Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

MAT633-FUZZY SET THEORY

CHAPTER THREE
FUZZY RELATIONS AND THE EXTENSION PRINCIPLES
LECTURE WEEK FIVE
MOHD FAZRIL IZHAR BIN MOHD IDRIS
TOPIC OF THIS WEEK

• From Classical Relations to Fuzzy Relations


• Binary Fuzzy Relations
• Operations and properties of binary fuzzy relations
• Inverse of Fuzzy Relations
FROM CLASSICAL RELATIONS TO FUZZY RELATIONS

• By definition, a relation of crisp set represents the existence (or non-existence) of association between
set through their elements
• The association can be in the form of interaction, connection or any sort of reasonable rules of
association
• Classical relations are given by set
• The boundary of a classical relation in general is definite and unambiguous
FROM CLASSICAL RELATIONS TO FUZZY RELATIONS

• A relation is normally represented by a characteristic function.


• The characteristic function for a relation, say R is defined as

• In other words, the membership value is 1 if the relation exists and 0 if the relation does not exist.
EXAMPLE
• Let R be a relation which associates a country with a currency

R Malaysia Britain Indonesia US


Ringgit 1 0 0 0
Rupiah 0 0 1 0
Pound 0 1 0 0
Dollar 0 0 0 1

• Let W be a relation “is the product of”


W Cow Sugar cane Wheat
Milk 1 0 0
Flour 0 0 1
Sugar 0 1 0
• A relation between two sets is called binary.
• A relation of three sets is called ternary.
• If four or five sets are involved, then the relation is called quarternary or quinary respectively.
• In general, a relation defined on n-sets is then called n-ary or n-dimensional.
FUZZY RELATIONS

• Vague and ambiguous relation between two objects often happens in daily communications
• For instance, statements like “object A is similar to object B” or “packaging P is bit bigger than packaging
Q” make it difficult to compare due to imprecise information or subjective description.
• However fuzzy relation can satisfactorily define this type of relationship.
• Introduced by Zadeh himself, fuzzy relation is a generalization of both classical relation and fuzzy sets.
• If the classical relation is given by set, fuzzy relations are represented by fuzzy sets.
• Fuzzy relations describe the presence of the association by giving the strength of the associations.
• Unlike the classical relation, the boundary of a fuzzy relation is vague.
BINARY FUZZY RELATIONS

• General expression of n-ary fuzzy relation


An n-ary fuzzy relation R is defined on the Cartesian product of X1 x X2 x X3 x … x Xn of sets X1, X2, X3, …, Xn
such as

 R x 1, x 2 ,, x n 
R 
X1xX2 x... xXn
x 1, x 2 ,, x n  (continuous relation )

Or  R x 1, x 2 ,, x n 
R 
X1xX2 x... xXn
x1, x 2 ,, x n  (discrete relation)

Where µR is the membership function of R such that µR : X1 x X2 x X3 x … x Xn → [0,1]


BINARY FUZZY RELATIONS

• Binary fuzzy relation


The binary fuzzy relation R between sets X and Y are defined as

R x, y 
R 
XxY
x, y 

Where µR : X x Y → [0,1]
FUZZY RELATIONAL MATRICES
• Let the universe X and Y be given as
X = { x1 , x2 , x3 , … , xm }
Y = { y1 , y2 , y3 , … , yn }
• The binary fuzzy relational matrix representing fuzzy relation R ⊆ X x Y is as follows.

y1 y2  yn
x1  R x 1, y 1   R x 1, y 2    R x 1, y n 
R  x2  R x 2 , y 1   R x 2 , y 2    R x 2 , y n 
    
xm  R x m , y 1   R x m , y 2    R x m , y n 

• A matrix expressing a fuzzy relation is also known as a fuzzy matrix and is useful for discrete expression
EXAMPLE
Let R be fuzzy relation “very far from” between two cities such that R ⊆ X x Y where
X = { Kangar (KGR), Johor Bahru (JB), Shah Alam (SA) }
Y = { Johor Bahru (JB), Kuala Lumpur (KL), Melaka (ML) }
The fuzzy relations R can be represented in the form of:
a) List notation:
R(x,y) = 0.95/(KGR,JB) + 0.7/(KGR,KL) + 0.8/(KGR,ML) + 0.5/(JB,KL) + 0.4/(JB,ML) + 0.5/(SA,JB) + 0.1/(SA,KL) +
0.15/(SA,ML)
b) Binary relational matrix:
R(x,y) =

c) Binary relational mappings:


OPERATION ON BINARY RELATIONS
• All operations on fuzzy sets (complements, intersections and unions) are applicable to fuzzy relations as
well. Let R and S be fuzzy relations on the Cartesian product X x Y.
a) Union of fuzzy relation: R ⋃ S

 RS x, y   s R x, y ,  S x, y 

b) Intersection of fuzzy raltions: R ∩ S


 RS x, y   t R x, y ,  S x, y 

c) Complement of fuzzy relation


 R
 x , y   c  R  x , y  

The operators s, t and c represent the s-norm, t-norm and fuzzy complement respectively.
EXAMPLE
Consider the following fuzzy relational matrices R and S.

y1 y2 y3 y1 y2 y3
x1 0.95 0.7 0 .8 x1 1 0 .2 0 .1
R S
x2 0 0 .5 0 .4 x2 0 0 .4 0 .2
x3 0 .5 0.1 0.15 x3 0 .3 0 .6 0 .8

a) The union of R and S can be calculated using:


i) the basic fuzzy union : µRUS(x,y) = max[µR(x), µS(x)]

ii) we may also find the union of fuzzy relation using other operators such, for example:
The algebraic sum: µRUS(x,y) = µR(x) + µS(x) - µR(x). µS(x)
b) The intersection of R and S using
i) the basic fuzzy intersection: µR∩S(x,y) = min[µR(x), µS(x)]

ii) similar to fuzzy union, we may obtain the intersection of fuzzy relation using other operators.
The algebraic product: µR∩S(x,y) = µR(x). µS(x)

c) The complement of S using the basic fuzzy complement: µS’(x,y) = 1 - µS(x,y)


PROPERTIES OF FUZZY RELATIONS

• Properties such as De Morgan’s Law and Law of Double Negation (Involutive) are also valid for fuzzy
relations
• The concepts of convexity, α-cuts, height and etc. introduced for fuzzy sets could also be applied for
fuzzy relations.
INVERSE OF FUZZY RELATION

• The operations of inverse and composition, which are not applicable to ordinary fuzzy sets could be
performed on fuzzy relation.
• Let R be fuzzy relation on the Cartesian product of set X and Y or X x Y. The inverse of the fuzzy binary
relation R-1 is a relation on Y x X such that R-1(y,x) = R(x,y) for all (x,y) in Y x X
• In matrix form, R-1 is simply the ‘transpose’ of R
• Example:

 1 0 .7 0 .8   1 0 0 .5 
   
R   0 0 .5 0 .4  R 1  0.7 0.5 0.6 
0.5 0.6 0.8  0.8 0.4 0.8 
PROPERTIES OF INVERSE FUZZY RELATION

• i) (R U S)-1 = R-1 U S-1


• ii) (R ∩ S)-1 = R-1 ∩ S-1
• iii) R ⊆ S = R-1 ⊆ S-1
• iv) (R-1)-1 = R
• v) (R-1) = (R)-1
EXAMPLE
JUNE 2016

JUNE 2015
EXAMPLE
DEC 2014

JUNE 2013
DIFFERENCES BETWEEN CLASSICAL RELATIONS AND
FUZZY RELATIONS
CLASSICAL RELATION FUZZY RELATION
Represented by set, which the boundary is definite Represented by fuzzy sets, which the boundary is
and unambiguous vague and ambiguous
The relation exists if the membership value is 1 and The strength of the relation between elements is
0 if the relation does not exist represented by a real number in the interval [0,1]

You might also like