*Zadeh, L. A. (1965), Fuzzy sets. Information and control, 8(3), 338-353
FUZZY SETS*
 Introduction
 Differences between crisp sets & Fuzzy sets
 Some definitions W.R.T to Fuzzy sets (FSS)
 Graphical Representation of union & intersection
 Algebraic operations on FSs
 Convex combination
 Fuzzy Relation
 Convexity
TABLE OF CONTENTS
 Consider a collection/set of animals. Dogs, horses, etc. are part of it but
rocks, fluids, etc. are not part of it.
 What about bacteria, starfish?
 Consider a set that contains “all real numbers much greater than 1”.
Where does ‘2’ & ‘10’ fall?
 Such classification problems occur frequently in real life problems.
INTRODUCTION
 Consider few more examples: class of beautiful women, class of tall
men.
 Even though such sets have an inherent type of imprecision in
information that they convey, still they are very important in a number of
fields such as pattern recognition, etc. Such sets are called FUZZY
SETS.
INTRODUCTION….
 Fuzzy Sets: Those collection of objects where it is not possible to make
a sharp distinction between the belongingness or non-belongingness to
the collection.
 These are useful in cases where the source of imprecision is the absence
of sharply defined criteria of the class of membership rather that the
probability theory.
INTRODUCTION….
 Let the universal set be denoted by X and its elements by x i.e. X= {x}.
We define a set A on X such that 𝐴 ⊂ 𝑋. We define the term grade of
membership denoted by fA(x) which represents the information
regarding the extent of belongingness of x to set A.
 If 𝑥 ∈ 𝑋, if fA(x)= 0 or 1 only and no intermediate value, then the set A
is called the crisp set and if the value of fA x belongs to the closed
interval [0, 1], then A is called the Fuzzy set. Eg: For the set X= set of
real numbers close to 1, we have fA 0 = 0; fA 10 = 0.2; fA 500 = 1
DIFFERENCES BETWEEN CRISP
SETS & FUZZY SETS
 Empty FS: 𝑓𝐴 𝑥 = 0 ∀𝑥 ∈ 𝑋
 Equal FSs: Given two FSs A & B, then if 𝑓𝐴 𝑥 = 𝑓𝐵(𝑥)∀𝑥 ∈ 𝑋, then A=B
 Complement(𝑨′): It is defined as 𝑓 𝐴′ 𝑥 = 1 − 𝑓𝐴(𝑥)
 Subset: 𝐴 ⊂ 𝐵 ↔ 𝑓𝐴 ≤ 𝑓𝐵 𝑖. 𝑒. 𝑓𝐴 𝑥 ≤ 𝑓𝐵 𝑥 ∀𝑥 ∈ 𝑋
 Union: Let 𝐶 = 𝐴 ∪ 𝐵, 𝑡ℎ𝑒𝑛 𝑓𝑐 𝑥 = 𝑀𝑎𝑥 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 , ∀𝑥 ∈ 𝑋
Corollary: The union of A & B is the smallest fuzzy set containing both A & B.
 Intersection: Let 𝐶 = 𝐴 ∩ 𝐵, 𝑡ℎ𝑒𝑛 𝑓𝑐 𝑥 = 𝑀𝑖𝑛 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 , ∀𝑥 ∈ 𝑋
Corollary: The intersection of A & B is the largest fuzzy set containing both A & B.
SOME DEFINITIONS W.R.T TO FUZZY SETS (FSS)
Graphically, the union is shown by sections 1 & 2 of the graph. Intersection
is shown by sections 3 & 4:
GRAPHICAL REPRESENTATION OF UNION &
INTERSECTION
 Algebraic product: Denoted by AB, is defined for FSs A & B as:
𝑓𝐴𝐵 = 𝑓𝐴 𝑓𝐵
 Algebraic sum: Denoted by A+B, is defined as:
𝑓𝐴+𝐵 = 𝑓𝐴 + 𝑓𝐵
𝑝𝑟𝑜𝑣𝑖𝑑𝑒𝑑 𝑓𝐴 + 𝑓𝐵 ≤ 1 ∀𝑥 ∈ 𝑋
 Absolute difference: Denoted by |A-B|, is defined as:
𝑓|𝐴−𝐵| = |𝑓𝐴 − 𝑓𝐵|
ALGEBRAIC OPERATIONS ON FSS
 Let A, B & Λ be three FSs, then their convex combination is denoted by
𝐴, 𝐵; Λ = Λ𝐴 + Λ′ 𝐵
𝑤ℎ𝑒𝑟𝑒Λ′ is the complement of Λ
 Or in terms of the membership function as:
𝑓 𝐴,𝐵;Λ 𝑥 = 𝑓Λ 𝑥 𝑓𝐴 𝑥 + [1 − 𝑓Λ 𝑥 ]𝑓𝐵(𝑥)
 A basic property of convex combination of A, B and Λ is given by:
𝐴 ∩ 𝐵 ⊂ 𝐴, 𝐵; Λ ⊂ 𝐴 ∪ 𝐵 ∀Λ
CONVEX COMBINATION
 For FSs, the above expression is rewritten as:
𝑀𝑖𝑛 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 ≤ 𝜆𝑓𝐴 𝑥 + 1 − 𝜆 𝑓𝐵 𝑥 ≤ 𝑀𝑎𝑥 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥
 Also, it is possible to find a FS ‘C’, s.t. C= 𝐴, 𝐵; Λ and the membership of this
set is given by:
𝑓Λ 𝑥 =
𝑓𝐶 𝑥 − 𝑓𝐵(𝑥)
𝑓𝐴 𝑥 − 𝑓𝐵(𝑥)
, 𝑥 ∈ 𝑋
CONVEX COMBINATION……
 A fuzzy relation in X is a fuzzy set in the product space 𝑋 × 𝑋.
 For example, the relation 𝑥 ≫ 𝑦, 𝑥, 𝑦 ∈ 𝑅, is expressed as a fuzzy set A
in 𝑅2, with the membership function value given by: 𝑓𝐴(𝑥, 𝑦). Lets say
𝑓𝐴 10,5 = 0; 𝑓𝐴 100,10 = 0.7 and 𝑓𝐴 100,1 = 1.
 However, the realtion values are subjective interpretations.
FUZZY RELATION
 A fuzzy set A is called the convex iff
𝑓𝐴 𝜆𝑥1 + 1 − 𝜆 𝑥2 ≥ 𝑀𝑖𝑛[𝑓𝐴 𝑥1 , 𝑓𝐴 𝑥2 ]
∀𝑥1, 𝑥2 ∈ 𝑋 𝑎𝑛𝑑𝜆 ∈ [0,1]
 However the above definition does not imply that 𝑓𝐴(𝑥) must be a
convex function of 𝑥.
CONVEXITY
THANK
YOU

Fuzzy sets

  • 1.
    *Zadeh, L. A.(1965), Fuzzy sets. Information and control, 8(3), 338-353 FUZZY SETS*
  • 2.
     Introduction  Differencesbetween crisp sets & Fuzzy sets  Some definitions W.R.T to Fuzzy sets (FSS)  Graphical Representation of union & intersection  Algebraic operations on FSs  Convex combination  Fuzzy Relation  Convexity TABLE OF CONTENTS
  • 3.
     Consider acollection/set of animals. Dogs, horses, etc. are part of it but rocks, fluids, etc. are not part of it.  What about bacteria, starfish?  Consider a set that contains “all real numbers much greater than 1”. Where does ‘2’ & ‘10’ fall?  Such classification problems occur frequently in real life problems. INTRODUCTION
  • 4.
     Consider fewmore examples: class of beautiful women, class of tall men.  Even though such sets have an inherent type of imprecision in information that they convey, still they are very important in a number of fields such as pattern recognition, etc. Such sets are called FUZZY SETS. INTRODUCTION….
  • 5.
     Fuzzy Sets:Those collection of objects where it is not possible to make a sharp distinction between the belongingness or non-belongingness to the collection.  These are useful in cases where the source of imprecision is the absence of sharply defined criteria of the class of membership rather that the probability theory. INTRODUCTION….
  • 6.
     Let theuniversal set be denoted by X and its elements by x i.e. X= {x}. We define a set A on X such that 𝐴 ⊂ 𝑋. We define the term grade of membership denoted by fA(x) which represents the information regarding the extent of belongingness of x to set A.  If 𝑥 ∈ 𝑋, if fA(x)= 0 or 1 only and no intermediate value, then the set A is called the crisp set and if the value of fA x belongs to the closed interval [0, 1], then A is called the Fuzzy set. Eg: For the set X= set of real numbers close to 1, we have fA 0 = 0; fA 10 = 0.2; fA 500 = 1 DIFFERENCES BETWEEN CRISP SETS & FUZZY SETS
  • 7.
     Empty FS:𝑓𝐴 𝑥 = 0 ∀𝑥 ∈ 𝑋  Equal FSs: Given two FSs A & B, then if 𝑓𝐴 𝑥 = 𝑓𝐵(𝑥)∀𝑥 ∈ 𝑋, then A=B  Complement(𝑨′): It is defined as 𝑓 𝐴′ 𝑥 = 1 − 𝑓𝐴(𝑥)  Subset: 𝐴 ⊂ 𝐵 ↔ 𝑓𝐴 ≤ 𝑓𝐵 𝑖. 𝑒. 𝑓𝐴 𝑥 ≤ 𝑓𝐵 𝑥 ∀𝑥 ∈ 𝑋  Union: Let 𝐶 = 𝐴 ∪ 𝐵, 𝑡ℎ𝑒𝑛 𝑓𝑐 𝑥 = 𝑀𝑎𝑥 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 , ∀𝑥 ∈ 𝑋 Corollary: The union of A & B is the smallest fuzzy set containing both A & B.  Intersection: Let 𝐶 = 𝐴 ∩ 𝐵, 𝑡ℎ𝑒𝑛 𝑓𝑐 𝑥 = 𝑀𝑖𝑛 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 , ∀𝑥 ∈ 𝑋 Corollary: The intersection of A & B is the largest fuzzy set containing both A & B. SOME DEFINITIONS W.R.T TO FUZZY SETS (FSS)
  • 8.
    Graphically, the unionis shown by sections 1 & 2 of the graph. Intersection is shown by sections 3 & 4: GRAPHICAL REPRESENTATION OF UNION & INTERSECTION
  • 9.
     Algebraic product:Denoted by AB, is defined for FSs A & B as: 𝑓𝐴𝐵 = 𝑓𝐴 𝑓𝐵  Algebraic sum: Denoted by A+B, is defined as: 𝑓𝐴+𝐵 = 𝑓𝐴 + 𝑓𝐵 𝑝𝑟𝑜𝑣𝑖𝑑𝑒𝑑 𝑓𝐴 + 𝑓𝐵 ≤ 1 ∀𝑥 ∈ 𝑋  Absolute difference: Denoted by |A-B|, is defined as: 𝑓|𝐴−𝐵| = |𝑓𝐴 − 𝑓𝐵| ALGEBRAIC OPERATIONS ON FSS
  • 10.
     Let A,B & Λ be three FSs, then their convex combination is denoted by 𝐴, 𝐵; Λ = Λ𝐴 + Λ′ 𝐵 𝑤ℎ𝑒𝑟𝑒Λ′ is the complement of Λ  Or in terms of the membership function as: 𝑓 𝐴,𝐵;Λ 𝑥 = 𝑓Λ 𝑥 𝑓𝐴 𝑥 + [1 − 𝑓Λ 𝑥 ]𝑓𝐵(𝑥)  A basic property of convex combination of A, B and Λ is given by: 𝐴 ∩ 𝐵 ⊂ 𝐴, 𝐵; Λ ⊂ 𝐴 ∪ 𝐵 ∀Λ CONVEX COMBINATION
  • 11.
     For FSs,the above expression is rewritten as: 𝑀𝑖𝑛 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥 ≤ 𝜆𝑓𝐴 𝑥 + 1 − 𝜆 𝑓𝐵 𝑥 ≤ 𝑀𝑎𝑥 𝑓𝐴 𝑥 , 𝑓𝐵 𝑥  Also, it is possible to find a FS ‘C’, s.t. C= 𝐴, 𝐵; Λ and the membership of this set is given by: 𝑓Λ 𝑥 = 𝑓𝐶 𝑥 − 𝑓𝐵(𝑥) 𝑓𝐴 𝑥 − 𝑓𝐵(𝑥) , 𝑥 ∈ 𝑋 CONVEX COMBINATION……
  • 12.
     A fuzzyrelation in X is a fuzzy set in the product space 𝑋 × 𝑋.  For example, the relation 𝑥 ≫ 𝑦, 𝑥, 𝑦 ∈ 𝑅, is expressed as a fuzzy set A in 𝑅2, with the membership function value given by: 𝑓𝐴(𝑥, 𝑦). Lets say 𝑓𝐴 10,5 = 0; 𝑓𝐴 100,10 = 0.7 and 𝑓𝐴 100,1 = 1.  However, the realtion values are subjective interpretations. FUZZY RELATION
  • 13.
     A fuzzyset A is called the convex iff 𝑓𝐴 𝜆𝑥1 + 1 − 𝜆 𝑥2 ≥ 𝑀𝑖𝑛[𝑓𝐴 𝑥1 , 𝑓𝐴 𝑥2 ] ∀𝑥1, 𝑥2 ∈ 𝑋 𝑎𝑛𝑑𝜆 ∈ [0,1]  However the above definition does not imply that 𝑓𝐴(𝑥) must be a convex function of 𝑥. CONVEXITY
  • 14.