You are on page 1of 22

Fuzzy Set Theory

Ch. 3: Fuzzy Rules and Fuzzy Reasoning

Outline
Extension principle Fuzzy relations Fuzzy if-then rules Compositional rule of inference Fuzzy reasoning

Extension Principle
A is a fuzzy set on X :

A = A ( x1 ) / x1 + A ( x2 ) / x2 ++ A ( x n ) / x n
he image of A under f( ) is a fuzzy set B:

B = B ( x1 ) / y1 + B ( x2 ) / y2 ++ B ( x n ) / y n
!here yi = f(xi), i = 1 to n. "f f( ) is a many-to-one mapping# then

B ( y ) = max A ( x )
x= f
1

( y)

Fuzzy Relations
$ fuzzy relation R is a %& 'F:
R = {(( x , y ), R ( x , y ))|( x , y ) X Y}

Examples:
x is close to y (x and y are num)ers* x depends on y (x and y are e+ents* x and y loo, ali,e (x# and y are persons or o)-ects* "f x is large# then y is small (x is an o)ser+ed reading and . is a corresponding action*

'ax-'in Composition
he max-min composition of t!o fuzzy relations R1 (defined on X and Y* and R2 (defined on Y and Z* is

R R ( x , z ) = [ R ( x , y ) R ( y , z )]
1 2

Properties:

$ssociati+ity: R ( S T ) = ( R S ) T &istri)uti+ity o+er union:


R (S T ) = ( R S ) ( R T )

/ee, distri)uti+ity o+er intersection: R ( S T ) ( R S ) ( R T ) 'onotonicity: S T ( R S ) ( R T )

'ax-0tar Composition
'ax-product composition:

R R ( x , z ) = [R ( x , y )R ( y , z )]
1 2

"n general# !e ha+e max-1 composition:

R R ( x , z ) = [R ( x , y ) * R ( y , z )]
1 2

!here 1 is a -norm operator.

2inguistic 3aria)les
$ numerical +aria)les ta,es numerical +alues: Age = 65 $ linguistic +aria)les ta,es linguistic +alues: Age is old $ linguistic +alues is a fuzzy set. $ll linguistic +alues form a term set:
(age* 4 5young# not young# +ery young# ... middle aged# not middle aged# ... old# not old# +ery old# more or less old# ... not +ery yound and not +ery old# ...6

2inguistic 3alues ( erms*

compl+.m

Operations on 2inguistic 3alues


Concentration: &ilation: Contrast intensification:
CON ( A) = A2

DIL( A) = A0.5
0 A ( x ) 0.5 2 A2 , INT ( A) = 2 2 ( A ) , 0.5 A ( x ) 1

intensif.m

Fuzzy "f- hen Rules


7eneral format:
"f x is $ then y is 8

Examples:
"f pressure is high# then +olume is small. "f the road is slippery# then dri+ing is dangerous. "f a tomato is red# then it is ripe. "f the speed is high# then apply the )ra,e a little.

Fuzzy "f- hen Rules


!o !ays to interpret 9"f x is $ then y is 8::
y

$ coupled !ith 8

$ entails 8

Fuzzy "f- hen Rules


!o !ays to interpret 9"f x is $ then y is 8::
$ coupled !ith 8: (A and B*
R = A B = A B = A ( x ) B ( y )|( x , y )
~

$ entails 8: (not A or B* - Material implication - Propositional calculus - Extended propositional calculus - Generalization of modus ponens

Fuzzy "f- hen Rules


Fuzzy implication function:
R ( x , y ) = f ( A ( x ), B ( y )) = f ( a, b )
$ coupled !ith 8

fuzimp.m

Fuzzy "f- hen Rules


$ entails 8

fuzimp.m

Compositional Rule of "nference


&eri+ation of y = b from x = a and y = f(x):
y y

b y = f(x) a
a and b: points y = f(x) : a cur+e
x

b y = f(x)
x

a
a and b: inter+als

y = f(x) : an inter+al-+alued function

Compositional Rule of "nference


a is a fuzzy set and y = f(x) is a fuzzy relation:

cri.m

Fuzzy Reasoning
0ingle rule !ith single antecedent
Rule: if x is $ then y is 8 Fact: x is $; Conclusion: y is 8;

7raphic Representation:
$; $
!

8
< .

$;
x is $; <

8;
y is 8; .

Fuzzy Reasoning
0ingle rule !ith multiple antecedent
Rule: if x is $ and y is 8 then z is C Fact: x is $; and y is 8; Conclusion: z is C;

7raphic Representation:
$; $
<

8; 8
.

-norm
!

C%
=

$;
x is $; <

8;
y is 8; .

C;
z is C; =

Fuzzy Reasoning
'ultiple rules !ith multiple antecedent
Rule >: if x is $> and y is 8> then z is C> Rule %: if x is $% and y is 8% then z is C% Fact: x is $; and y is 8; Conclusion: z is C;

7raphic Representation: (next slide*

Fuzzy Reasoning
7raphics representation:
$; $>
<

8; 8>
!>

C>
. =

$; $%
<

8; 8%
!%

C%
. =
-norm

$;
x is $; <

8; C;
y is 8; . z is C; =

Fuzzy Reasoning: '$ 2$8 &emo


?? rule+ie! mam%>

Other 3ariants
0ome terminology:
&egrees of compati)ility (match* Firing strength @ualified (induced* 'Fs O+erall output 'F

You might also like