B219 Intelligent Systems
Fuzzy Expert Systems
What is fuzzy thinking?
§ Experts rely on common sense when they solve
problems
§ How can we represent expert knowledge that uses
vague and ambiguous terms in a computer?
§ Fuzzy logic is not logic that is fuzzy, but logic that is
used to describe fuzziness. Fuzzy logic is the theory
of fuzzy sets, sets that calibrate vagueness.
§ Fuzzy logic is based on the idea that all things admit
of degrees. Temperature, height, speed, distance,
beauty – all come on a sliding scale.
§ The motor is running really hot. Tom is a very tall
guy.
§ Boolean logic used sharp distinctions. It forces us to
draw lines between members of a class and non-
members.
§ For instance, we may say, Tom is tall because his
height is 181 cm. If we drew a line at 180cm, we
would find that David, who is 179cm, is short.
Week 6 Lecture Notes page 1 of 1
B219 Intelligent Systems
§ Fuzzy Logic reflects how people think. It attempts to
model our sense of words, our decision making and
our common sense. As a result, it is leading to new,
more human, intelligent systems.
What are fuzzy systems?
§ It works on fuzzy logic, which superset of
conventional (Boolean) logic that has been extended
to handle the concept of partial truth -- truth values
between "completely true" and "completely false".
§ It provides a systematic, intuitive and mathematical
means of handling uncertainty in natural and artificial
systems
Different types of uncertainty:
§ Classical uncertain
o “Will I get a HD for this unit?”
o Uncertain and precise
o Mostly can be handle by probability theory
§ Vague
o “Steve is tall”
o Certain but imprecise
o Can be handle by fuzzy sets and fuzzy logic
Week 6 Lecture Notes page 2 of 2
B219 Intelligent Systems
§ Imprecise
o “Mark weighs between 50kg and 65 kg”
o Uncertain and imprecise
o Probability or possibilities theory
Fuzzy Logic?
§ Fuzzy Logic is a set of mathematical principles for
knowledge representation based on degrees of
membership.
§ Unlike two-valued Boolean logic, fuzzy logic is multi-
valued. It deals with degrees of membership and
degrees of truth.
§ Fuzzy logic uses the continuum of logical values
between 0 (completely false) and 1 (completely true).
Instead of just black and white, it employs the
spectrum of colours, accepting that things can be
partly true and partly false at the same time.
Week 6 Lecture Notes page 3 of 3
B219 Intelligent Systems
What are fuzzy sets?
§ is a class of objects in which there is no sharp
boundary between those objects that belong to the
class and those that do not. (ì – membership)
§ The classical example in fuzzy sets is tall men. The
elements of his fuzzy set “tall men” are all men, but
their degrees of membership depend on their height.
Week 6 Lecture Notes page 4 of 4
B219 Intelligent Systems
§ The x-axis represents the universe of discourse – the
range of all possible values applicable to a chosen
variable. In our case, the variable is the man height.
According to this representation, the universe of
men’s heights consists of all tall men.
§ The y-axis represents the membership value of the
fuzzy sets. In our case, the fuzzy set of “tall men”
maps height values into corresponding membership
values.
Week 6 Lecture Notes page 5 of 5
B219 Intelligent Systems
§ Crisp set A = {1, 2, 3}, ì 1(A)=1, ì 5=(A)=0
A 1
3
2
5
§ Fuzzy set A={1.0/1, 0.6/2, 0.1/5}, ì 1(A)=1, ì 5(A)=0.1
A 2
1
5
Week 6 Lecture Notes page 6 of 6
B219 Intelligent Systems
Fuzzy memberships
§ distribution of truth of a variable
µ(x)
low normal raised strong
1
0
36°C 37°C 38°C 39°C 40°C 41°C 42°C
§ The distribution of the four membership functions are
as follows:-
low temp={(1,35),(1,36),(0,37)}
normal temp={(0,36),(1,36.8),(0,37.2)}
raised temp={(0,37),(1,37.8),(0.9,38),(0,39.2)}
strong fever={(0,37.5),(0.5,39.5),(0.9,41)}
§ A person with temperature at 38 C has pretty much
raised temperature but just slightly strong fever
§ The membership value at 38 C for raised
temperature is 0.9 and for strong fever is 0.1, and for
others is 0.
§ show the degree of truth or confidence
Week 6 Lecture Notes page 7 of 7
B219 Intelligent Systems
§ 4 common types of fuzzy membership functions:
o triangular (3 parameters) trapezoidal(4 para)
o gaussian (2 parameters)
o generalised bell (3 parameters)
Week 6 Lecture Notes page 8 of 8
B219 Intelligent Systems
Logic operation
§ is a superset of standard Boolean logic
§ in fuzzy logical reasoning, AND is the min operation,
OR is the max operation, and NOT becomes 1-A
§ in the case if A=0.5, B=0.7, and C = A OR B, then C
= max(0.5,0.7)=0.7
§ if we are looking for F = A and B, then
F=min(0.5,0.7)=0.5
Week 6 Lecture Notes page 9 of 9
B219 Intelligent Systems
Linguistic variables and hedges
§ At the root of fuzzy set theory lies the idea of
linguistic variables
§ A linguistic variable is a fuzzy variable. For example,
the statement “John is tall” implies that the linguistics
variable John takes the linguistic value tall.
§ In fuzzy expert systems, linguistic variables are used
in fuzzy rules.
Main Components of a fuzzy system
§ There are basically three main components:-
1. Fuzzification – convert values to fuzzy inputs by
using membership functions
2. Fuzzy inference – using fuzzy rules to form
fuzzy output
3. Defuzzification – using output membership
functions and defuzzification techniques to
produce system output value.
• the commonly used defuzzification method is
known as centroid
Week 6 Lecture Notes page 10 of 10
B219 Intelligent Systems
Measured Variables 2. Fuzzy-Inference Command Variables
(Linguistic Values) (Linguistic Values)
Linguistic
Level
1. Fuzzification 3. Defuzzification
Numerical
Level
Measured Variables Plant Command Variables
(Numerical Values) (Numerical Values)
Week 6 Lecture Notes page 11 of 11
B219 Intelligent Systems
Fuzzy Rules
§ Using if-then rule statements to formulate the
conditional statements that comprise fuzzy logic.
§ A single fuzzy if-then rule assumes the form
if x is A then y is B
where A and B are linguistic values defined by
fuzzy sets on the ranges (universes of discourse) X
and Y, respectively
§ Example:
if service is good then tip is average
Week 6 Lecture Notes page 12 of 12
B219 Intelligent Systems
Fuzzy inference systems
§ Fuzzy inference is the process of formulating the
mapping from a given input to an output using
fuzzy logic
§ The process of fuzzy inference involves all of the
pieces that are described in the previous slides:
membership functions, fuzzy logic operators, and
fuzzy rules
Week 6 Lecture Notes page 13 of 13
B219 Intelligent Systems
§ The two main types of inference systems:-
1. Mamdani-type
2. Sugeno-type
§ These two types of inference systems vary somewhat
in the way outputs are determined
Week 6 Lecture Notes page 14 of 14
B219 Intelligent Systems
Advantages of using fuzzy logic
§ Conceptually easy to understand.
The mathematical concepts behind fuzzy reasoning
are very simple.
§ Flexible.
You can modify and add on fuzzy rules without
starting from scratch.
§ Tolerant of imprecise data.
Everything is imprecise if you look closely enough,
but more than that, most things are imprecise even on
careful inspection.
§ Can model nonlinear functions of arbitrary
complexity.
Can create a fuzzy system to match any set of input-
output data.
§ Can be built on top of the experience of experts.
§ Is close to natural language.
Week 6 Lecture Notes page 15 of 15
B219 Intelligent Systems
Disadvantages of using fuzzy logic
§ Creating the fuzzy rules base
It is difficult to create the fuzzy rules base from input-
output data if no fuzzy rule extraction technique is
used
§ Accuracy of the inference depends directly to the
number of fuzzy rules used in complex problem
§ The increase in input variables and fuzzy membership
used will increase the number of fuzzy rules
exponentially.
Number of fuzzy rules = MI
where M = number of membership function
I = number of input variables
Week 6 Lecture Notes page 16 of 16
B219 Intelligent Systems
Fuzzy Clustering
§ Traditionally, each data point is said to be belonging
to a cluster or not belonging to a cluster
§ In fuzzy clustering, each data point belongs to a
cluster to some degree that is specified by a
membership grade.
§ Non-overlapping cluster
Cluster 1 Cluster 2 Cluster 3
John 1 0 0
Mary 0 0 1
Terry 0 1 0
Patrick 1 0 0
§ Overlapping cluster
Cluster 1 Cluster 2 Cluster 3
John 1 1 0
Mary 0 0 1
Terry 0 1 1
Patrick 1 0 1
Week 6 Lecture Notes page 17 of 17
B219 Intelligent Systems
§ Fuzzy clustering
Cluster 1 Cluster 2 Cluster 3
John 0.5 0.4 0.1
Mary 0.1 0.1 0.8
Terry 0 0.6 0.4
Patrick 0.6 0.1 0.3
§ Fuzzy C-means (FCM) originally introduced by Jim
Bezdek in 1981 is popular in generating fuzzy
clusters.
Week 6 Lecture Notes page 18 of 18
