x2 ∈ X
and any λ ∈ [0, 1]
µA (λx1 + (1 -λ)x2 ) ≥ min(µA(x1 ), µA(x2 ))

Note :
• A is convex if all its α- level sets are convex.
• Convexity (Aα ) =⇒ Aα is composed of a single line segment only.
Membership function is Non-convex
convex Membership function

1.0 1.0

29 / 69
Fuzzy terminologies: Height

Height:

In A fuzzy set A the highest point is defined as the height of that fuzzy set.

31 / 69
Fuzzy terminologies: Bandwidth

Bandwidth :

For a normal and convex fuzzy set, the bandwidth (or width) is defined
as the distance the two unique crossover points:
Bandwidth(A) = | x1 - x2 |
where µA(x1 ) = µA(x2 ) = 0.5

30 / 69
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.

31 / 69
• The fuzzy approach uses a premise that humans don't
represent classes of objects
• The term "fuzzy systems" refers mostly to systems that
are governed by fuzzy IF-THEN rules. The IF part of an
implication is called the antecedent whereas the THEN
is called a consequent
• a fuzzy IF-THEN rule is a scheme for capturing
knowledge that involves imprecision.
• Fuzzy rules => partial matching capability, which
enables an inference to be made from a fuzzy rule even
when the rule's condition is only partially satisfied.
Fuzzy terminologies: Open and Closed
A fuzzy set A is
Open left
If limx →−∞ µA(x) = 1 and limx →+∞ µA(x) = 0
Open right:
If limx →−∞µA(x) = 0 and limx →+∞ µA(x) = 1
Closed
If : limx →−∞ µA(x) = limx →+∞ µA(x) = 0

Open left Open right

Closed

32 / 69
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 patient is
suffering from flue. So, the disease will be cured with certainty of 60%
and uncertainty 40%. Here, in stead of flue, other diseases with some
other certainties may be.

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.

33 / 69
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.

The main difference:

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

34 / 69
Fuzzy Membership
Functions

35 / 69
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).
Note: A membership function can be on
(a) a discrete universe of discourse and (b)
a continuous universe of discourse.
Example:

1.0 1.0
0.8 0.8
0.6 0.6
µA

µB
0.4 0.4
0.2 0.2

0 2 4 6 8 10 0 10 20 30 40 50
60

Number of children (X) Age (X)

A = Fuzzy set of “Happy family” B = “Young age”

36 / 69
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.
Following figures shows a typical examples of membership functions.

µ
µ

x x x

< triangular > < trapezoidal > < curve >

µ

x x

< non-uniform > < non-uniform >

Fuzzy MFs : Formulation and parameterization

38 / 69
Fuzzy MFs: Trapezoidal

a b c d

39 / 69
Fuzzy MFs: Gaussian

A Gaussian MF is specified by two parameters {c, σ} and can be

defined as below:

1 x−c
gaussian(x;c,σ) =e − 2 ( σ)2 .

c
0.1 0.1c 0.9c

40 / 69
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:
bell(x; a, b, c)= 1
1+| x −c
a
|
2b

Slope at x = b
2a
Slope at y = − b
xx b
b yy 2a

c-a c c+a

41 / 69
Example: Generalized bell MFs

Example: µ(x)= 1+x1 2 ;

a = b = 1 and c = 0;

1.0

-1 0 1

42 / 69
Generalized bell MFs: Different shapes

Changing a Changing b

Changing a
Changing a and b

43 / 69
Fuzzy MFs: Sigmoidal MFs

Parameters: {a, c } ; where c = crossover point and a = slope at c;

Sigmoid(x;a,c)= 1
a
−[ x −c ]
1+e

1.0

Slope = a
0.5

44 / 69
Fuzzy MFs : Example

Example : Consider the following grading system for a course.

Excellent = Marks ≤ 90
Very good = 75 ≤ Marks ≤ 90
Good = 60 ≤ Marks ≤ 75
Average = 50 ≤ Marks ≤ 60
Poor = 35 ≤ Marks ≤ 50
Bad= Marks ≤ 35

45 / 69
Grading System

A fuzzy implementation will look like the following.

Bad poor Average Good Very Good Excellent

1
.8
.6
.4
.2
0
10 20 30 40 50 60 70 80 90
marks

You can decide a standard fuzzy MF for each of the fuzzy garde.

46 / 69