Fuzzy Sets and Logic

Basics of Fuzzy Logic

MSc IS/ISR
 Overview
• This lecture introduces the basics of fuzzy logic, with a review of
classical logic and its operations, logical implication, and certain
classical inference mechanisms such as Tautologies or equivalences.
• The operations of disjunction, conjunction and negation as well as
classical implication and equivalences are reviewed.
• Operations on propositions are shown to be isomorphic with
operations on sets; hence an algebra of propositions is developed
using the algebra of operations covered in the last lesson.
• Fuzzy Logic is then shown to be an extension of classical logic when
partial truth values between true and not true are included to extend
bi-valued logic.
 Objectives
• To know the main five classical operations with propositions:
• disjunction, conjunction, negation, implication and equivalence.
• To learn that a proposition can be modelled as a set.
• To know how to derive the truth value of the conjunction, disjunction,
implication and equivalence of two propositions from the truth values
associated of the individual propositions.
• To obtain the membership function associated to the conjunction,
disjunction, implication and equivalence of two propositions
described using vague concepts modelled via fuzzy sets.
 Fuzzy Logic
• The notion fuzzy logic has three different meanings.
• In most cases, the term fuzzy logic refers to fuzzy logic in the broader sense,
including all applications and theories where fuzzy sets or concepts are
involved.
• The second and narrower meaning of the term fuzzy logic focuses on the field
of approximative reasoning where fuzzy sets are used and propagated within
an inference mechanism as it is for instance common in expert systems/fuzzy
control.
• Finally, fuzzy logic in the narrow sense, which is the topic of this section,
considers fuzzy logic as an extension of classical logic – from two valued logic
to multi-valued logic – and is devoted to issues connected to logical calculi
and the associated deduction mechanisms.
 From Classical Logic to Fuzzy Logic
• We cannot provide a complete introduction to fuzzy logic as a multi-
valued logic.
• In this session, we will introduce those notions of fuzzy logic which
are necessary or useful to understand fuzzy controllers, the following
topic in this module.
• We mainly need the concepts of fuzzy logic to introduce the set
theoretical operations for fuzzy sets.
• The basis for operations like union, intersection and complement are the
logical connectives disjunction, conjunction and negation, respectively.
• Therefore, we briefly repeat some fundamental concepts from classical logic
in order to generalize them to the field of fuzzy logic.
 Propositions and Truth Values
• Classical propositional logic deals with the formal handling of
statements (propositions) to which one of the two truth values 1 (for
true ) or 0 (for false ) can be assigned.
• We represent these propositions by Greek letters ϕ , ψ etc.
• Typical propositions, for which the formal symbols ϕ1 and ϕ2 may
stand are
ϕ1 : Four is an even number.
ϕ2 : 2+ 5 = 9.
• The truth value which is assigned to a proposition ϕ is denoted by
[[ϕ]] .
• For the above propositions we obtain [[ϕ1]] = 1 and [[ϕ2]] = 0.
 Combined Propositions
• If the truth values of single propositions are known, we can
determine the truth values of combined propositions using
• truth tables that define the interpretation of the corresponding logical
connectives.
• The most important logical connectives are
• the logical AND ∧ (conjunction),
• the logic OR ∨ (disjunction),
• the negation NOT ⌐, and
• the IMPLICATION→ .
 Truth Values of Combined Propositions
• The conjunction ϕ ∧ ψ of two propositions ϕ and ψ is true, if and
only if both ϕ and ψ are true.
• The disjunction ϕ ∨ ψ of ϕ and ψ obtains the truth value 1 (true), if
and only if at least one of the two propositions is true.
• The implication ϕ→ψ is only false, if the antecedent ϕ is true and
the consequent ψ is false.
• The negation ⌐ ϕ of the proposition ϕ is false, if and only if ϕ is true.
 Truth Tables
Conjunction

[[ϕ]] [[ψ]] [[ϕ ∧ ψ]]

1 1 1

1 0 0

0 1 0

0 0 0
Truth Tables
Disjunction

[[ϕ]] [[ψ]] [[ϕ ∨ ψ]]

1 1 1

1 0 1

0 1 1

0 0 0
Truth Tables
Implication

[[ϕ]] [[ψ]] [[ϕ → ψ]]

1 1 1

1 0 0

0 1 1

0 0 1
 Truth Tables
Negation

[[ϕ]] [[¬ϕ]]
1 0

0 1
 Example
This definition implies that the propositions
Four is an even number AND 2+ 5 = 9.
Four is an even number IMPLICATION 2 + 5 = 9.

are false, whereas the propositions

Four is an even number OR 2+ 5 = 9.
NOT 2+ 5 = 9.
are true.
Formally expressed, this means that we have
[[ϕ1∧ϕ2]] = 0, [[ϕ1→ϕ2]] = 0, [[ϕ1 ∨ ϕ2]] = 1 and [[¬ ϕ2]]= 1
 Fuzzy Propositions
• The assumption that a statement is either true or false is suitable for
mathematical issues.
• For many expressions formulated in natural language such a strict
separation between true and false statements would be unrealistic
and would lead to counterintuitive consequences.
• A driver will not calculate the distance he will need for stopping his
car abruptly on a wet road by using another friction constant in some
mathematical formula to calculate this distance.
• He will consider the rule: the wetter the road, the longer the distance
needed for breaking.
• In order to model this human information processing in a more
appropriate way, we use gradual truth values for statements.
• This means that a statement can not only be true (truth value 1) or false (truth value 0) but also more
or less true expressed by a value between 0 and 1.
 From Fuzzy Sets to Fuzzy Statements
• A fuzzy set models a property that elements of the universe of discourse can
have more or less.
• For example, let us consider the fuzzy set of high velocities
That means the degree of membership of a specific velocity v to the
fuzzy set of high velocities corresponds to the “truth value” which is
assigned to the statement “v is a high velocity”. In this sense, a fuzzy
set determines the corresponding truth values for a set of
statements—in our example for all statements we obtain, when we
consider in a concrete velocity value for v . In order to understand how
to operate with fuzzy sets, it is first of all useful to consider classical
crisp propositions.
 From Fuzzy Sets to Fuzzy Statements
A fuzzy set models a property that
elements of the universe of discourse can
have more or less.
For example, let us consider the fuzzy set of
high velocities (right hand side)
That means the degree of membership of a
specific velocity `v’ to the fuzzy set of `high
velocities’ corresponds to the “truth value”
which is assigned to the statement “v is a high
velocity”.
In this sense, a fuzzy set determines the
corresponding truth values for a set of
statements—in our example for all statements
we obtain its truth value when we consider in a
concrete velocity value for v .
In order to understand how to operate with
fuzzy sets, it is first of all useful to consider
classical crisp propositions.
 Dealing with combined fuzzy statements
• Dealing with combined propositions like
“160 km/h is a high velocity AND the stopping distance is about 110 m”
requires the extension of the truth tables of the classical logical
connectives for conjunction, disjunction, implication or negation.
• For conjunction, disjunction and implication this truth function
assigns to each combination of two truth values (the truth value
assigned to ϕ and ψ ) one truth value (the truth value of the
conjunction, disjunction of ϕ and ψ or the implication ϕ→ψ ).
• The truth function assigned to the negation has only one truth value
as argument.
 Truth Functions
• If we denote the truth function by w∗ associated with the classical
logical connective ∗ ∈ {∧,∨,→, ⌐ } , then w∗ is a binary or unary
function.

w∧, w∨, w→ : { 0, 1}2 →{ 0, 1}, while w⌐ : { 0, 1} → { 0, 1}

• For fuzzy propositions, where the unit interval [ 0, 1] replaces the

binary set { 0, 1} as set of possible truth values, we have to assign
truth functions to the logic connectives accordingly.
 Truth Functions (2)
• These truth functions have to be defined on the unit square or the
unit interval.
w∧, w∨, w→ : [0, 1]2 →[0, 1], while w⌐ : [0, 1] → [0, 1]

• A minimum requirement we demand of these functions is that,

limited to the values 0 and 1, they should provide the same values as
the corresponding truth function associated with the connectives of
classical logic.
• This requirement says that a combination of fuzzy propositions which
are actually crisp (non-fuzzy), because their truth values are 0 or 1
were, coincide with the usual combination of classical crisp
propositions.
 Truth Functions (3)
• The most frequently used truth functions for conjunction and
disjunction in fuzzy logic are the minimum or maximum.
w∧(α,β) = min{α,β} , w∨(α,β) = max{α,β}
• Normally the negation is defined by w⌐ (α) = 1 − α .
• The implication is often understood in the sense of the Łukasiewicz
implication
𝑤→ 𝛼, 𝛽 = min{1 − 𝛼 + 𝛽, 1}
or the Gödel implication
1 𝑖𝑓 𝛼 ≤ 𝛽
𝑤→ 𝛼, 𝛽 = 3
𝛽 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒.
 Truth Functions (4)
In Fuzzy Logic Control, however, the main implication membership
functions used are:

Mamdani implication: 𝑤→ 𝛼, 𝛽 = min 𝛼, 𝛽

Larsen implication: 𝑤→ 𝛼, 𝛽 = 𝛼 ⋅ 𝛽

and not the the Łukasiewicz implication and the Gödel implication
obtained extending propositional logic.
 Truth Functions (4)
• In the framework of fuzzy logic, we should always choose a t-norm as the
truth function for conjunction.
• The minimum can be considered as a special t-norm, since it is the only
idempotent t-norm which means that only the minimum satisfies the
property
t (α,α) = α for all α ∈ [ 0, 1] .
• Only the idempotence of a t-norm can guarantee that the truth values of
the proposition ϕ and ϕ ∧ ϕ coincide, which at first sight seems to be a
canonical requirement, letting the minimum seem to be the only
reasonable choice for the truth functions for the conjunction in the context
of fuzzy logic.
• However, the following example shows that the idempotency property is
not always desirable.
 Example of conjunction with t-norm
A buyer has to decide between the houses A and B . The houses are very
similar in most aspects. So, the buyer makes the decision considering the
criteria good price and good location. After careful consideration the
following “truth values” are assigned to the decisive aspects:
 Example (cont.)
• He chooses house x ∈ {A,B} for which the proposition
“The price of house x is good AND The location of house x is good”
yields the greater truth value.
• This means that the buyer will choose house A if
[[ϕ1 ∧ ϕ2]] > [[ϕ3 ∧ ϕ4]],
and house B otherwise.
• When we determine the truth value of the conjunction by the minimum, we
would obtain the value 0.6 for both of the houses and thus the houses
would be regarded as equally good.
• But this is counterintuitive because house A has definitely a better price
than house B and the locations are equally good.
• However, when we choose a non-idempotent t-norm, for example, the
algebraic product or the Łukasiewicz t-norm, as truth function for the
conjunction, we will always favour house A .
 Truth Functions (5)
• Equally, in the framework of fuzzy logic, we should always choose a t-
conorm as the truth function for disjunction.
• Like the minimum is the only idempotent t-conorm which means that
only the minimum satisfies the property
s (α,α) = α for all α ∈ [ 0, 1] .
• Applying the disjunction in the sense of the maximum only the
proposition with the greatest truth value determines the truth value
of the disjunction of the propositions.
• We can avoid this disadvantage, if we give up idempotency.
 t-norms and Implications
• In addition to the connection between t-norms and t-conform
(duality), we can also find connections between t-norms and
implications.

• A continuous t-norm t induces the residuated implication 𝑡⃗ by the

formula
𝑡⃗ 𝛼, 𝛽 = sup{ 𝛾 ∈ [0,1]|𝑡 𝛼, 𝛾 ≤ 𝛽}.

• Thus, by residuation we obtain

• the Łukasiewicz implication from the Łukasiewicz t-norm, and
• the Gödel implication from the minimum t-norm.
 Biimplication or Equivalence
• In classical logic the biimplication or equivalence is defined as

[[ϕ↔ψ]] = [[(ϕ→ψ) ∧ (ψ →ϕ) ]]

• If α = [[ϕ]] and β= [[ψ]] then using the residuated implication 𝑡⃗ we

would have that the formula for the biimplication would be

⃡
𝑡 (𝛼, 𝛽) = 𝑡⃗ max{𝛼, 𝛽}, min{𝛼, 𝛽}
= t (𝑡⃗(𝛼, 𝛽), 𝑡⃗(𝛽, 𝛼))
=min { 𝑡⃗(𝛼, 𝛽), 𝑡⃗(𝛽, 𝛼)}.
 Quantifiers
• Besides the logical operators like conjunction, disjunction, implication or negation in (fuzzy) logic,
there also exist the quantifiers ∀ (all) and ∃ (exists).
• The universal quantifier ∀ and the existential quantifier ∃ are closely related to the conjunction
and the disjunction, respectively.
• If the universal set X is finite, e.g. X = {x1, . . . , xn} , then

(∀x ∈ X)(P (x)) is equivalent to [P(x1)∧…∧P(xn)]

The truth value of the statement (∀x ∈ X)(P (x)) would be:

[[(∀x ∈ X)(P (x))]]= inf{[[P(x)]]; x ∈ X}

• Other t-norms than the minimum are normally not used for the universal quantifier, because the
non-idempotent property leads easily to the truth value zero in the case of an infinite universe of
discourse.
 Quantifiers (2)
The same consideration about the existential quantifier

(∃ x ∈ X)(P (x)) is equivalent to [P(x1) ∨ … ∨ P(xn)]

leads to
[[(∃ x ∈ X)(P (x))]]= sup{[[P(x)]]; x ∈ X}
 Example
Consider the predicate P(x) with the interpretation “x is a high velocity”. Let the truth
value [[P(x)]] be given by the fuzzy set of the high velocities from Figure shown before
[[P(x)]] = μhv(x)
We have for instance
[[P( 150)]] = 0, [[P( 170)]] = 0. 5 and [[P( 190)]] = 1.

Thus, the statement

(∀x ∈ [ 170, 200])(P(x)) = (“All velocities between 170 km/h and 200 km/h are high velocities”)
has the truth value
[[∀x ∈ [ 170, 200] : P(x)]]= inf{[[P(x)]]; x ∈ [ 170, 200]}= inf{μhv(x) | x ∈ [ 170, 200]}= 0. 5.

Analogously, we obtain
[[∃x ∈ [ 100, 180] : P(x)]] = 0. 75.
 Summary
• In this session about fuzzy logic we have discussed
• various ways of combining fuzzy propositions.
• An essential assumption we have used is that of truth functionality.
• This means that the truth value of the combination of several propositions depends only on the
truth values of the propositions, but not on the individual propositions.
• If we understand the conjunction in its classical sense, a conjunctive combination of
a proposition with itself should be equivalent to itself, which is not satisfied for
nonidempotent t-norms.
• Another possibility is to understand the conjunction as a list of pro and con
arguments for a thesis or as a proof.
• In any case, the repeated use of the same (fuzzy) argument within a proof might
result in a loss of credibility and thus idempotency is not desirable, even for a
conjunction of a proposition with itself.
• Fortunately, for fuzzy control these consideration are of minor importance, because
in this application area fuzzy logic is used in a more restricted context, where we do
not have to worry about combining the same proposition with itself.