5주차 - chapter 4 5

Chapter 4
Fuzzy Relations and

the Extension Principle
Contents
• From Classical Relations to Fuzzy Relations
• Relations
• Projections and Cylindric Extensions
• Composition of Fuzzy Relations
• The Extension Principle
• Summary and Further Readings
• Exercises
Chap. 4 Fuzzy relations and the extension principle 2

From Classical Relations to Fuzzy Relations
Relation
• Let U and V be two arbitrary classical (nonfuzzy, crisp) sets.
• The Cartesian product of U and V :

𝑈 × 𝑉 = {(𝑢, 𝑣)|𝑢 ∈ 𝑈 𝑎𝑛𝑑 𝑣 ∈ 𝑉}
• The Cartesian product of arbitrary n nonfuzzy sets :

𝑈1 × 𝑈2 ×. . .× 𝑈𝑛 = {(𝑢1 , 𝑢2 , . . . , 𝑢𝑛 )|𝑢1 ∈ 𝑈1 , . . . , 𝑢𝑛 ∈ 𝑈𝑛 }
• Relation 𝑄 is a subset with meaning

𝑄 𝑈1 , . . . , 𝑈𝑛 ⊂ 𝑈1 ×. . .× 𝑈𝑛
𝐼𝑛 𝑐ℎ𝑎𝑝. 3, 𝑥,𝑦∈𝑈. 𝐼𝑛 𝑐ℎ𝑎𝑝. 4, 𝑥∈𝑈, 𝑦∈𝑉 and the fitness value of is

denoted by relation matrix.

• Example. 4.1 Let 𝑈 = {1,2,3}, 𝑉 = {2,3,4}. Find a relation
𝑄 ⊂ 𝑈 × 𝑉 meaning "the first element is no smaller that the second
element".
𝑄(𝑈, 𝑉) = {(2,2), (3,2), (3,3)}
• The membership function

1 if (𝑢1 , . . . , 𝑢𝑛 ) ∈ 𝑄 𝑈1 , . . . , 𝑈𝑛 )
𝜇Q (𝑢1 , . . . , 𝑢𝑛 ) = ቊ
0 otherwise
• The relation 𝑄 𝑈, 𝑉 of (4,4)

𝑉
2 3 4
1 0 0 0
𝑈 2 1 0 0
3 1 1 0

• Example. 4.2 Let 𝑈 = 𝑆𝑎𝑛𝐹𝑟𝑎𝑛𝑐𝑖𝑠𝑐𝑜, 𝐻𝑜𝑛𝑔𝐾𝑜𝑛𝑔, 𝑇𝑜𝑘𝑦𝑜 ,
𝑉 = {𝐵𝑜𝑠𝑡𝑜𝑛, 𝐻𝑜𝑛𝑔𝐾𝑜𝑛𝑔}. Find a relation of "very far".
𝑉
𝐵𝑜𝑠𝑡𝑜𝑛 𝐻𝐾
SF 0.3 0.9
𝐻𝐾 1 0
𝑈 𝑇𝑜𝑘𝑦𝑜 0.95 0.1
0.3 0.9
𝑅= 1 0
0.95 0.1
• Definition. 4.1 A fuzzy relation is a fuzzy set defined in the Cartesian

product of crisp sets.
𝑄 = {((𝑢1 , 𝑢2 , . . . , 𝑢𝑛 ), 𝜇Q (𝑢1 , 𝑢2 , . . . , 𝑢𝑛 ))| 𝑢1 , 𝑢2 , . . . , 𝑢𝑛
∈ 𝑈1 × 𝑈2 ×. . .× 𝑈𝑛 }

AE (approximately equal) & ML
Chap. 2 Mathematical Models of Systems 6

Projection and Cylindric Extensions

• Definition. 4.2 The projection of 𝑄:
𝜇Qp (𝑢𝑖1 , . . . , 𝑢𝑖𝑘 ) = 𝑚𝑎𝑥 𝜇Q (𝑢1 , . . . , 𝑢𝑛 )
𝑢𝑗1 ∈𝑈𝑗1 ,...,𝑢𝑗 𝑛−𝑘 ∈𝑈𝑗 𝑛−𝑘
where {𝑢𝑗1 , . . . , 𝑢𝑗(𝑛−𝑘)} is the complement of {𝑢𝑖1 , . . . , 𝑢𝑖𝑘 } with respect to
{𝑢1 , . . . , 𝑢𝑛 }.
• As a special case, if 𝑄 is a binary fuzzy

relation in 𝑈 × 𝑉, then the projection of 𝑄
on 𝑈, denoted by 𝑄1
𝜇Qp 𝑥 = max 𝜇𝑄 (𝑥, 𝑦)

𝑦∈𝑉

• Example. 4.4 the projection of fuzzy relation 𝑄
𝑄1 (projection to 𝑈) = 0.9/SF + 1/HK + 0.95/Tokyo : 0.9, 1, 0.95 T
𝑄2 (projection to 𝑉) = 1/Boston + 0.9/HK : (1, 0.9)
• The projections of AE(approximately equal) defined by:

• Definition. 4.3 Let 𝑄𝑃 be a fuzzy relation is a subsequence, then the
cylindric extension is a fuzzy relation defined by
• As a special case, if 𝑄1 is a fuzzy set in 𝑈, then cylindric extension of

𝑄1 to 𝑈 × 𝑉 is a fuzzy relations 𝑄1𝐸 in 𝑈 × 𝑉 defined by

• Example. 4.5 cylindric extensions to 𝑈 × 𝑉.
𝑄1𝐸 = 0.9/(SF, Boston) + 0.9/(SF, HK) + 1/(HK, Boston)
+ 1/(HK,HK) + 0.95/(Tokyo, Booston)
+ 0.95/(Tokyo, HK)
𝑄2𝐸 = 1/(SF, Boston) + 1/(HK, Boston) + 1/(Tokyo, Boston)
+ 0.9/(SF, HK) + 0.9/(HK, HK) + 0.9/(Tokyo, HK)
0.9 0.9 1 0.9

𝑄1𝐸 = 1 1 , 𝑄2𝐸 = 1 0.9
0.95 0.95 1 0.9
• The cylindric extensions of 𝐴𝐸1 and 𝐴𝐸2 defined by:

• The Cartesian product of 𝐴1 , … , 𝐴𝑛 , denoted by 𝐴1 × ⋯ × 𝐴𝑛 , is a
fuzzy relation in 𝑈1 × ⋯ × 𝑈𝑛 , whose membership function is
defined as
where ∗ represents any t-norm operator.
• Lemma. 4.1 If 𝑄 is a fuzzy relation in 𝑈1 × ⋯ × 𝑈𝑛 , and 𝑄1 , … , 𝑄𝑛 , are

its projections on 𝑈1 , … , 𝑈𝑛 , respectively, then
𝑄 ⊂ 𝑄1 × ⋯ × 𝑄𝑛
Where we use min for the t-norm.

• Proof. Substituting Definition 4.2 into Definition 4.3
𝜇Qp𝐸 𝑢𝑖1 , . . . , 𝑢𝑖𝑘 = 𝑚𝑎𝑥 𝜇Q (𝑢1 , . . . , 𝑢𝑛 )
𝑢𝑗1 ∈𝑈𝑗1 ,...,𝑢𝑗 𝑛−𝑘 ∈𝑈𝑗 𝑛−𝑘
𝑄 ⊂ 𝑄𝑖𝐸
• where 𝑄𝑖𝐸 is the cylindric extension of 𝑄𝑖 to 𝑈1 × ⋯ × 𝑈𝑛 .
𝑄 ⊂ 𝑄𝑖𝐸 ∩ ⋯ ∩ 𝑄𝑛𝐸 = 𝑄1 × ⋯ × 𝑄𝑛

Composition of Fuzzy Relations
• Definition. 4.4 𝑃 ∘ 𝑄 is the composition if and only if
𝜇𝑃∘𝑄 (𝑥, 𝑧) = 𝑚𝑎𝑥 𝑡[𝜇𝑃 (𝑥, 𝑦), 𝜇𝑄 (𝑦, 𝑧)], 𝑥 ∈ 𝑈, 𝑦 ∈ 𝑉, 𝑧 ∈ 𝑊.
𝑦∈𝑉
• The max-min composition of fuzzy relations 𝑃 𝑈, 𝑉 and 𝑄 𝑉, 𝑊 is a

fuzzy relation 𝑃 ∘ 𝑄 in 𝑈 × 𝑊 defined by the membership function
𝜇𝑃∘𝑄 𝑥, 𝑧 = max min[𝜇𝑃 𝑥, 𝑦 , 𝜇𝑄 𝑦, 𝑧 ]
𝑦∈𝑉
where 𝑥, 𝑧 ∈ 𝑈 × 𝑊.
• The max-product composition of fuzzy relations 𝑃 𝑈, 𝑉 and 𝑄 𝑉, 𝑊 is

a fuzzy relation 𝑃 ∘ 𝑄 in 𝑈 × 𝑊 defined by the membership function
𝜇𝑃∘𝑄 𝑥, 𝑧 = max[𝜇𝑃 𝑥, 𝑦 𝜇𝑄 𝑦, 𝑧 ]
𝑦∈𝑉
where 𝑥, 𝑧 ∈ 𝑈 × 𝑊.

• Example. 4.6 Let 𝑈 = 𝑆𝑎𝑛𝐹𝑟𝑎𝑛𝑐𝑖𝑠𝑐𝑜, 𝐻𝑜𝑛𝑔𝐾𝑜𝑛𝑔, 𝑇𝑜𝑘𝑦𝑜 , 𝑉 =
𝐵𝑜𝑠𝑡𝑜𝑛, 𝐻𝑜𝑛𝑔𝐾𝑜𝑛𝑔 , W= 𝑁𝑌𝐶, 𝐵𝑒𝑖𝑗𝑖𝑛𝑔 . Let P(U,V) denote the fuzzy
relation “very far”, Q(V,W) denote “very near”
• Find a compositions
𝑉
𝐵𝑜𝑠𝑡𝑜𝑛 𝐻𝐾 𝑊
0.3 0.9 𝑁𝑌𝐶 𝐵𝑒𝑖𝑗𝑖𝑛𝑔
SF
𝐻𝐾 𝐵𝑜𝑠𝑡𝑜𝑛 0.3 0.9
1 0 𝑉
𝑈 𝑇𝑜𝑘𝑦𝑜 0.95 0.1 𝐻𝐾 0.1 0.9
𝑃 = 0.3/(SF, Boston) + 0.9/(SF, HK) + 1/(HK, Boston)

+ 0/(HK, HK) + 0.95/(Tokyo, Booston) + 0.1/(Tokyo, HK)
𝑄= 0.95/(Boston, NYC) + 1/(Boston, Beijing) + 0.1/(Tokyo, Boston)
+ 0.9/(HK, Beijing)

• Using the max-min composition.
• The final 𝑃 ∘ 𝑄 is

• Using the max-product composition.
• Specifically, max-min composition and max-product composition.

The Extension Principle
• Let 𝑓: 𝑈 → 𝑉 be a function from crisp set 𝑈 to crisp set 𝑉. Suppose
that a fuzzy set 𝐴 in 𝑈 is given and we want to determine a fuzzy set
𝐵 = 𝑓 𝐴 in 𝑉 that is induced by 𝑓. If 𝑓 is an one-to-one mapping,
then we can define
𝜇𝐵 (𝑦) = 𝜇𝐴 𝑓 −1 𝑦 𝑥 , 𝑦∈𝑉
where 𝑓 −1 𝑦 is the inverse of 𝑓, that is, 𝑓 𝑓 −1 𝑦 = 𝑦.
• The membership function for 𝐵 is defined as

𝜇𝐵 (𝑦) = 𝑚𝑎𝑥
−1
𝜇𝐴 (𝑥), 𝑦 ∈ 𝑉
𝑥∈𝑓 (𝑦)
where 𝑓 −1 𝑦 denotes the set of all points 𝑥 ∈ 𝑈 such that 𝑓 𝑥 = 𝑦.

The identity is called the extension principle.

• Example. 4.8 Let 𝑈 = 1,2, … , 10 and 𝑓 𝑥 = 𝑥 2 . Let small be a
fuzzy set in 𝑈 defined by
1 1 0.8 0.6 0.4
𝑠𝑚𝑎𝑙𝑙 = + + + +
1 2 3 4 5
• Then, is consequence of the extension principle,

1 1 0.8 0.6 0.4
𝑠𝑚𝑎𝑙𝑙 2 = + + + +
1 4 9 16 25

0.5 0.8 1 0.4
• Exercises. 4.4 Consider fuzzy set 𝐴 = + + +
−1 0 1 2
2
and function 𝑓 𝑥 = 𝑥 . Determine the fuzzy set 𝑓 𝑥
using the extension principle.
Chap. 2 Mathematical Models of Systems 19

Summary and Further Readings
• The concepts of fuzzy relations, projections, and

cylindric extensions.
• The max-min and max-product compositions of fuzzy
relations.
• The extension principle and its applications.

Exercises
• Exercises. 4.1 Given an n-ary relation, how many different
projections of the relation can be taken?
• Exercises. 4.2 Consider the fuzzy relation 𝑄 defined in 𝑈1 × ⋯ ×

𝑈4 where 𝑈1 = {𝑎, 𝑏, 𝑐 }, 𝑈2 = {𝑠, 𝑡}, 𝑈3 = {𝑥, 𝑦} and 𝑈4 = {𝑖, 𝑗}:
• (a) Compute the projections of 𝑄 on 𝑈1 × 𝑈2 × 𝑈4 , 𝑈1 × 𝑈3 and 𝑈4 .

• (b) Compute the cylindric extensions of the projections in (a) to 𝑈1 ×
𝑈2 × 𝑈3 × 𝑈4 .

Exercises
• Exercises. 4.3 Consider the three binary fuzzy relations defined by
the relational matrices:
Compute the max-min and max-product compositions 𝑄1 ∘ 𝑄2 , 𝑄2 ∘

𝑄3 and 𝑄1 ∘ 𝑄2 ∘ 𝑄3 .
0.5 0.8 1 0.4

• Exercises. 4.4 Consider fuzzy set 𝐴 = + + + and function
−1 0 1 2
𝑓 𝑥 = 𝑥 2 . Determine the fuzzy set 𝑓 𝑥 using the extension
principle.
• Exercises. 4.5 Compute the 𝜇𝐴𝐸∘𝑀𝐿 (𝑥, 𝑧) in Example 4.7 𝑥, 𝑧 =

0,0 , 0,1 , 1,0 , 1,1 .

Chapter 5
Linguistic variables and fuzzy

IF-THEN rules
Contents
• 5.1 From Numerical Variables to Linguistic Variables

• 5.2 Linguistic Hedges
• 5.3 Fuzzy IF-THEN Rules
5.3.1 Fuzzy Propositions
5.3.2 Interpretations of Fuzzy IF-THEN Rules
• 5.4 Summary and Further Readings
• 5.5 Exercises
Chap. 5 Linguistic variables and fuzzy IF-THEN rules 26

From numerical variables to linguistic
variables
• Definition 5.1
𝒍𝒊𝒏𝒈𝒖𝒊𝒔𝒕𝒊𝒄 𝒗𝒂𝒓𝒊𝒂𝒃𝒍𝒆 : If a variable can take words in natural languages as
131its values (where the words are characterized by fuzzy sets defined in the
universe of discourse in which the variable is defined. )
• Example 5.1
The speed of a car is a variable x that takes values in the interval [0, Vmax]
Vmax is the maximum speed of the car
define three fuzzy
1. slow
2. medium
3. fast
Figure 5.1 : The speed of a car as a linguistic variable that can take fuzzy sets “slow”, “medium” and
“fast” as its values

variables
• Definition 5.2
– 𝐴 𝑙𝑖𝑛𝑔𝑢𝑖𝑠𝑡𝑖𝑐 𝑣𝑎𝑟𝑎𝑏𝑙𝑒 is characterized by 𝑋, 𝑇, 𝑈, 𝑀 ,
where :
• 𝑋 : name of the linguistic variable
• 𝑇 : set of linguistic values that 𝑋 can take
• 𝑈 : actual physical domain in which the 𝑋 takes its
quantitative (crisp) values
• 𝑀 : semantic rule that relates each linguistic value
in 𝑇 with a fuzzy set in 𝑈
Example 5.1
𝑋 : speed of the car 𝑇 : {slow, medium, fast}
𝑈 : [0, Vmax]
𝑀 : relates "slow," "medium," and "fast" with the membership
functions shown in Fig. 5.1.
variables
𝐰𝐡𝐲 𝐢𝐬 𝐭𝐡𝐞 𝐜𝐨𝐧𝐜𝐞𝐩𝐭 𝐨𝐟 𝐥𝐢𝐧𝐠𝐮𝐢𝐬𝐭𝐢𝐜 𝐯𝐚𝐫𝐢𝐚𝐛𝐥𝐞 𝐢𝐦𝐩𝐨𝐫𝐭𝐚𝐧𝐭?
Because linguistic variables are the most fundamental elements in human
knowledge representation When we use sensors to measure a variable,
they give us numbers; when we ask human experts to evaluate a variable,
they give us words.

Linguistic Hedges
In our daily life, we often use more than one word to describe a variable
if we view the speed of a car as a linguistic variable, then its values might
be "not slow," "very slow," "slightly fast," "more or less medium“
the value of a linguistic variable is a composite term x = ~x1,x2 . . .xn, that
is a concatenation of atomic terms x1,x2, …, xn. These atomic terms may
be classified into three groups:
• Primary terms, which are labels of fuzzy sets; in Example 5.1, they are
“slow," "medium," and "fast."
• Complement "not" and connections "and" and "or."
• Hedges, such as "very," “slightly," "more or less”

Linguistic Hedges
Definition 5.3
Let 𝐴 be a fuzzy set in 𝑈, then 𝑣𝑒𝑟𝑦 𝐴 is defined as
𝜇𝑣𝑒𝑟𝑦 𝐴 𝑥 = 𝜇𝐴 𝑥 2
𝑚𝑜𝑟𝑒 𝑜𝑟 𝑙𝑒𝑠𝑠 𝐴 is defined as
𝜇𝑚𝑜𝑟𝑒 𝑜𝑟 𝑙𝑒𝑠𝑠 𝐴 𝑥 = 𝜇𝐴 𝑥 1/2

Linguistic Hedges
• Example 5.2
– Let 𝑈 = { 1, 2, … , 5} and the fuzzy set 𝑠𝑚𝑎𝑙𝑙 be defined as
𝑠𝑚𝑎𝑙𝑙 = 1/1 + 0.8/2 + 0.6/3 + 0.4/4 + 0.2/5
Then, according to Definition 5.3, we have
𝑣𝑒𝑟𝑦 𝑠𝑚𝑎𝑙𝑙 = 1/1 + 0.64/2 + 0.36/3 + 0.16/4 + 0.04/5
𝑣𝑒𝑟𝑦 𝑣𝑒𝑟𝑦 𝑠𝑚𝑎𝑙𝑙 = 𝑣𝑒𝑟𝑦(𝑣𝑒𝑟𝑦 𝑠𝑚𝑎𝑙𝑙)

= 1/1 + 0.4096/2 + 0.1296/3 + 0.0256/4
+ 0.0016/5
𝑚𝑜𝑟𝑒 𝑜𝑟 𝑙𝑒𝑠𝑠 𝑠𝑚𝑎𝑙𝑙 = 1/1 + 0.8944/2 + 0.7746/3 + 0.6325/4

+ 0.4472/5

Fuzzy IF-THEN Rules
• Fuzzy IF-THEN rules (Conditional state expressed)

– 𝐼𝐹 < 𝑓𝑢𝑧𝑧𝑦 𝑝𝑟𝑜𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 >, 𝑇𝐻𝐸𝑁 < 𝑓𝑢𝑧𝑧𝑦 𝑝𝑟𝑜𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 >
• There are two types of fuzzy propositions: Atomic fuzzy propositions, and
compound fuzzy propositions
• Atomic fuzzy proposition is single statement

– 𝑥 𝑖𝑠 𝐴 (𝑤ℎ𝑒𝑟𝑒 𝑥 𝑖𝑠 𝑎 𝑙𝑖𝑛𝑔𝑢𝑖𝑠𝑡𝑖𝑐 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒, 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑎 𝑙𝑖𝑛𝑔𝑢𝑖𝑠𝑡𝑖𝑐 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑥)
• A compound fuzzy proposition is a composition of atomic fuzzy propositions

using the connectives "and," "or," and "not"
− 𝑥 𝑖𝑠 𝑆 − 𝑥 𝑖𝑠 𝑆 𝑜𝑟 𝑥 is not 𝑀
− 𝑥 𝑖𝑠 𝑀 − 𝑥 𝑖𝑠 not 𝑆 𝑎𝑛𝑑 𝑥 𝑖𝑠 not 𝐹
− 𝑥 𝑖𝑠 𝐹 − 𝑥 𝑖𝑠 𝑆 𝑎𝑛𝑑 𝑥 𝑖𝑠 not 𝐹 𝑜𝑟 𝑥 𝑖𝑠 𝑀
Fuzzy set – S : slow M : medium F : fast

Fuzzy IF-THEN Rules
For connective "and" use fuzzy intersections. Specifically, let x and y be

linguistic variables in the physical domains U and V, and A and B be
fuzzy sets in U and V, respectively, then the compound fuzzy
proposition
𝑥 𝑖𝑠 𝐴 𝑎𝑛𝑑 𝑦 𝑖𝑠 𝐵 ∶
Fuzzy relation : 𝑥 ∈ 𝑈, 𝑦 ∈ 𝑉, 𝐴, 𝐵 ∶ 𝑓𝑢𝑧𝑧𝑦 𝑠𝑒𝑡𝑠 ∈ 𝑈, 𝑉
– 𝑥 𝑖𝑠 𝐴 𝑎𝑛𝑑 𝑦 𝑖𝑠 𝐵 ∶ 𝜇𝐴∩𝐵 𝑥, 𝑦 = 𝑡[𝜇𝐴 𝑥 , 𝜇𝐵 𝑦 ]

t : [O,1] x [O,1] → [O, 1] is any t-norm.
– 𝑥 𝑖𝑠 𝐴 𝑜𝑟 𝑦 𝑖𝑠 𝐵 ∶ 𝜇𝐴∪𝐵 𝑥, 𝑦 = 𝑠[𝜇𝐴 𝑥 , 𝜇𝐵 𝑦 ]
s : [O, 1] x [0, I] → [O, 11 is any s-norm.
– 𝑥 𝑖𝑠 𝑛𝑜𝑡 𝐴 ∶ 𝜇𝐴ҧ 𝑥 = 𝑐𝑜𝑚𝑝𝑙𝑒𝑚𝑒𝑛𝑡(𝜇𝐴 𝑥 )

Fuzzy IF-THEN Rules
• Example 5.3
FP = 𝑥 𝑖𝑠 𝑆 𝑎𝑛𝑑 𝑦 𝑖𝑠 𝑛𝑜𝑡 𝐹 𝑜𝑟 𝑥 𝑖𝑠 𝑀
is a fuzzy relation in the product space [0, 𝑉𝑚𝑎𝑥 ]3 with the
membership function
→ 𝜇𝐹𝑃 𝑥1 , 𝑦2 , 𝑧3 = 𝑠 𝑡 𝜇𝑠 𝑥1 , 𝑐 𝜇𝐹 𝑦2 , 𝜇𝑀 𝑥3
where s, t and c are s-norm, t-norm and fuzzy complement

operators, respectively, the fuzzy sets S = small, M = medium, and
F = fast are defined in Fig. 5.1, and 𝑥1 =𝑥2 =𝑥3 =x.

Fuzzy IF-THEN Rules
• Interpretation of Fuzzy IF-THEN Rules
– IF < FP1 > THEN < FP2 >
– the expression IF p THEN q is written as p → q with the

implication → regarded as a connective defined by Table 5.1,
where p and q are propositional variables whose values are
either truth (T) or false (F).
𝑝 → 𝑞 ⇔ 𝑝ҧ ∨ 𝑞 ⇔ (𝑝 ∧ 𝑞) ∨ 𝑝ҧ

Fuzzy IF-THEN Rules
In the following, we rewrite (5.7) as IF < FPl > THEN < FP2 > and
replace the p and q in (5.21) and (5.22) by FPl and FP2,
respectively, where FPl and FP2 are fuzzy propositions. We assume
that FPl is a fuzzy relation defined in U =𝑈1 × 𝑈2 ×…𝑈𝑛 , FP2 is a
fuzzy relation defined in V =𝑉1 × 𝑉2 ×…𝑉𝑚 , and x and y are
linguistic variables (vectors) in U and V, respectively.

Fuzzy IF-THEN Rules
• IF < FPl > THEN < FP2 >

– Dienes-Rescher Implication :
𝜇𝑄𝐷 𝑥, 𝑦 = 𝑚𝑎𝑥 1 − 𝜇𝐹𝑃1 𝑥 , 𝜇𝐹𝑃2 𝑦 ⟺ 𝑝ҧ ∨ 𝑞
If we replace the logic operators - and V in (5.21) by the basic fuzzy

complement (3.1) and the basic fuzzy union (3.2), respectively,
then we obtain the so-called Dienes-Rescher implication.

Fuzzy IF-THEN Rules

– Lukasiewicz Implication :
𝜇𝑄𝐿 𝑥, 𝑦 = 𝑚𝑖𝑛 1, 1 − 𝜇𝐹𝑃1 𝑥 + 𝜇𝐹𝑃2 𝑦 ⟺ (𝑝 ∨ 𝑝) ∧ (𝑝ҧ ∨ 𝑞)
If we use the Yager s-norm (3.10) with w = 1 for the V and basic
fuzzy complement (3.1) for the-in (5.21), we obtain the Lukasiewicz
implication

Fuzzy IF-THEN Rules

– Zadeh Implication
– Here the fuzzy IF-THEN rule IF < FPl > THEN < FP2 > is
interpreted as a fuzzy relation 𝑄𝑧 in U x V with the membership
function
𝜇𝑄𝑍 𝑥, 𝑦 = 𝑚𝑎𝑥 min[𝜇𝐹𝑃1 𝑥 , 𝜇𝐹𝑃2 𝑦 , 1 − 𝜇𝐹𝑃1 𝑥 ] ⟺ (𝑝 ∧ 𝑞) ∨ 𝑝ҧ

Fuzzy IF-THEN Rules

– Gödel Implication
The Godel implication is a well-known implication formula in
classical logic. By generalizing it to fuzzy propositions, we obtain
the following: the fuzzy IF-THEN rule IF < FPl > THEN < FP2 >
is interpreted as a fuzzy relation 𝑄𝐺 in U x V with the membership
function
1 𝑖𝑓 𝜇𝐹𝑃1 𝑥 ≤𝜇𝐹𝑃2 𝑦
𝜇𝑄𝐺 𝑥, 𝑦 = ቊ
𝜇𝐹𝑃2 𝑦 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Fuzzy IF-THEN Rules
• For example, when we say "IF speed is high, THEN

resistance is high," we are concerned only with a local
situation in the sense that this rule tells us nothing about
the situations when "speed is slow," "speed is medium,"
etc. Therefore, the fuzzy IF-THEN rule
• should be interpreted as
• where NOTHING means that this rule does not exist. In

logic terms, it becomes

Fuzzy IF-THEN Rules
• Mamdani Implications
– IF <FP1> THEN <FP2> ELSE < NOTHING>
𝑝 ⟶𝑞 =𝑝∧𝑞
𝜇𝑄𝑀𝑀 𝑥, 𝑦 = min[𝜇𝐹𝑃1 𝑥 , 𝜇𝐹𝑃2 𝑦 ]

or
𝜇𝑄𝑀𝑃 𝑥, 𝑦 = 𝜇𝐹𝑃1 𝑥 𝜇𝐹𝑃2 𝑦
– Mamdani implications are the most widely used implications in

fuzzy systems and fuzzy control. They are supported by the
argument that fuzzy IF-THEN rules are local. However, one may
not agree with this argument.
– Different implications are needed to cope with the diversity of
interpretations.

Fuzzy IF-THEN Rules
• Example 5.4.
• Let x1 be the speed of a car, x2 be the acceleration, and
y be the force applied to the accelerator. Consider the
following fuzzy IF-THEN rule
where "slow" is the fuzzy set defined in Fig. 5.1, that is
"small" is a fuzzy set in the domain of acceleration with the membership

function

Fuzzy IF-THEN Rules
• Example 5.4.
and "large" is a fuzzy set in the domain of force applied to the
accelerator with the membership function

Fuzzy IF-THEN Rules
• Example 5.4.

Fuzzy IF-THEN Rules
• Example 5.4.
If we use the Dienes-Rescher implication then the fuzzy IF-
THEN rule is interpreted as a fuzzy relation QD(x1,x2, y)
From we have

Fuzzy IF-THEN Rules
• Example 5.4.

Fuzzy IF-THEN Rules
• Example
• From Example 5.4 we see that if the membership
functions in the atomic fuzzy propositions are not smooth
functions the computation of the final membership
functions 𝜇 𝑄𝐷, 𝜇Qz, etc., is cumbersome, although it is
straightforward. A way to resolve this complexity is to
use a single smooth function to approximate the
nonsmooth functions; see the following example. 5.4.

Fuzzy IF-THEN Rules
• Example 5.4 (Cont 'd). Suppose we use
to approximate the 𝜇slow(xl)
to approximate the 𝜇small(x2) and

Fuzzy IF-THEN Rules
• Example 5.4 (Cont 'd). Suppose we use
To approximate the μ𝑙𝑎𝑟𝑔𝑒 𝑦 , if we use Mamdani product

implication and algebraic product for the t-norm then

Fuzzy IF-THEN Rules
• Example 5.5
– 𝑈 = 1,2,3,4 , 𝑉 = 1,2,3 , 𝑥 ∈ 𝑈, 𝑦 ∈ 𝑉
IF 𝑥 is large THEN 𝑦 is small

where the fuzzy sets "large" and "small" are defined as
𝑙𝑎𝑟𝑔𝑒 = 0/1 + 0.1/2 + 0.5/3 + 1/4

𝑠𝑚𝑎𝑙𝑙 = 1/1 + 0.5/2 + 0.1/3
If we use the Dienes-Rescher implication (5.23), then the fuzzy IF-

THEN rule is interpreted as the following fuzzy relation 𝑄𝐷 in U x V:

Fuzzy IF-THEN Rules
• Example 5.5
• If we use the Lukasiewicz implication, the rule becomes
For the Zadeh implication and the Godel implication we have
and

Fuzzy IF-THEN Rules
• Example 5.5
• Finally, if we use the Mamdani implication then the fuzzy
IF- THEN rule becomes
• and

Summary and Further Readings
In this chapter, we have demonstrated the following:
-The concept of linguistic variables and the characterization of hedges.
-The concept of fuzzy propositions and fuzzy IF-THEN rules.
-Different interpretations of fuzzy IF-THEN rules, including Dienes-Rescher,

Lukasiewicz, Zadeh, Godel and Mamdani implications.
-Properties and computation of these implications.

HW
Exercise 5.4. Use basic fuzzy operators (3.1, 3.2, 3.3) for "not,"
"or," and "and," respectively, and determine the membership
functions for the fuzzy propositions (5.12, 5.13). Plot these
membership functions
Exercise 5.5. Consider the fuzzy IF-THEN rule (5.33) with the
fuzzy sets "slow" "small" and "large" defined by (5.42, 43, 44)
respectively. Use min for the t-norm (5.16) and compute the fuzzy
relations QD, QL, Qz, QG, QMM and QMP.

5주차 - chapter 4 5

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5주차 - chapter 4 5

Uploaded by

Copyright:

Available Formats

Chapter 4

Fuzzy Relations and

Chap. 4 Fuzzy relations and the extension principle 2

• The Cartesian product of U and V :

• The Cartesian product of arbitrary n nonfuzzy sets :

• Relation 𝑄 is a subset with meaning

𝐼𝑛 𝑐ℎ𝑎𝑝. 3, 𝑥,𝑦∈𝑈. 𝐼𝑛 𝑐ℎ𝑎𝑝. 4, 𝑥∈𝑈, 𝑦∈𝑉 and the fitness value of is

Chap. 4 Fuzzy relations and the extension principle 3

• The membership function

• The relation 𝑄 𝑈, 𝑉 of (4,4)

Chap. 4 Fuzzy relations and the extension principle 4

• Definition. 4.1 A fuzzy relation is a fuzzy set defined in the Cartesian

Chap. 4 Fuzzy relations and the extension principle 5

Chap. 2 Mathematical Models of Systems 6

Projection and Cylindric Extensions

• As a special case, if 𝑄 is a binary fuzzy

𝜇Qp 𝑥 = max 𝜇𝑄 (𝑥, 𝑦)

Chap. 4 Fuzzy relations and the extension principle 7

𝑄1 (projection to 𝑈) = 0.9/SF + 1/HK + 0.95/Tokyo : 0.9, 1, 0.95 T

𝑄2 (projection to 𝑉) = 1/Boston + 0.9/HK : (1, 0.9)

• The projections of AE(approximately equal) defined by:

Chap. 4 Fuzzy relations and the extension principle 8

• As a special case, if 𝑄1 is a fuzzy set in 𝑈, then cylindric extension of

Chap. 4 Fuzzy relations and the extension principle 9

0.9 0.9 1 0.9

Chap. 4 Fuzzy relations and the extension principle 10

where ∗ represents any t-norm operator.

• Lemma. 4.1 If 𝑄 is a fuzzy relation in 𝑈1 × ⋯ × 𝑈𝑛 , and 𝑄1 , … , 𝑄𝑛 , are

Where we use min for the t-norm.

Chap. 4 Fuzzy relations and the extension principle 11

Chap. 4 Fuzzy relations and the extension principle 12

• The max-min composition of fuzzy relations 𝑃 𝑈, 𝑉 and 𝑄 𝑉, 𝑊 is a

• The max-product composition of fuzzy relations 𝑃 𝑈, 𝑉 and 𝑄 𝑉, 𝑊 is

Chap. 4 Fuzzy relations and the extension principle 13

𝑃 = 0.3/(SF, Boston) + 0.9/(SF, HK) + 1/(HK, Boston)

Chap. 4 Fuzzy relations and the extension principle 14

Chap. 4 Fuzzy relations and the extension principle 15

• Specifically, max-min composition and max-product composition.

Chap. 4 Fuzzy relations and the extension principle 16

where 𝑓 −1 𝑦 is the inverse of 𝑓, that is, 𝑓 𝑓 −1 𝑦 = 𝑦.

• The membership function for 𝐵 is defined as

where 𝑓 −1 𝑦 denotes the set of all points 𝑥 ∈ 𝑈 such that 𝑓 𝑥 = 𝑦.

Chap. 4 Fuzzy relations and the extension principle 17

• Then, is consequence of the extension principle,

Chap. 4 Fuzzy relations and the extension principle 18

Chap. 2 Mathematical Models of Systems 19

• The concepts of fuzzy relations, projections, and

Chap. 4 Fuzzy relations and the extension principle 20

• Exercises. 4.2 Consider the fuzzy relation 𝑄 defined in 𝑈1 × ⋯ ×

• (a) Compute the projections of 𝑄 on 𝑈1 × 𝑈2 × 𝑈4 , 𝑈1 × 𝑈3 and 𝑈4 .

Chap. 4 Fuzzy relations and the extension principle 21

Compute the max-min and max-product compositions 𝑄1 ∘ 𝑄2 , 𝑄2 ∘

0.5 0.8 1 0.4

• Exercises. 4.5 Compute the 𝜇𝐴𝐸∘𝑀𝐿 (𝑥, 𝑧) in Example 4.7 𝑥, 𝑧 =

Chap. 4 Fuzzy relations and the extension principle 22

Linguistic variables and fuzzy

• 5.1 From Numerical Variables to Linguistic Variables

Chap. 5 Linguistic variables and fuzzy IF-THEN rules 26

define three fuzzy

Chap. 5 Linguistic variables and fuzzy IF-THEN rules 27

Chap. 5 Linguistic variables and fuzzy IF-THEN rules 29

• Complement "not" and connections "and" and "or."

• Hedges, such as "very," “slightly," "more or less”