2004/2/18

1
Page 1
Uncertainty Systems
Ia-4002
Fuzzifiers and Defuzzifiers
Rogelio Soto
Centro de Sistemas Inteligentes
ITESM, Campus Monterrey
http://www-csi.mty.itesm.mx/~rsoto
[email protected]
2
Fuzzy Systems: Fuzzifiers and Defuzzifiers
Outline:
Fuzzifiers
- Singleton, Gaussian and Triangular
Defuzzifiers
- Center of gravity, center average and
Maximum Defuzzifier
- Comparison of the Defuzzifiers
Exercise
Summary
3
Fuzzifiers and Defuzzifiers
In most applications the input and
output of the fuzzy system are real-
valued numbers.
The interfaces between the fuzzy
inference engine and the environment
needto be devised.
These interfaces are the fuzzifier and
defuzzifier
2004/2/18
2
Page 2
4
Fuzzifiers and Defuzzifiers
x in U y in V
Fuzzy sets Fuzzy sets
in U in V
Basic Configuration of fuzzy systems with
fuzzifiers and defuzzifiers
Fuzzy Rule
Base
Fuzzifier Defuzzifier
Fuzzy inference
Engine
5
Fuzzifiers
The fuzzifier is defined as a mapping
from a real-valued point
to a fuzzy set A in U.
The fuzzy set A should have a large
membership value at x*
The fuzzifier should help to suppress the
noise of the input
The fuzzifier should help to simplify the
computations involved in the fuzzy
inference engine
n
R U x *
6
Fuzzifiers
Singleton Fuzzifier
The singleton fuzzifier maps a real-valued
point into a fuzzy singleton A in U,
which has a membership value 1 at x* and 0
at all other points in U

'

Otherwise
if
A
0
* 1
) (
'
x x
x
U * x
2004/2/18
3
Page 3
7
Fuzzifiers
GaussianFuzzifier
The Gaussianfuzzifier maps into a
fuzzy set A in U, which has the following
Gaussian membership function:
2
*
2
1
*
1 1
) (
'

,
_

,
_

n
n n
a
x x
a
x x
A
e e L x
U * x
8
Fuzzifiers
Triangular Fuzzifier
The Triangular fuzzifier maps into a
fuzzy set A in U, which has the following
triangular membership function:

'

,
_

,
_

otherwise
n i b x x if
b
x x
b
x x
A
i i i
n
n n
0
) (
, , 2 , 1 , 1 1
'
*
*
1
*
1 1
K L
x
U * x
9
Fuzzifiers
The singleton fuzzifier simplifies the
computations involved in the fuzzy
inference engine
If the fuzzy sets have Gaussian or
triangular membership functions, then
the Gaussian or triangular fuzzifier
also will simplify some fuzzy inference
engines
2004/2/18
4
Page 4
10
Fuzzifiers
Suppose that a fuzzy rule base has M
rules with Gaussian MFs then the
product inference engine with
algebraic product for the t-norm * is
1
1
1
1
1
]
1

,
_

,
_

n
i
B
a
x x x x
M
l
B
y e e y l
i
i
l
iP
l
i
l
i
l
iP
1
1
) ( max ) (
2
*
2
'

11
Fuzzifiers
And for the minimum inference engine
with min for the t-norm * is
1
1
1
1
1
]
1

,
_

,
_

,
_

) ( , , , min max ) (
2 2
1
1 1
'
1
y e e y l
l
n
l
n
l
nM
l
l l
M
B
x x x x
M
l
B


K
12
Fuzzifiers
The singleton fuzzifier simplifies the
computation in the fuzzy inference engine
for any type of MF of the fuzzy rules
The Gaussian or triangular fuzzifiers
simplifies the computation if the MFs of the
fuzzy rules are Gaussian or triangular,
respectively
The Gaussian and triangular fuzzifiers can
suppress noise in the input
2004/2/18
5
Page 5
13
Defuzzifiers
The defuzzifier is defined as a
mapping from fuzzy set
(which is the output of the fuzzy
inference engine) to a crisp point
Conceptually, the task of the
defuzzifier is to specify a point in V
that best represents the fuzzy set B
R V in B
V y *
14
Defuzzifiers
The following criteria are consider in
choosing a defuzzification method
Plausibility: The point y* should represent B from
an intuitive point of view
Computational simplicity: Specially important for
fuzzy systems designed to operate in real -time
Continuity: A small change in B should not result
in a large change in y*.
15
Defuzzifiers
Three types of defuzzifiers are proposed. It
is assumed that the fuzzy set B is obtained
from one of the fuzzy inference engines
discussed earlier. Also, B is the union or
intersection of M individual fuzzy sets
Center of Gravity
Center Average (Center of Area,
COA)
Maximum
2004/2/18
6
Page 6
16
Center of Gravity Defuzzifier
This method specifies the y* as the
center of the area covered by the MF
of B

V B
V
B
dy y
dy y y
y
) (
) (
*

17
Center of Gravity Defuzzifier
18
Center of Gravity Defuzzifier
The advantage of the center of gravity
lies in its intuitive plausibility
The disadvantage is that it is
computationally intensive, the
integrations may be difficult to
compute for irregular MFs
2004/2/18
7
Page 7
19
Center Average Defuzzifier (Center
of Area, COA)
The fuzzy set B is the union or
intersection of M fuzzy sets, a good
approximation to get y* is the
weighted average of the centers of the
M fuzzy sets, with the weights equal
the heights of the corresponding fuzzy
sets

M
l
l
M
l
l
l
y
y
1
1
*

20
Center Average Defuzzifier
2 1
2
2
1
1
*
w w
w y w y
y
+
+

* y
2
w
1
w
V
1
y
2
y
21
Center Average Defuzzifier
Example 1. Suppose that the fuzzy set
Bis the one of the previous figure
with The center of
average is
but the center of gravity defuzzifier
more complex and computationally
more intensive.
1
2
0
1
y and y
2 1
2
*
w w
w
y
+

2004/2/18
8
Page 8
22
Center Average Defuzzifier
The center of gravity defuzzifier for
this example is:

V B
V
B
dy y
dy y y
y
) (
) (
*

( )
2 1
2 1
2 1
2
2 1
3
1
2 1
2
1
6
1
6 / 1
w w
w w
w w
w w
w
w w
+
+
+
+ +

23
Comparison of Center of Gravity
and Area Defuzzifiers
W1 W2 COG CA Error
0.9
0.9
0.9
0.6
0.6
0.6
0.3
0.3
0.3
0.7
0.5
0.2
0.7
0.5
0.2
0.7
0.5
0.2
0.4258
0.5457
0.7313
0.3324
0.4460
0.6471
0.1477
0.2155
0.3818
0.4375
0.5385
0.7000
0.3571
0.4545
0.6250
0.1818
0.2500
0.4000
0.0275
0.0133
0.0428
0.0743
0.0192
0.0342
0.2308
0.1600
0.0476
24
Maximum Defuzzifier
This method chooses the y* as the
point in V at which achieves its
maximum value
hgt(B) is the set of all points in V at
which achieves its maximum
value
B

'

) ( sup ) ( ) (

y y V y B hgt
B
V y
B

2004/2/18
9
Page 9
25
Maximum Defuzzifier
y* is any point in hgt (B)
If hgt(B) contains more than one
point then:
a) Smallest of the maxima
b) Largest of maxima
c) Mean of maxima
{ ( inf * B hgt y y
{ } ) ( sup * B hgt y y

) (
) (
*
B hgt
B hgt
dy
dy y
y
26
Maximum Defuzzifier
The mean of maxima may give results
which are contradictory to the
intuition, this may be due to non-
convex nature of
B
The method is computationally simple
and plausible
27
Maximum Defuzzifier
2004/2/18
10
Page 10
28
Comparison of the Defuzzifiers
CG CA Max
Plausibility yes yes yes
Computacional
Simplicity
no yes yes
Continuity yes yes no
29
Product Inference Engine and CA
Exercise 1
Consider a two-input-one-output system
that is constructed from the following two
rules:
IF x
1
is A
1
and x
2
is A
2
, THEN y is A
1
IF x
1
is A
2
and x
2
is A
1
, THEN y is A
2

'

otherwise
u if u
u
A
0
1 1 1
) (
1

'

otherwise
u if u
u
A
0
2 0 1 1
) (
2

30
Product Inference Engine and CA
Exercise 1
The input to the system is (x
1
*,x
2
*) =
(0.3,0.6)
Determine the output of the fuzzy system y*
for a singleton fuzzifier, product inference
engine and center average defuzzifier
2004/2/18
11
Page 11
31
Summary
In this session, we have shown:
The definition of singleton, Gaussian and
triangular fuzzifiers
Different methods of defuzzifiers like
center of gravity, center average and
maximum
32
Homework # 5
Due date: February 25, 2003
Problem 7.13, Neural Fuzzy Systems, Lin &
Lee, page 178
http://
www-csi.mty.itesm.mx/~rsoto/Tarea%204.pdf