Upload
brian-mccormick
View
75
Download
3
Tags:
Embed Size (px)
DESCRIPTION
Fuzzy sets II. Prof. Dr. Jaroslav Ramík. Content. Extension principle Extended binary operations with fuzzy numbers Extended operations with L-R fuzzy numbers Extended operations with t-norms Probability, possibility and fuzzy measure Probability and possibility of fuzzy event - PowerPoint PPT Presentation
Citation preview
Fuzzy sets II 1
Fuzzy sets II
Prof. Dr. Jaroslav Ramík
Fuzzy sets II 2
Content
• Extension principle• Extended binary operations with fuzzy numbers• Extended operations with L-R fuzzy numbers• Extended operations with t-norms• Probability, possibility and fuzzy measure• Probability and possibility of fuzzy event• Fuzzy sets of the 2nd type• Fuzzy relations
Fuzzy sets II 3
Extension principle (EP)by L. Zadeh, 1965
• EP makes possible to extend algebraical operations with NUMBERS to FUZZY SETS
• Even more: EP makes possible to extend REAL FUNCTIONS of real variables to FUZZY FUNCTIONS with fuzzy variables
• Even more: EP makes possible to extend CRISP CONCEPTS to FUZZY CONCEPTS
(e.g. relations, convergence, derivative, integral, etc.)
Fuzzy sets II 4
Example 1. Addition of fuzzy numbers N~
M~
S~
N~
M~
S~
EP:
Fuzzy sets II 5
Theorem 1.
Let
• the operation denotes + or · (add or multiply)
• - fuzzy numbers, [0,1]
• - -cuts
Then is defined by its -cuts as follows
N~
,M~
]d,c[N~
],b,a[M~
]d*c,b*a[S~
N~~M
~S~
[0,1]
Fuzzy sets II 6
Extension principle for functions• X1, X2,…,Xn, Y - sets
• n - fuzzy sets on Xi , i = 1,2,…,n
• g : X1X2 …Xn Y - function of n variables
i.e. (x1,x2 ,…,xn ) y = g (x1,x2 ,…,xn )
Then the extended function
is defined by
iA~
otherwise0
)y(gif)x(),...,x(MinSup)y(
1nA1A
)y(g)x,...,(xBn11-
n1
)A~
,...,A~
,A~
(g~B~
)A~
,...,A~
,A~
( n21n21
Fuzzy sets II 7
Remarks
• g-1(y) = {(x1,x2 ,…,xn ) | y = g (x1,x2 ,…,xn )} - co-image of y
• Special form of EP: g (x1,x2) = x1+x2 or g (x1,x2) = x1*x2
• Instead of Min any t-norm T can be used - more general for of EP
)y(),x(MinSup2R)y,x(
)z( NMNM
yxz
Fuzzy sets II 8
Example 2. Fuzzy Min and Max
))y(),x((MinSup)z( NM)y,x(Maxz
Max
))y(),x((MinSup)z( NM)y,x(Minz
Min
)N~
,M~
(ni~
M)N~
,M~
(xa~M
Fuzzy sets II 9
Extended operations with L-R fuzzy numbers
• L, R : [0,+) [0,1] - decreasing functions - shape functions
• L(0) = R(0) = 1, m - main value, > 0, > 0
• = (m, , )LR - fuzzy number of L-R-type if
•
.mxifmx
R
,mxifxm
L
)x(A~
A~
Left spread Right spread
Fuzzy sets II 10
Example 3. L-R fuzzy number “About eight”
28xA e)x( 1,8nm,e)x(R)x(L
2mx
Fuzzy sets II 11
Example 4.
L(u) = Max(0,1 ‑ u) R(u) = 2u1
1
LRLR )1,2,3(),,m(M~
Fuzzy sets II 12
Theorem 2.Let
= (m,,)LR , = (n,,)LR
where L, R are shape functions
Then is defined as
Example: (2,3,4)LR (1,2,3)LR = (3,5,7)LR
N~
M~
N~~M
~S~
LR),,nm(S~
~
Addition
Fuzzy sets II 13
Opposite FN
= (m,,)LR - FN of L-R-type
= (m,, )LR - opposite FN of L-R-type to
M~
M~~ M
~
“Fuzzy minus” M~
Fuzzy sets II 14
Theorem 3.Let
= (m,,)LR , = (n,,)LR
where L, R are shape functions
Then is defined as
Example: (2,3,4)LR (1,2,3)LR = (1,6,6)LR
N~
M~
N~~M
~S~
LR),,nm(S~
~
Subtraction
Fuzzy sets II 15
Example 5. Subtraction
Fuzzy sets II 16
Theorem 4.Let
= (m,,)LR , = (n,,)LR
where L, R are shape functions
Then is defined by approximate formulae:
Example by 1.: (2,3,4)LR (1,2,3)LR (2,7,10)LR
N~
M~
N~~M
~S~
LR).m.n,.m.n,n.m(S~
~
Multiplication
LR)..m.n,..mn,n.m(S~
1.
2.
Fuzzy sets II 17
Example 6. Multiplication
M~
N~
= (2,1,2)LR , = (4,2,2)LR
N~~M
~S~
(8,8,12)LR LR)16,6,8(S~
formula 1. - - - - formula 2. ……. exact function
Fuzzy sets II 18
Inverse FN
= (m,,)LR > 0 - FN of L-R-type
- approximate formula 1
- approximate formula 2
M~
RL221 )
m,
m,
m
1(M
~
RL1 )
)m(m,
)m(m,
m
1(M
~
11 M~~)M
~~(
We define inverse FN only for positive (or negative) FN !
Fuzzy sets II 19
Example 7. Inverse FN
1M
~= (2,1,2)LR
RL1
1 )4
1,
2
1,
2
1(M
~ RL
11 )
2
1,
4
1,
2
1(M
~
formula 1. - - - - formula 2. ……. exact function
f.1: f.2:
Fuzzy sets II 20
= (m,,)LR , = (n,,)LR > 0
where L, R are shape functions
Define
Combinations of approximate formulae, e.g.
N~M
~
1N~~M
~N~
/~
M~
S~
LR))n(n
nm,
)n(n
nm,
n
m(S
~
Division
Fuzzy sets II 21
Probability, possibility and fuzzy measure
Sigma Algebra (-Algebra) on :
F - collection of classical subsets of the set satisfying:
(A1) F
(A2) if A F then CA F
(A3) if Ai F, i = 1, 2, ... then i Ai F
- elementary space (space of outcomes - elementary events)
F - -Algebra of events of
Fuzzy sets II 22
Probability measure
F - -Algebra of events of p : F [0,1] - probability measure on Fsatisfying:(W1) if A F then p(A) 0 (W2) p() = 1
(W3) if Ai F , i = 1, 2, ..., Ai Aj = , ij
then p(i Ai ) = i p(Ai ) - -additivity
(W3*) if A,B F , AB= ,
then p(AB ) = p(A ) + p(B) - additivity
Fuzzy sets II 23
Fuzzy measure
F - -Algebra of events of g : F [0,1] - fuzzy measure on F
satisfying:
(FM1) p() = 0
(FM2) p() = 1
(FM3) if A,B F , AB then p(A) p(B) - monotonicity
(FM4) if A1, A2,... F , A1 A2 ...
then g(Ai ) = g( Ai ) - continuityi
limi
lim
Fuzzy sets II 24
Properties
• Additivity condition (W3) is stronger than monotonicity (MP3) & continuity (MP4) i.e.
• (W3) (MP3) & (MP4)
• Consequence: Any probability measure is a fuzzy measure but not contrary
Fuzzy sets II 25
Possibility measure
P() - Power set of (st of all subsets of ) : P() [0,1] - possibility measure on satisfying:(P1) () = 0 (P2) () = 1
(P3) if Ai P() , i = 1, 2, ...
then (i Ai ) = Supi {p(Ai )}
(P3*) if A,B P() ,
then (AB ) = Max{(A ), (B)}
Fuzzy sets II 26
Properties
• Condition (P3) is stronger than monotonicity (MP3) & continuity (MP4) i.e.
• (P3) (MP3) & (MP4)
• Consequence: Any possibility measure is a fuzzy measure but not contrary
Fuzzy sets II 27
Example 8.
A B C A B A C B C
P 0 0,3 0,5 0,2 0,8 0,5 0,7 1
g 0 0,4 0,6 0,3 0,8 0,6 0,7 1 0 0,7 1 0,3 1 0,7 1 1
= ABC
F = {, A, B, C, AB, BC, AC, ABC}
Fuzzy sets II 28
Possibility distribution - possibility measure on P()
• Function : [0,1] defined by
(x) = ({x}) for xis called a possibility distribution on
Interpretation: is a membership function of a fuzzy set , i.e. (x) = A(x) x ,
A(x) is the possibility that x belongs to
A~
Fuzzy sets II 29
Probability and possibility of fuzzy event
Example 1: What is the possibility (probability) that tomorrow will be a nice weather ?
Example 2: What is the possibility (probability) that the profit of the firm A in 2003 will be high ?
• nice weather, high profit - fuzzy events
Fuzzy sets II 30
Probability of fuzzy event Finite universe
={x1, …,xn} - finite set of elementary outcomes
F - -Algebra on P - probability measure on F
- fuzzy set of , with the membership
function A(x) - fuzzy event,
A F for [0,1]
P( ) = - probability of fuzzy event
A~
A~
n
1iiiA })x({P).x(
A~
Fuzzy sets II 31
Probability of fuzzy event Real universe
= R - real numbers - set of elementary outcomes
F - -Algebra on R
P - probability measure on F given by density fction g
- fuzzy set of R, with the membership
function A(x) - fuzzy event
A F for [0,1]
P( ) = - probability of fuzzy event
A~
A~ dx)x(g).x(A
A~
Fuzzy sets II 32
Example 9.
= (4, 1, 2)LR L(u) = R(u) = e-u
A~
.otherwise0
0xfore2)x(g
x2
- density function of random value
dxe2edxe2e)A
~(P x2
4
x24
0
)x4( 24x
225
25
e5
2e2)
e
1e(e2e
5
2e2limee2 2
044
z
5
x2
A
4
0x4
= 0,036
- “around 4”
Fuzzy sets II 33
Possibility of fuzzy event
- set of elementary outcomes
: [0,1] - possibility distribution
- fuzzy set of , with the membership
function A(x) - fuzzy event
A F for [0,1]
P( ) = - possibility of fuzzy event
A~
A~ )}x(),x({MinSup A
x
A~
Fuzzy sets II 34
Fuzzy sets of the 2nd type
• The function value of the membership function is again a fuzzy set (FN) of [0,1]
Fuzzy sets II 35
Example 10.
Fuzzy sets II 36
Example 11.Linguistic variable “Stature”- Height of the body
Stature
Tall Short Very tall Middle Very short
Fuzzy sets II 37
Fuzzy relations
• X - universe• - (binary) fuzzy (valued) relation on X = fuzzy set on XX
is given by the membership function R : XX [0,1]
FR is:
• Reflexive: R (x,x) = 1 xX
• Symmetric: R (x,y) = R (y,x) x,yX
• Transitive: Supz[Min{R (x,z), R (z,y)}] R (x,y)
• Equivalence: reflexive & symmetric & transitive
R~
R~
Fuzzy sets II 38
Example 12.Binary fuzzy relation : “x is much greater than y”
xy10for1
y10xyfory9
yxyxfor0
)y,x(R
e.g. R(8,1) = 7/9 = 0,77…
- is antisymmetric: If R (x,y) > 0 then R (y,x) = 0 x,yX
R~
R~
Fuzzy sets II 39
Example 13.Binary fuzzy relation : “x is similar to y”R
~
x/y 1 2 3 4 51 1,0 0,5 0,3 0,2 02 0,5 1,0 0,6 0,5 0,23 0,3 0,6 1,0 0,7 0,44 0,2 0,5 0,7 1,0 0,85 0 0,2 0,4 0,4 1,0
R~
is equivalence !
X = {1,2,3,4,5}
Fuzzy sets II 40
Summary• Extension principle• Extended binary operations with fuzzy numbers• Extended operations with L-R fuzzy numbers• Extended operations with t-norms• Probability, possibility and fuzzy measure• Probability and possibility of fuzzy event• Fuzzy sets of the 2nd type• Fuzzy relations
Fuzzy sets II 41
References
[1] J. Ramík, M. Vlach: Generalized concavity in fuzzy optimization and decision analysis. Kluwer Academic Publ. Boston, Dordrecht, London, 2001.
[2] H.-J. Zimmermann: Fuzzy set theory and its applications. Kluwer Academic Publ. Boston, Dordrecht, London, 1996.
[3] H. Rommelfanger: Fuzzy Decision Support - Systeme. Springer - Verlag, Berlin Heidelberg, New York, 1994.
[4] H. Rommelfanger, S. Eickemeier: Entscheidungstheorie - Klassische Konzepte und Fuzzy - Erweiterungen, Springer - Verlag, Berlin Heidelberg, New York, 2002.