Fuzzy Systems

Introduction

Rudolf Kruse Christian Moewes

{kruse,cmoewes}@iws.cs.uni-magdeburg.de

Otto-von-Guericke University of Magdeburg

Faculty of Computer Science
Department of Knowledge Processing and Language Engineering
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 1 / 50
 Fuzzy Systems
What are we going to talk about?!

Research on fuzzy systems wants to establish

• theoretical and methodological bases for computational
intelligence
• tools and techniques for design of intelligent systems

Fuzzy systems focus on applications

• where vagueness, imprecision and uncertainty play important role

Fuzzy set theory

• supplies us with the basic mathematical foundation

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 2 / 50

 Literatur for the Lecture
Dubois, D. and Prade, H., editors (2000).
Fundamentals of Fuzzy Sets.
Kluwer Academic Publishers, Boston, MA, USA.

Höppner, F., Klawonn, F., Kruse, R., and Runkler, T. (1999).

Fuzzy Cluster Analysis: Methods for Classification, Data Analysis and Image Recognition.
John Wiley & Sons Ltd, New York, NY, USA.

Klir, G. and Yuan, B. (1995).

Fuzzy Sets and Fuzzy Logic: Theory and Applications.
Prentice Hall, Upper Saddle River, NJ, USA.

Kruse, R., Gebhardt, J., and Klawonn, F. (1994).

Foundations of Fuzzy Systems.
John Wiley & Sons Ltd, Chichester, United Kingdom.

Kruse, R., Gebhardt, J., and Klawonn, F. (1995).

Fuzzy-Systeme.
Teubner, Stuttgart, Germany, 2nd edition.

Michels, K., Klawonn, F., Kruse, R., and Nürnberger, A. (2006).

Fuzzy Control: Fundamentals, Stability and Design of Fuzzy Controllers, volume 200 of Studies in Fuzziness and
Soft Computing.
Springer, Berlin / Heidelberg, Germany.

Nauck, D., Klawonn, F., and Kruse, R. (1997).

Foundations of Neuro-Fuzzy Systems.
John Wiley & Sons Ltd, Chichester, United Kingdom.

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 3 / 50

 Organization of the Lecture
Questions about the lecture...
• consultation: on Wednesdays, from 11 a.m. to 12 noon, G29-008
• preferred way of contact: [email protected]

Questions about the exercises...

• tutor: Christian Moewes, mailto:[email protected]
• consultation: usually from 8 a.m. to 5 p.m., G29-019
• 1st exercise was yesterday :-)
• next one is on Tuesday, October 27, 2009 in G22A-208

Updated information on the course

• http://fuzzy.cs.uni-magdeburg.de

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 4 / 50

 Conditions for Certificate (“Schein”) and Exam

No matter if certificate or exam, everybody must...

• contribute well in exercises every week
• present ≥ 2 solutions to written assignment during exercises

Only for diploma students: Certificate will get who...

• tick off ≥ 66% of all written assignments
• submit ≥ 2 running implementations of programming assignments
• small colloquium (≈ 10 minutes) or written test (if > 20 students)

For everybody else: Exam or marked certificate will get who...

• tick off ≥ 50% of all written assignments
• oral exam (≈ 25 minutes) or written exam (if > 20 students)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 5 / 50

 Class Schedule in October 2009

Calendar Week 43
• Exercise: Assignment Sheet 1 (already discussed...)
• Lecture: Introduction

Calendar Week 44
• No exercise next week on October 27th, 2009.
• No lecture next week on October 28th, 2009.

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 6 / 50

 Class Schedule in November 2009
Calendar Week 45
• Exercise: Assignment Sheet 2
• Lecture: Fuzzy Set Theory
Calendar Week 46
• Exercise: Assignment Sheet 3
• Lecture: Fuzzy Arithmetic (with C. Moewes)
Calendar Week 47
• Exercise: Assignment Sheet 4
• Lecture: Fuzzy Relations (with C. Moewes)
Calendar Week 48
• Exercise: Assignment Sheet 5
• Lecture: Possibility Theory
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 7 / 50
 Class Schedule in December 2009
Calendar Week 49
• Exercise: Assignment Sheet 6
• Lecture: Fuzzy Control I
Calendar Week 50
• Exercise: Assignment Sheet 7
• Lecture: Fuzzy Control II
Calendar Week 51
• Exercise: Assignment Sheet 8
• Lecture: Fuzzy Pattern Recognition (with C. Moewes)
Calendar Week 52
• Exercise: Assignment Sheet 9
• No lecture on December 23rd, 2009 due to Christmas break!
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 8 / 50
 Class Schedule in January 2010

Calendar Week 2
• Exercise: Assignment Sheet 10
• Lecture: Fuzzy Rule Generation

Calendar Week 3
• Exercise: Assignment Sheet 11
• Lecture: Neuro-Fuzzy and Evolutionary Fuzzy Systems

Calendar Week 4
• Exercise: Assignment Sheet 12
• Lecture: Kernel Methods

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 9 / 50

 Outline

1. Motivation

2. Vagueness, Imprecision, Uncertainty

3. From Logic to Fuzzy Logic

4. From Sets to Fuzzy Sets

5. Fuzzy Sets

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 10 / 50

 Motivation

• every day humans use vague/imprecise/uncertain linguistic terms

• e.g., “big”, “fast”, “about 12 o’clock”, “old”, etc.
• all complex human actions are decisions based on such concepts
• driving and parking
• financial/business decisions
• law and justice
• giving a lecture
• listening to the professor/tutor
• so, these terms and the way they are processed play crucial role

• computers need a mathematical model to express and process

such complex semantics
• concepts in classical mathematics are inadequate for such models

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 11 / 50

 Lotfi Asker Zadeh (1965)

• classes of objects in real world do not have

precisely defined criteria of membership
• such imprecisely defined “classes” play an
important role in human thinking
• particularly in domains of pattern
recognition, communication of information,
and abstraction
Zadeh in 2004 (born 1921)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 12 / 50

 Outline

1. Motivation

2. Vagueness, Imprecision, Uncertainty

Vagueness
Imprecision
Uncertainty

3. From Logic to Fuzzy Logic

4. From Sets to Fuzzy Sets

5. Fuzzy Sets

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 13 / 50

 Vagueness
Any notion is said to be vague when its meaning is not fixed by
sharp boundaries.

• vagueness is a particular case of ambiguity

• vagueness can be applied fully/to a certain degree/not at all
• this gradualness (“membership gradience”) is also called fuzziness

• a proposition is vague if it contains gradual predicates

• such propositions may be neither true nor false, but in-between
• so, true to a certain degree ⇒ partial truth
• forms of such degrees can be found in natural language
• e.g., “very”, “rather”, “almost not”, etc.

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 14 / 50

 The Sorites Paradox

If a heap of sand is small, adding one grain of sand to it leaves it small.

A heap of sand with a single grain is small.
Hence all heaps of sand are small.

• paradox comes from all-or-nothing treatment of small

• degree of truth of “heap of sand is small” decreases by adding one
grain after another
• certain number of words refer to continuous numerical scales

Why is there fuzziness and ambiguity in all languages?

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 15 / 50

 Why is there fuzziness and ambiguity?
Any language is discrete and real world is continuous!
• gap between discrete representation and continuous perception
⇒ prevalence of ambiguity in languages

• consider the word young, applied to humans

• the more fine-grained the scale of age...
• e.g., going from years to months, weeks, days, etc.
• ... the more difficult is it to fix threshold
• below which young fully applies
• above which young does not at all

• conflict between linguistic and numerical representation

• finite term set {young, mature, old}
• real-valued interval [0, 140] years for humans

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 16 / 50

 Why is there fuzziness and ambiguity?

Natural language tolerates the presence of exceptions!

• for example, some birds fly, others do not

• ostrich, penguin, emu, cassowary, rhea, kiwi, etc.
• among birds, some are more typical than others
• some are “better” members of that category than others
• set of birds is thus partially ordered

• degree of memberships evaluate how “normal” individuals are

• it is not always easy to numerically represent these degrees

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 17 / 50

 Vagueness

Is there a membership threshold for vaguely defined classes?

• consider the notion bald

• man without hair on his head is bald
• a hairy man is not bald
• usually, bald is only partly applicable
• where to set baldness/non baldness threshold?

• Fuzzy set theory does not assume any threshold!

• this has consequences for the logic behind fuzzy set theory
• will be discussed in this course later

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 18 / 50

 Imprecision

Any measurement yields results of limited accuracy!

• also characterizes language
• pertains to measurable concepts (especially metric properties)
⇒ numerical imprecision

• intervals instead of single numbers

• amount of imprecision might be length of assigned interval
• in logic, imprecision appears as disjunctions p ∨ q
• not know if p or q or both are true ⇒ incompleteness

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 19 / 50

 Poincaré’s Paradox – Indistinguishability

A = B, B = C and A < C

• it may be impossible to distinguish physical quantities

• A from B and B from C
• might sometimes distinguish A from C ⇒ contradicts classical
logic
• however, if d(A, B) < ε for distance function d and threshold ε
⇒ observation A = a is imprecise
• should be interval I(A) of center a and width ε
⇒ A = B means I(A) intersects I(B)
• imprecision is also modeled by fuzzy subsets of real numbers

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 20 / 50

 Uncertainty

• imprecision and vagueness refer to contents of a piece of

information

Uncertainty
The ability of an agent to claim whether a proposition holds or not.

• two different kinds of uncertainty

1. uncertainty modeled by propositional logic
2. uncertainty modeled by probability

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 21 / 50

 Uncertainty modeled by Propositional Logic

• uncertainty stems from incomplete or imprecise information

• uncertainty about Boolean propositions can be gradual
• compare to probability theory
• there, partial belief ranges on unit interval [0, 1]

• in propositional logic, uncertainty is three-valued

1. sure about a proposition
2. sure about its negation
3. unsure about both

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 22 / 50

 Uncertainty modeled by Probability

• uncertainty also comes from conflicting but precisely observed

pieces of information
• usually in statistics
• consider random experiment
• run several times and does not produce same outcomes

• if information is fuzzy, then there is gradual uncertainty

• uncertainty due to lack of information
• so, fuzzy sets ⇒ gradual theory of uncertainty
• possibility theory will be discussed in this lecture later

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 23 / 50

 Examples of Uncertainty
Uncertainty differs from imprecision and vagueness.
It only results from them.

• “This car is between 10 and 15 years old.” (pure imprecision)

• lack of knowledge
• lack of ability to measure or to evaluate numerical features

• “This car is very big.” (imprecision and vagueness)

• lack of precise definition of notion big
• modifier very indicates rough degree of “bigness”

• “This car was probably made in Germany.” (uncertainty)

• uncertainty about well-defined proposition made in Germany
• perhaps based on statistics (random experiment)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 24 / 50

 Outline

1. Motivation

2. Vagueness, Imprecision, Uncertainty

3. From Logic to Fuzzy Logic

The Traditional Logic
Graded Truth

4. From Sets to Fuzzy Sets

5. Fuzzy Sets

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 25 / 50

 The Traditional or Aristotlelian Logic I
What is logic about? Different schools speak different languages!

• there are raditional, linguistic,

psychological, epistemological and
mathematical schools
• traditional logic founded by
Aristotle (384-322 B.C.)
• Aristotlelian logic as formal
approach to human reasoning
• still used today in Artificial
Intelligence for
• knowledge representation
• reasoning about knowledge
Detail of “The School of Athens” by R. Sanzio (1509)
• also used in fuzzy logic showing Plato (left) and his student Aristotle (right).

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 26 / 50

 The Traditional or Aristotlelian Logic II
Short History

• Aristotle introduced logic of terms and drawing

conclusion from two premises
• Great Greeks (Chrisippus) also developed logic
of propositions Gottlob Frege (1848-1925)

• Gottlob Frege dealt with soundness of deductive

method
• Jan Łukasiewicz founded multi-valued logic
• multi-valued logic is to fuzzy set theory
what classical logic is to set theory

Jan Łukasiewicz (1878-1956)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 27 / 50

 The Basic Principle of Classical Logic

• The Principle of Bivalence

• “Every proposition is either true or false.”
• formally developed by Tarski
• Gödel proved it is wrong (incompleteness
theorems) Alfred Tarski (1902-1983)
• Łukasiewicz suggested to replace it as follows

• The Principle of Valence

• “Every proposition has a truth value.”
• propositions can have intermediate truth value
• expressed by number from unit interval [0, 1]

Kurt Gödel (1906-1978)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 28 / 50

 Ancestor of Fuzzy Set Calculus I

• Hermann Weyl devised the following logic

• let any proposition be assigned a number from
fixed subset L of [0, 1]
• L must be closed under operation 1 − x

a and b = min(a, b)
a or b = max(a, b)
not a = 1 − a
Hermann Weyl (1885-1955)
a implies b = 1 − a + min(a, b)
= min(1, 1 − a + b)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 29 / 50

 Ancestor of Fuzzy Set Calculus II
• Weyl also introduced calculus of vague predicates (like functions)
• assume predicate f : U 7→ [0, 1] for fixed universe U
• he then defined operations as follows

conjunction: f ∩ g = min(f , g)
disjunction: f ∪ g = max(f , g)
negation: fc =1−f

• Weyl’s logic is an ancestor of the fuzzy logic

• but interpreting his definitions considers truth values as
probabilities
• probability and truth address different issues!
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 30 / 50
 Outline

1. Motivation

2. Vagueness, Imprecision, Uncertainty

3. From Logic to Fuzzy Logic

4. From Sets to Fuzzy Sets

Cantor’s Theory
Boole’s Theory
Menger’s Theory

5. Fuzzy Sets

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 31 / 50

 Cantor’s Theory

• “By a set we understand every collection made

into a whole of definite, distinct objects of our
intuition or of our thought.” (Georg Cantor)

• for a set in Cantor’s sense, the following

properties hold:

1. x 6= {x }
2. if x ∈ X and X ∈ Y then x ∈ /Y
Georg Cantor (1845-1918)
3. set of all subsets of X is denoted as F (X )
4. ∅ is the empty set and thus very important

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 32 / 50

 Operations on Sets
Connection to Traditional Logic

A ∩ B = {x ∈ X | x ∈ A ∧ x ∈ B}
A ∪ B = {x ∈ X | x ∈ A ∨ x ∈ B}
Ac = {x ∈ X | x ∈
/ A} = {x ∈ X | ¬(x ∈ A)}

A ⊆ B if and only if (x ∈ A) → (x ∈ B) for all x ∈ X

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 33 / 50

 Boole’s Theory I

• George Boole assumed two classes

1. class of all properties
2. class of all objects
• he established two laws connecting properties
with objects
1. (Boolean) algebra on first class w.r.t. the words
and, or, not, implies
2. to any property A corresponds the set  of all
objects possessing this property George Boole (1815-1864)

• “x is A” means object x has property A

• then  = {x | x is A}

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 34 / 50

 Boole’s Theory II
The Characteristic Function

• if X denotes universe of all objects

• then subset  can be defined by the characteristic function
(
1 if x ∈ Â
χ (x) =
0 if x ∈
/ Â

1 )

IR
0 [
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Characteristic function of interval [0, 1.05)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 35 / 50

 Menger’s Theory

• element-set relation should be replaced by

probability of an element belonging to a set
• so, he defined a notion hazy set
• qualities should be replaced by quantities
• hazy set A is defined as function pA : X 7→ [0, 1]
• interpret pA (x) as probability s.t. x belongs to A
Karl Menger (1902-1985)

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 36 / 50

 Outline

1. Motivation

2. Vagueness, Imprecision, Uncertainty

3. From Logic to Fuzzy Logic

4. From Sets to Fuzzy Sets

5. Fuzzy Sets
Membership Functions
Fuzzy Numbers
Linguistic Variables and Linguistic Values
Semantics
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 37 / 50
 Membership Functions I
Lotfi A. Zadeh (1965)
“A fuzzy set is a class with a continuum of membership grades.”

• fuzzy set M ⊂ X is characterized by membership function µM

• µM associates real number in [0, 1] with each element x ∈ X
• value of µM at x represents grade of membership of x in M
• fuzzy set M is thus defined as mapping

µM : X 7→ [0, 1]

⇒ µM generalizes traditional characteristic function

χM : X 7→ {0, 1}

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 38 / 50

 Membership Functions II
• µM (u) = 1 reflects full membership in M
• µM (u) = 0 expresses absolute non-membership in M
• sets can be viewed as special case of fuzzy sets where
• only full membership and absolute non-membership are allowed
• such sets are called crisp sets or Boolean sets
• membership degrees 0 < µM < 1 represent partial membership

µM

0 Age
0 20 40
Representing young in “a young person”
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 39 / 50
 Membership Functions III
• membership function attached to given word (such as young)
depends on context
• young retired person is certainly older than young student
• even idea of young student depends on the user

• membership degrees are fixed only by convention

• unit interval as range of membership grades is arbitrary
• natural for modeling membership grades of fuzzy sets of real
numbers
• membership degree as degree of proximity between x and
prototypes
• prototypes are elements v of M with µM (v ) = 1
• membership decreases as elements are located farther from
prototypes

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 40 / 50

 Membership Functions IV
• consider again representation for predicate young
• there is no precise threshold between
1. prototypes of young and
2. prototypes of not young
• fuzzy sets offer natural interface between linguistic and numerical
representations

µM

0 Age
0 20 40
Representing young in “a young person”
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 41 / 50
 Examples of Fuzzy Sets I
Body Height of Four Year Old Boys

1 µtall

0.65

h [m]
0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Membership function for µtall

• 1.5 m is for sure tall, 0.7 m is for sure small, but in-between?!
• vague predicate tall modeled as sigmoid function
• e.g., height of 1.1 m has membership degree of 0.65
• so, height of 1.1 m satisfies predicate tall with 0.65
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 42 / 50
 Examples of Fuzzy Sets II
Velocity of Rotating Hard Disk

1.0 µ

b b
0.5
v
0
a b x c d

Fuzzy set µ characterizing velocity of rotating hard disk

• let x be velocity v of rotating hard disk in revolutions per minute
• if no observations about x available, use expert’s knowledge
• “It’s impossible that v drops under a or exceeds d.”
• “It’s highly certain that any value between [b, c] can occur.”
• additionally, values of v with membership degree of 0.5 are
provided
• interval [a, d] is called support of the fuzzy set
• interval [b, c] is denoted as core of the fuzzy set
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 43 / 50
 Fuzzy Numbers
Exactly Two and Around Two

1 µ b
1 µ

IR IR
0 0
0 1 2 3 4 0 1 2 3 4
exactly two around two

• exact numerical value has membership degree of 1

• left: monotonically increasing, right: monotonically decreasing
⇒ unimodal function
• terms like around modeled using triangular or Gaussian function

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 44 / 50

 Linguistic Variables and Linguistic Values
Living Area of an Appartment

• linguistic variables represent attributes in fuzzy systems

• they are partitioned into linguistic values (not numerical!)
• partition is usually chosen subjectively (based on human intuition)
• all linguistic values have a meaning, not a precise numerical value
µt µs µm µl µh
1

0
0 10 15 25 30 40 45 55 60 [m2 ]
• linguistic variable living area A stores the linguistic values
• e.g., tiny, small, medium, large, huge
• every x ∈ A has µ(x) ∈ [0, 1] to each value, e.g., let a = 42.5m2
⇒ µt (a) = µs (a) = µh (a) = 0, µm (a) = µl (a) = 0.5
R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 45 / 50
 Semantics of Fuzzy Sets

What membership grades may mean?

• fuzzy sets are relevant in three types of information-driven tasks

1. classification and data analysis
2. decision-making problems
3. approximate reasoning

• these three tasks exploit three semantics of membership grades

1. similarity
2. preference
3. uncertainty

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 46 / 50

 Degree of Similarity (Zadeh, 1966)

• oldest interpretation of membership grades

• µ(u) is degree of proximity of u from prototype elements of µ
• goes back to interests of fuzzy set concept in pattern classification

• still used today for cluster analysis, regression, etc.

• here, proximity between pieces of information is modeled
• also, in fuzzy control: similarity degrees are measured between
• current situation and
• prototypical ones

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 47 / 50

 Degree of Preference (Zadeh, 1970)

• µ represents both
• set of more or less preferred objects
• values of a decision variable X
• µ(u) represents both
• intensity of preference in favor of object u
• feasibility of selecting u as value of X
• fuzzy sets then represent criteria or flexible constraints

• has been used in

• fuzzy optimization (especially fuzzy linear programming) and
• decision analysis
• typical applications: engineering design and scheduling problems

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 48 / 50

 Degree of Uncertainty (Zadeh, 1978)

• this interpretation implicitly proposed by Zadeh when he

• introduced possibility theory and
• developed his theory of approximate reasoning

• µ(u) can be viewed as

• degree of plausibility that parameter X has value u
• given the only information “X is µ”

• then support values are mutually exclusive and

• membership degrees rank these values by their plausibility
• this view has been used in expert systems and artificial intelligence

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 49 / 50

 Semantics of Fuzzy Sets – Examples
• classifying cars of known dimensions into big, regular, small
• computation of membership degree of each car to category big by
• choosing prototype of big car
• measuring distance between car and prototype

• buying a big car

• membership grade of our dream to class of big cars
= degree of satisfaction according to criterion “size”

• somebody says that (s)he just saw a big car (so, what is known)
• membership grade of car to class of big cars
= degree of plausibility that this car is same one observed
• high membership degree ⇒ confidence that car is known might be
low
• low membership degree ⇒ implausible candidate can be rejected

R. Kruse, C. Moewes Fuzzy Systems – Introduction 2009/10/21 50 / 50