SlideShare a Scribd company logo
1 of 21
• Fuzzy set  Fuzzy measure
To which degree does an What is the degree of evidence,
individual belong to a set that an individual belongs to a
defined in an inexact way? certain (crisp) set?
• Fuzzy measure g: P(X)  [0,1]
If g(A)g(B) BP(X) then A is the "best guess"
 E.g.: diagnose of a patient with given symptoms
• Usual interpretation: subjective probability,
however: properties are less strict!
• Axioms of fuzzy measures:
g1. g()=0 and g(X)=1 (boundary conditions)
g2. " A,BP(X): if A  B then g(A)g(B) (monotonicity)
g3. " sequence (AiP(X)  i, Ai X):
if A1A2... or A1A2... then
(continuity)
i
i
i
i
g A g A
 

lim lim
( ) ( )
4 / 1
• More general definition of fuzzy measure g:
g: B  [0,1]
where BP(X) is a family of subsets:
1. B and XB
2. If AB then B
3. If A,BB then ABB
(B is Borel-field or s-field)
 As ABA and ABB, we have
max(g(A),g(B))g(AB)
also ABA and ABB, so
min(g(A),g(B))g(AB)
• Fuzzy measures are "too general". Practical considerations lead to some
restrictions, so we obtain special classes of measures.
 Clearly, probability is one of them!
4 / 2
• Belief measure :
Bel: P(X)  [0,1]
Axioms: g1-g3 and
g4Bel.:
for arbitrary AiP(X)
 n=2 Bel(A1A2) Bel(A1)+Bel(A2)-Bel(A1A2)
n=3 Bel(A1A2A3)Bel(A1)+Bel(A2)+Bel(A3)-Bel(A1A2)-Bel(A1A3)-
-Bel(A2A3)+Bel(A1A2A3)
• Interpretation: Bel(A) is the degree of Belief that a particular xX belongs to A
• If (AiAj)= then
Probability is a special Bel measure!
 g4Belg2 (not independent axiom) so {g1, g3, g4Bel } defines Bel measures
Bel A A Bel A
n i
i
n
( ... ) ( )
1
1
  


Bel A A Bel A Bel A A Bel A A A
n i
i
i j
i j
n
n
( ... ) ( ) ( ) ... ( ) ( ... )
1
1
1 2
1
          
 


4 / 3
• Plausability measure :
Pl: P(X)  [0,1]
Axioms: g1-g3 and
g4Pl.:
for arbitrary AiP(X), n
 Pl is dual with Bel!
,
(*)
 Connection of g4Bel and g4Pl:
(*) applied to g4Bel we obtain:
I.e. g4Pl for
)
...
(
)
1
(
...
)
(
)
(
)
...
( 2
1
1
1
1 n
n
j
i
j
i
n
i
i
n A
A
A
Pl
A
A
Pl
A
Pl
A
A
Pl 









 




Pl A Bel A
( ) ( )
 
1
Bel A Pl A
( ) ( )
 
1
1 1 1 1 1
1 1
1
1
         
  


   
Pl A Pl A Pl A A Pl A
i
n
i i
i
n
i j
i j
n
i
n
i
( ) ( ( ) ( ( )) ... ( ) ( ( ))
Pl A Pl A Pl A A Pl A
i
n
i i
i
n
i j
i j
n
i
n
i
( ) ( ) ( ) ... ( ) ( )
  


   
     
1 1
1
1
1
Ai
4 / 4
 From g4Bel-Pl we obtain:
• Every Bel and its dual Pl can be expressed by the "basic (probabilistic)
assignment" m:
m: P(X)  [0,1],
m()=0 and
m is the degree of evidence that xX belongs to A but not to any particular
subset of A.
 Different from probabilistic p: x  [0,1] !
 m(X)=1 not necessarily!
m(A) m(B) when AB: not necessarily!
m(A) and m( ) have no relation
m is not a fuzzy measure!
• (**)
Bel A Bel A
( ) ( )
 1
Pl A Pl A
( ) ( )
  1
m A
( ) 

 1
A P(X)
A



A
B
B
m
A
Bel )
(
)
(
Pl A m B
B A
( ) ( )

 

4 / 5
Interpretation:
m(A): degrees of evidence that xX belongs to A alone
Bel(A): degrees of evidence that xX belongs to some subsets of A
Pl(A): degrees of evidence that xX belongs to A or any subsets of X which
overlaps with A
 Pl(A)Bel(A) " AP(X)
• If m(A)>0 : A is a focal element (of m)
• Body of evidence: <F, m>
F: set of focal elements
m: basic assignment
 F={X}, m(X)=1 : total ignorance
 Bel(X)=1, Bel(A)=0 " AX
 Pl()=0, Pl(A)=1 " A
• Simple support function (focused at A):
m(A)=s, m(X)=1-s,
and m(B)=0 " BA,X
4 / 6
 Basic assignment derived from Bel:
 Example X={a,b,c}
P(X)={,{a},{b},{c},{a,b},{a,c},{b,c},{a,b,c}}
m: {a}  0.1
{b}  0.2
{a,b}  0.3 F={{a},{b},{a,b},{b,c},{a,b,c}}
{b,c}  0.3
{a,b,c}  0.1
Bel({a,b})=m()+m{a}+m{b}+m{a,b}=0+0.1+0.2+0.3=0.6
Pl({a,b})=m{a,b,c}+m{a,b},+m{a,c}+m{b,c}+m{a}+m{b}=
=0.1+0.3+0+0.3+0.1+0.2=1.0
Bel({c})=m()+m{c}=0+0=0
Pl({c})=m{a,b,c}+m{a,c}+m{b,c}+m{c}=0.1+0+0.3+0=0.4
Bel()=0 Bel({a})=0.1 Bel({b})=0.2
Bel({a,c}=0.1 Bel({b,c})=0.5 Bel({a,b,c})=1
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
m()=0 m({a})=Bel({a})=0.1
m({b})=Bel({b})=0.2 m({c})=0
m({a,b})=Bel({a,b})-Bel({a})-Bel({b})=0.6-0.1-0.2=0.3
Etc.
m A Bel B
A B
B A
( ) ( ) ( )
  

 1
4 / 7
• Evidence from two independent sources: m1,m2
Joint basic assignment m1,2:
A (***)
m1,2()=0 (Dempster's rule of combination)
The sum of products m1(B)m2(C) for all focal elements of B (m1) and C (m2) so that
BC, equals (in the denominator); this is why the sum of
products is normalized
 Example (from Klir-Folger textbook)
Assume that an old painting was discovered that strongly resembles paintings by Raphael. Such a discovery
is likely to generate various questions regarding the status of the painting. Assume the following three
questions:
1. Is the discovered painting a genuine painting by Raphael?
2. Is the discovered painting a product of one of Raphael's many disciples?
3. Is the discovered painting a counterfeit?
Let R, D and C denote subsets of our universal set X - the set of paintings - that contain the set of all
paintings by Raphael, the set of all paintings by disciples of Raphael, and the set of all counterfeits of
Raphael's paintings, respectively.
Assume now, that two experts performed careful examinations of the painting and subsequently provided us
with basic assignments m1 and m2 specified in the table below. These are the degrees of evidence that each
expert obtained by the examination and that support the various claims that the painting belongs to one of the
sets of our concern. For example, m1(RD)=0.15 is the degree of evidence obtained by the first expert that
the painting was done by Raphael himself or that the painting was done by one of his disciples.
m A
m B m C)
m B m C)
B C A
B C
1 2
1 2
1 2
1
, ( )
( ) (
( ) (


 
 


1 1 2
 
 
m B m C)
B C
( ) (
4 / 8
U
singEq. (**) wecan easilycalculatethetotal evidence, Bel1 and Bel2, in each set, as shown in thetable.
Applying Dempster's rule (Eq. (***)) to m1 and m2, we obtain the joint basic assignment m1,2, which is also
shown in thetable. To determinethevalues of m1,2, wecalculatethenormalization factor 1-Kfirst:

K m B m C
B C
 
 
1 2
( ) ( )
K=m1(R)m2(D)+m1(R)m2(C)+m1(R)m2(DC)+m1(D)m2(R)+m1(D)m2(C)+m1(D)m2(RC)+m1(C)m2(R)+
 +m1(C)m2(D)+m1(C)m2(RD)+m1(RD)m2(C)+m1(RC)m2(D)+m1(DC)m2(R)=0.03
Thenormalization factor is then 1-K=0.97. Values of m1,2 arecalculated byEq. (***). For example:
m1,2(R)=[m1(R)m2(R)+m1(R)m2(RD)+m1(R)m2(RC)+m1(R)m2(RDC)+m1(RD)m2(R)+
 +m1(RD)m2(RC)+m1(RC)m2(R)+m1(RC)m2(RD)+m1(RDC)m2(R)]/0.97=0.21
m1,2(D)=[m1(D)m2(D)+m1(D)m2(RD)+m1(D)m2(DC)+m1(D)m2(RDC)+m1(RD)m2(D)+
 +m1(RD)m2(DC)+m1(DC)m2(D)+m1(DC)m2(RD)+m1(RDC)m2(D)]/0.97=0.01
m1,2(RC)=[m1(RC)m2(RC)+m1(RC)m2(RDC)+m1(RDC)m2(RC)]/0.97=0.2
m1,2(RDC)=[m1(RDC)m2(RDC)]/0.97=0.31
and similarly for the remaining
focal elements C, RD and DC.
The joint basic assignment can now
be used to calculate the joint Belief
Bel1,2 (see table) and joint
plausibilityPl1,2.

Expert 1 Expert 2 Combined evidence
Focal
elements m1 Bel1 m2 Bel2 m1,2 Bel1,2
R 0.05 0.05 0.15 0.15 0.21 0.21
D 0 0 0 0 0.01 0.01
C 0.05 0.05 0.05 0.05 0.09 0.09
RD 0.15 0.2 0.05 0.2 0.12 0.34
RC 0.1 0.2 0.2 0.4 0.2 0.5
DC 0.05 0.1 0.05 0.1 0.06 0.16
RDC 0.6 1 0.5 1 0.31 1
Combination of degrees of evidence fromtwo independent sources
4 / 9
• Marginal basic assignment:
Given m: P(X×Y)  [0,1],
Focal elements are binary relations R on X×Y
If RX is RX and RY is RY, the projections of m on X and Y, resp. are the
marginal basic assignments:
" AP(X)
" AP(X) (****)
• Marginal bodies of evidence:
<FX,mX>, <FY,mY>
• Noninteractive marginal basic assignment:
m(A×B)=mX(A)mY(B) " AFX, " BFY
and m(R)=0 for RA×B
m A m R
X
R A RX
( ) ( )
:
 

m B m R
Y
R B RY
( ) ( )
:
 

4 / 10
 Example: body of evidence (only focal elements are given)
This is also a case of noninteractive marginal basic assignments
E.g.: m(R1)=0.0625 m(R1)=mX({2,3})*mY({b,c})=0.25*0.25
{2,3}x{b,c}={2b,2c,3b,3c}=R1
Or m(R7)=0.0375 m(R7)=mX({1,3})*mY({a})=0.15*0.25
XxY
1a 1b 1c 2a 2b 2c 3a 3b 3c m(Ri)
R1 = 0 0 0 0 1 1 0 1 1 0.0625
R2 = 0 0 0 1 0 0 1 0 0 0.0625
R3 = 0 0 0 1 1 1 1 1 1 0.125
R4 = 0 1 1 0 0 0 0 1 1 0.0375
R5 = 0 1 1 0 1 1 0 0 0 0.075
R6 = 0 1 1 0 1 1 0 1 1 0.075
R7 = 1 0 0 0 0 0 1 0 0 0.0375
R8 = 1 0 0 1 0 0 0 0 0 0.075
R9 = 1 0 0 1 0 0 1 0 0 0.075
R10 = 1 1 1 0 0 0 1 1 1 0.075
R11 = 1 1 1 1 1 1 0 0 0 0.15
R12 = 1 1 1 1 1 1 1 1 1 0.15
m: P(XxY)  [0,1]
By (****) we obtain:
X
1 2 3 mX(A
)
A = 0 1 1 0.25
1 0 1 0.15
1 1 0 0.3
1 1 1 0.3
mx: P(X)  [0,1]
Y
a b c mY(B)
B = 0 1 1 0.25
1 0 0 0.25
1 1 1 0.5
mY
: P(Y)  [0,1]
4 / 11
• Probability measures
If g4Bel is replaced by g4P:
Bel(AB)=Bel(A)+Bel(B), if AB= (additivity)
we obtain a special class of belief measures: probability measures (Bayesian
belief measures)
 Bel is P iff the basic assignment:
m({x})=Bel({x}) and m(A)=0 if A is not singleton
 Probability measures are fully represented by
p: x  [0,1] probability distribution function (as focal elements are only
singletons)
 " AP(X)
i.e.
• Total ignorance:
• Joint and marginal probability distributions - similar way
 Very Broad field
Bel A Pl A m x
x A
( ) ( ) ({ })
  

Bel A Pl A p x
x A
( ) ( ) ( )
  

 
p x
X
m x
( ) ({ })
 
1
4 / 12
• Nested subsets:
• Consonant Bel / Pl
The family of focal elements is nested (consonant body of evidence <F, m> )
 Given a consonant <F, m> then
" A,B P(X)
• Necessity measure = consonant Bel
Possibility measure = consonant Pl
(very typical – ‘axiomatic’ – prop’s)
 Every possibility measure can be uniquely determined by a possibility distribution f’n
So that
4 / 13
X
A
A
A
A i
n 


 
2
1
))
(
),
(
max(
)
(
))
(
),
(
min(
)
(
B
Pl
A
Pl
B
A
Pl
B
Bel
A
Bel
B
A
Bel




))
(
),
(
max(
)
(
))
(
),
(
min(
)
(
B
A
B
A
B
A
B
A










)
(
1
)
( A
A 
 

]
1
,
0
[
: 
X
r
)
(
max
π(A) x
r
A
x

• Let
where (possibility distribution) length of r: n
• Let when i < j
is the set of all possibility distributions of length n
• Given two possibility distributions (length n)
 is a lattice: lattice of possibility distributions of length n
n = 3
4 / 14
}
,
,
,
{ 2
1 n
x
x
x
X 

)
,
,
,
( 2
1 n
r 

 

)
( i
i x
r


j
i 
 
R
n
R
R
N
n
n



))
,
min(
,
),
,
(min(
))
,
max(
,
),
,
(max(
)
,
,
,
(
)
,
,
,
(
1
1
1
1
2
1
2
1
2
2
2
1
2
2
1
2
1
1
1
1
n
j
n
i
j
i
j
i
n
j
n
i
j
i
j
i
n
i
i
n
n
n
n
r
r
r
r
N
i
IFF
r
r
R
r
R
r

























"







,
R
n



 3
3
3 r
r
r i
)
1
,
1
,
1
(
3 

r
)
,
,
( 3
2
1
3 

 i
i
i
i
r 
)
0
,
0
,
1
(
3 

r
Total ignorance
Perfect evidence
• ,  is defined on P(X) in terms of its m. So all
focal elements are nested subsets:
4 / 15
}
,
,
,
{ 2
1 n
x
x
x
X 

)
(
2
1 X
A
A
A n 


 
i
n
i
i
A
A
if
A
m
N
i
x
x
A




0
)
(
}
,
,
{ 1 
x1 x2 x3 x4 xn
m(A1)
m(A2) m(A3) m(A4)
m(An)
r(x1)=1
r(x2)=2
r(x3)=3 r(x4)=4
r(xn)=n
…
…
…
Complete sequence of nested subsets of X
• Example
4 / 16
x1 x2 x3 x4
m(A2) = 0.3
m(A3) = 0.4 m(A6) = 0.1
r(x1)=1=1
r(x2)=2=1
r(x3)=3=0.7 r(x4)=4=0.3 r(x6)=6=0.3
x5 x6 x7
m(A7) = 0.2
r(x5)=5=0.3
r(x7)=7=0.2
i m(Ai)
1 0
2 0.3
3 0.4
4 0
5 0
6 0.1
7 0.2
A possibility measure defined on X
• Basic distribution m
• The set of all basic distributions
 m and r represent each other unambiguously
• (one-to-one correspondance)
4 / 17
1
,
)
(
)
,
,
,
(
1
2
1





n
i
i
i
i
n
A
m
m




 
)
0
(
(*)
:
.
.
)
(
})
({
1
1
2
2
2
1
1
convention
by
i
E
I
N
i
A
m
x
Pl
n
i
i
i
n
n
n
n
n
n
i
k
k
n
i
k
k
i
i

"











"



























M
M
M
N
n
n
n



(length n)
M
R
t 
:
)
(
)
(
(*)
)
(
2
1
1
1
2
1
m
t
m
t
IFF
m
m
satisfied
is
IFF
m
r
t





Partial ordering defined on according to on
M
n
 R
n
• Let’s calculate r from m in the example on 4/16
• Possibility measure:
• Marginal possibility distributions rX and rY of joint possibility distribution r
• Noninteractive sets (possibilistic sense) if
4 / 18






i
j
j
i
i
i
i
1
1





7
.
0
)
,
,
max(
})
,
,
({ 5
4
3
5
4
3 
 


 x
x
x
))
1
,
,
1
,
1
(
(
)
1
,
,
0
,
0
,
0
(
)
(
))
0
,
,
0
,
1
(
(
)
0
,
,
0
,
0
,
1
(
)
(












n
n
n
n
r
r
t
r
r
t
))
,
(
(
max
)
(
))
,
(
(
max
)
(
y
x
r
y
r
y
x
r
x
r
X
x
Y
Y
y
X




Y
y
X
x
y
r
x
r
y
x
r Y
X 
"

"
 ))
(
),
(
min(
)
,
(
• Example for noninteractive consonant bodies of evidence 4 / 19
y1 y2
x1 x2 x3 x4 y3
1,1 1,2 2,1 2,2 3,1 3,2 4,1 4,2 1,3 2,3 3,3 4,3
X Y
X×Y
1=0.4
3=0.1 4=0.5
12=0.4
32=0.1 42=0.2
43=0.3
2´=0.7
3´=0.3
1=1
2=0.6 3=0.6
4=0.5
11=1
12=1
21=0.6 22=0.6 31=0.6 32=0.6
41=0.5 42=0.5
13=0.3
23=0.3
33=0.3
43=0.3
1´=1
2´=1
3´=0.3
Marginal consonant bodies of evidence
Joint consonant body of evidence
)
,
min( '
j
i
ij 

 
This definition of
noninteraction does not
conform with
Dempster’s rule (4/8)
• Example: two consonant bodies of evidence 4 / 20
1 2
.x1 .x2
.y1 .y2
.x1y1
.x2y1
.x1y2
.x2y2
.x1y2
.x1y1
.x2y1
.x2y2
1 0.8
X Y
X×Y
1 0.6
1´ 2´
0.2 0.8
1 2
0.4
0.6
1´
2´
0.2
0.2
0.6
1
0.8
0.6
0.6
0.12
0.08
0.32
0.48
Marginal
possibility
distributions
Marginal
basic
assignment
))
(
),
(
min(
)
,
( y
r
x
r
y
x
r Y
X
 K
C
m
B
m
y
x
m
Y
A
C
B
X




1
)
(
)
(
)
,
(
Joint basic
assignment
Joint possibility
distribution
Joint basic
assigmnet
m r m’
Largest joint poss. distr. that
satisfies the marginal distr’s
Not consonant!
Not a subject of
possibility theory
X×Y
• Classification of fuzzy measures
4 / 21
Fuzzy measures
Plausibility m’s
Belief m’s
Nec. m’s
crisp
Poss. m’s
crisp
Probability m’s (additivity)
Nested focal
elements
Focal element is
singleton
Mathematical theory of evidence

More Related Content

What's hot

Counting, mathematical induction and discrete probability
Counting, mathematical induction and discrete probabilityCounting, mathematical induction and discrete probability
Counting, mathematical induction and discrete probabilitySURBHI SAROHA
 
Continutiy of Functions.ppt
Continutiy of Functions.pptContinutiy of Functions.ppt
Continutiy of Functions.pptLadallaRajKumar
 
Lesson02 Vectors And Matrices Slides
Lesson02   Vectors And Matrices SlidesLesson02   Vectors And Matrices Slides
Lesson02 Vectors And Matrices SlidesMatthew Leingang
 
Eigen values and eigenvectors
Eigen values and eigenvectorsEigen values and eigenvectors
Eigen values and eigenvectorsAmit Singh
 
Fuzzy logic and application in AI
Fuzzy logic and application in AIFuzzy logic and application in AI
Fuzzy logic and application in AIIldar Nurgaliev
 
Gaussian Process Regression
Gaussian Process Regression  Gaussian Process Regression
Gaussian Process Regression SEMINARGROOT
 
Fuzzy Logic ppt
Fuzzy Logic pptFuzzy Logic ppt
Fuzzy Logic pptRitu Bafna
 
Predicates and quantifiers presentation topics
Predicates  and quantifiers  presentation topicsPredicates  and quantifiers  presentation topics
Predicates and quantifiers presentation topicsR.h. Himel
 
3. Linear Algebra for Machine Learning: Factorization and Linear Transformations
3. Linear Algebra for Machine Learning: Factorization and Linear Transformations3. Linear Algebra for Machine Learning: Factorization and Linear Transformations
3. Linear Algebra for Machine Learning: Factorization and Linear TransformationsCeni Babaoglu, PhD
 
Regularization and variable selection via elastic net
Regularization and variable selection via elastic netRegularization and variable selection via elastic net
Regularization and variable selection via elastic netKyusonLim
 
The Multivariate Gaussian Probability Distribution
The Multivariate Gaussian Probability DistributionThe Multivariate Gaussian Probability Distribution
The Multivariate Gaussian Probability DistributionPedro222284
 
Bias and variance trade off
Bias and variance trade offBias and variance trade off
Bias and variance trade offVARUN KUMAR
 

What's hot (20)

L7 fuzzy relations
L7 fuzzy relationsL7 fuzzy relations
L7 fuzzy relations
 
Counting, mathematical induction and discrete probability
Counting, mathematical induction and discrete probabilityCounting, mathematical induction and discrete probability
Counting, mathematical induction and discrete probability
 
Continutiy of Functions.ppt
Continutiy of Functions.pptContinutiy of Functions.ppt
Continutiy of Functions.ppt
 
03 Machine Learning Linear Algebra
03 Machine Learning Linear Algebra03 Machine Learning Linear Algebra
03 Machine Learning Linear Algebra
 
Fuzzy logic
Fuzzy logicFuzzy logic
Fuzzy logic
 
Lesson02 Vectors And Matrices Slides
Lesson02   Vectors And Matrices SlidesLesson02   Vectors And Matrices Slides
Lesson02 Vectors And Matrices Slides
 
Eigen values and eigenvectors
Eigen values and eigenvectorsEigen values and eigenvectors
Eigen values and eigenvectors
 
Introduction to optimization Problems
Introduction to optimization ProblemsIntroduction to optimization Problems
Introduction to optimization Problems
 
topology definitions
 topology definitions topology definitions
topology definitions
 
Fuzzy logic and application in AI
Fuzzy logic and application in AIFuzzy logic and application in AI
Fuzzy logic and application in AI
 
Gaussian Process Regression
Gaussian Process Regression  Gaussian Process Regression
Gaussian Process Regression
 
Fuzzy Logic ppt
Fuzzy Logic pptFuzzy Logic ppt
Fuzzy Logic ppt
 
Predicates and quantifiers presentation topics
Predicates  and quantifiers  presentation topicsPredicates  and quantifiers  presentation topics
Predicates and quantifiers presentation topics
 
3. Linear Algebra for Machine Learning: Factorization and Linear Transformations
3. Linear Algebra for Machine Learning: Factorization and Linear Transformations3. Linear Algebra for Machine Learning: Factorization and Linear Transformations
3. Linear Algebra for Machine Learning: Factorization and Linear Transformations
 
Regularization and variable selection via elastic net
Regularization and variable selection via elastic netRegularization and variable selection via elastic net
Regularization and variable selection via elastic net
 
Ridge regression
Ridge regressionRidge regression
Ridge regression
 
Definition of banach spaces
Definition of banach spacesDefinition of banach spaces
Definition of banach spaces
 
FUZZY COMPLEMENT
FUZZY COMPLEMENTFUZZY COMPLEMENT
FUZZY COMPLEMENT
 
The Multivariate Gaussian Probability Distribution
The Multivariate Gaussian Probability DistributionThe Multivariate Gaussian Probability Distribution
The Multivariate Gaussian Probability Distribution
 
Bias and variance trade off
Bias and variance trade offBias and variance trade off
Bias and variance trade off
 

Similar to fuzzy_measures.ppt

00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt
00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt
00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).pptDediTriLaksono1
 
Discrete probability
Discrete probabilityDiscrete probability
Discrete probabilityRanjan Kumar
 
Ch1 sets and_logic(1)
Ch1 sets and_logic(1)Ch1 sets and_logic(1)
Ch1 sets and_logic(1)Kwonpyo Ko
 
Data mining assignment 2
Data mining assignment 2Data mining assignment 2
Data mining assignment 2BarryK88
 
Briefnts1 events
Briefnts1 eventsBriefnts1 events
Briefnts1 eventsilathahere
 
Deep learning .pdf
Deep learning .pdfDeep learning .pdf
Deep learning .pdfAlHayyan
 
Econometrics 2.pptx
Econometrics 2.pptxEconometrics 2.pptx
Econometrics 2.pptxfuad80
 
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...Alberto Maspero
 
Spanos lecture+3-6334-estimation
Spanos lecture+3-6334-estimationSpanos lecture+3-6334-estimation
Spanos lecture+3-6334-estimationjemille6
 
Probability and Statistics Exam Help
Probability and Statistics Exam HelpProbability and Statistics Exam Help
Probability and Statistics Exam HelpStatistics Exam Help
 
Finance Enginering from Columbia.pdf
Finance Enginering from Columbia.pdfFinance Enginering from Columbia.pdf
Finance Enginering from Columbia.pdfCarlosLazo45
 

Similar to fuzzy_measures.ppt (20)

Estimation rs
Estimation rsEstimation rs
Estimation rs
 
00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt
00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt
00_1 - Slide Pelengkap (dari Buku Neuro Fuzzy and Soft Computing).ppt
 
Discrete probability
Discrete probabilityDiscrete probability
Discrete probability
 
Ch1 sets and_logic(1)
Ch1 sets and_logic(1)Ch1 sets and_logic(1)
Ch1 sets and_logic(1)
 
Course material mca
Course material   mcaCourse material   mca
Course material mca
 
Update 2
Update 2Update 2
Update 2
 
BOOLEAN ALGEBRA.ppt
BOOLEAN ALGEBRA.pptBOOLEAN ALGEBRA.ppt
BOOLEAN ALGEBRA.ppt
 
Data mining assignment 2
Data mining assignment 2Data mining assignment 2
Data mining assignment 2
 
Briefnts1 events
Briefnts1 eventsBriefnts1 events
Briefnts1 events
 
Deep learning .pdf
Deep learning .pdfDeep learning .pdf
Deep learning .pdf
 
Econometrics 2.pptx
Econometrics 2.pptxEconometrics 2.pptx
Econometrics 2.pptx
 
Unit II PPT.pptx
Unit II PPT.pptxUnit II PPT.pptx
Unit II PPT.pptx
 
9108528.ppt
9108528.ppt9108528.ppt
9108528.ppt
 
math camp
math campmath camp
math camp
 
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...
Birkhoff coordinates for the Toda Lattice in the limit of infinitely many par...
 
Spanos lecture+3-6334-estimation
Spanos lecture+3-6334-estimationSpanos lecture+3-6334-estimation
Spanos lecture+3-6334-estimation
 
Probability and Statistics Exam Help
Probability and Statistics Exam HelpProbability and Statistics Exam Help
Probability and Statistics Exam Help
 
Random Variable
Random Variable Random Variable
Random Variable
 
Finance Enginering from Columbia.pdf
Finance Enginering from Columbia.pdfFinance Enginering from Columbia.pdf
Finance Enginering from Columbia.pdf
 
Probability and Statistics Assignment Help
Probability and Statistics Assignment HelpProbability and Statistics Assignment Help
Probability and Statistics Assignment Help
 

Recently uploaded

Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physicsvishikhakeshava1
 
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |aasikanpl
 
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...Sérgio Sacani
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfnehabiju2046
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsAArockiyaNisha
 
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...Sérgio Sacani
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxUmerFayaz5
 
GFP in rDNA Technology (Biotechnology).pptx
GFP in rDNA Technology (Biotechnology).pptxGFP in rDNA Technology (Biotechnology).pptx
GFP in rDNA Technology (Biotechnology).pptxAleenaTreesaSaji
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Nistarini College, Purulia (W.B) India
 
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxPhysiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxAArockiyaNisha
 
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...jana861314
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Patrick Diehl
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxkessiyaTpeter
 
Isotopic evidence of long-lived volcanism on Io
Isotopic evidence of long-lived volcanism on IoIsotopic evidence of long-lived volcanism on Io
Isotopic evidence of long-lived volcanism on IoSérgio Sacani
 
Nanoparticles synthesis and characterization​ ​
Nanoparticles synthesis and characterization​  ​Nanoparticles synthesis and characterization​  ​
Nanoparticles synthesis and characterization​ ​kaibalyasahoo82800
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaPraksha3
 
Unlocking the Potential: Deep dive into ocean of Ceramic Magnets.pptx
Unlocking  the Potential: Deep dive into ocean of Ceramic Magnets.pptxUnlocking  the Potential: Deep dive into ocean of Ceramic Magnets.pptx
Unlocking the Potential: Deep dive into ocean of Ceramic Magnets.pptxanandsmhk
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real timeSatoshi NAKAHIRA
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...RohitNehra6
 

Recently uploaded (20)

Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physics
 
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Mukherjee Nagar(Delhi) |
 
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdf
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based Nanomaterials
 
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptx
 
GFP in rDNA Technology (Biotechnology).pptx
GFP in rDNA Technology (Biotechnology).pptxGFP in rDNA Technology (Biotechnology).pptx
GFP in rDNA Technology (Biotechnology).pptx
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...
 
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxPhysiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
 
The Philosophy of Science
The Philosophy of ScienceThe Philosophy of Science
The Philosophy of Science
 
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
Traditional Agroforestry System in India- Shifting Cultivation, Taungya, Home...
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
 
Isotopic evidence of long-lived volcanism on Io
Isotopic evidence of long-lived volcanism on IoIsotopic evidence of long-lived volcanism on Io
Isotopic evidence of long-lived volcanism on Io
 
Nanoparticles synthesis and characterization​ ​
Nanoparticles synthesis and characterization​  ​Nanoparticles synthesis and characterization​  ​
Nanoparticles synthesis and characterization​ ​
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
 
Unlocking the Potential: Deep dive into ocean of Ceramic Magnets.pptx
Unlocking  the Potential: Deep dive into ocean of Ceramic Magnets.pptxUnlocking  the Potential: Deep dive into ocean of Ceramic Magnets.pptx
Unlocking the Potential: Deep dive into ocean of Ceramic Magnets.pptx
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real time
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...
 

fuzzy_measures.ppt

  • 1. • Fuzzy set  Fuzzy measure To which degree does an What is the degree of evidence, individual belong to a set that an individual belongs to a defined in an inexact way? certain (crisp) set? • Fuzzy measure g: P(X)  [0,1] If g(A)g(B) BP(X) then A is the "best guess"  E.g.: diagnose of a patient with given symptoms • Usual interpretation: subjective probability, however: properties are less strict! • Axioms of fuzzy measures: g1. g()=0 and g(X)=1 (boundary conditions) g2. " A,BP(X): if A  B then g(A)g(B) (monotonicity) g3. " sequence (AiP(X)  i, Ai X): if A1A2... or A1A2... then (continuity) i i i i g A g A    lim lim ( ) ( ) 4 / 1
  • 2. • More general definition of fuzzy measure g: g: B  [0,1] where BP(X) is a family of subsets: 1. B and XB 2. If AB then B 3. If A,BB then ABB (B is Borel-field or s-field)  As ABA and ABB, we have max(g(A),g(B))g(AB) also ABA and ABB, so min(g(A),g(B))g(AB) • Fuzzy measures are "too general". Practical considerations lead to some restrictions, so we obtain special classes of measures.  Clearly, probability is one of them! 4 / 2
  • 3. • Belief measure : Bel: P(X)  [0,1] Axioms: g1-g3 and g4Bel.: for arbitrary AiP(X)  n=2 Bel(A1A2) Bel(A1)+Bel(A2)-Bel(A1A2) n=3 Bel(A1A2A3)Bel(A1)+Bel(A2)+Bel(A3)-Bel(A1A2)-Bel(A1A3)- -Bel(A2A3)+Bel(A1A2A3) • Interpretation: Bel(A) is the degree of Belief that a particular xX belongs to A • If (AiAj)= then Probability is a special Bel measure!  g4Belg2 (not independent axiom) so {g1, g3, g4Bel } defines Bel measures Bel A A Bel A n i i n ( ... ) ( ) 1 1      Bel A A Bel A Bel A A Bel A A A n i i i j i j n n ( ... ) ( ) ( ) ... ( ) ( ... ) 1 1 1 2 1                4 / 3
  • 4. • Plausability measure : Pl: P(X)  [0,1] Axioms: g1-g3 and g4Pl.: for arbitrary AiP(X), n  Pl is dual with Bel! , (*)  Connection of g4Bel and g4Pl: (*) applied to g4Bel we obtain: I.e. g4Pl for ) ... ( ) 1 ( ... ) ( ) ( ) ... ( 2 1 1 1 1 n n j i j i n i i n A A A Pl A A Pl A Pl A A Pl                 Pl A Bel A ( ) ( )   1 Bel A Pl A ( ) ( )   1 1 1 1 1 1 1 1 1 1                    Pl A Pl A Pl A A Pl A i n i i i n i j i j n i n i ( ) ( ( ) ( ( )) ... ( ) ( ( )) Pl A Pl A Pl A A Pl A i n i i i n i j i j n i n i ( ) ( ) ( ) ... ( ) ( )                1 1 1 1 1 Ai 4 / 4
  • 5.  From g4Bel-Pl we obtain: • Every Bel and its dual Pl can be expressed by the "basic (probabilistic) assignment" m: m: P(X)  [0,1], m()=0 and m is the degree of evidence that xX belongs to A but not to any particular subset of A.  Different from probabilistic p: x  [0,1] !  m(X)=1 not necessarily! m(A) m(B) when AB: not necessarily! m(A) and m( ) have no relation m is not a fuzzy measure! • (**) Bel A Bel A ( ) ( )  1 Pl A Pl A ( ) ( )   1 m A ( )    1 A P(X) A    A B B m A Bel ) ( ) ( Pl A m B B A ( ) ( )     4 / 5
  • 6. Interpretation: m(A): degrees of evidence that xX belongs to A alone Bel(A): degrees of evidence that xX belongs to some subsets of A Pl(A): degrees of evidence that xX belongs to A or any subsets of X which overlaps with A  Pl(A)Bel(A) " AP(X) • If m(A)>0 : A is a focal element (of m) • Body of evidence: <F, m> F: set of focal elements m: basic assignment  F={X}, m(X)=1 : total ignorance  Bel(X)=1, Bel(A)=0 " AX  Pl()=0, Pl(A)=1 " A • Simple support function (focused at A): m(A)=s, m(X)=1-s, and m(B)=0 " BA,X 4 / 6
  • 7.  Basic assignment derived from Bel:  Example X={a,b,c} P(X)={,{a},{b},{c},{a,b},{a,c},{b,c},{a,b,c}} m: {a}  0.1 {b}  0.2 {a,b}  0.3 F={{a},{b},{a,b},{b,c},{a,b,c}} {b,c}  0.3 {a,b,c}  0.1 Bel({a,b})=m()+m{a}+m{b}+m{a,b}=0+0.1+0.2+0.3=0.6 Pl({a,b})=m{a,b,c}+m{a,b},+m{a,c}+m{b,c}+m{a}+m{b}= =0.1+0.3+0+0.3+0.1+0.2=1.0 Bel({c})=m()+m{c}=0+0=0 Pl({c})=m{a,b,c}+m{a,c}+m{b,c}+m{c}=0.1+0+0.3+0=0.4 Bel()=0 Bel({a})=0.1 Bel({b})=0.2 Bel({a,c}=0.1 Bel({b,c})=0.5 Bel({a,b,c})=1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ m()=0 m({a})=Bel({a})=0.1 m({b})=Bel({b})=0.2 m({c})=0 m({a,b})=Bel({a,b})-Bel({a})-Bel({b})=0.6-0.1-0.2=0.3 Etc. m A Bel B A B B A ( ) ( ) ( )      1 4 / 7
  • 8. • Evidence from two independent sources: m1,m2 Joint basic assignment m1,2: A (***) m1,2()=0 (Dempster's rule of combination) The sum of products m1(B)m2(C) for all focal elements of B (m1) and C (m2) so that BC, equals (in the denominator); this is why the sum of products is normalized  Example (from Klir-Folger textbook) Assume that an old painting was discovered that strongly resembles paintings by Raphael. Such a discovery is likely to generate various questions regarding the status of the painting. Assume the following three questions: 1. Is the discovered painting a genuine painting by Raphael? 2. Is the discovered painting a product of one of Raphael's many disciples? 3. Is the discovered painting a counterfeit? Let R, D and C denote subsets of our universal set X - the set of paintings - that contain the set of all paintings by Raphael, the set of all paintings by disciples of Raphael, and the set of all counterfeits of Raphael's paintings, respectively. Assume now, that two experts performed careful examinations of the painting and subsequently provided us with basic assignments m1 and m2 specified in the table below. These are the degrees of evidence that each expert obtained by the examination and that support the various claims that the painting belongs to one of the sets of our concern. For example, m1(RD)=0.15 is the degree of evidence obtained by the first expert that the painting was done by Raphael himself or that the painting was done by one of his disciples. m A m B m C) m B m C) B C A B C 1 2 1 2 1 2 1 , ( ) ( ) ( ( ) (         1 1 2     m B m C) B C ( ) ( 4 / 8
  • 9. U singEq. (**) wecan easilycalculatethetotal evidence, Bel1 and Bel2, in each set, as shown in thetable. Applying Dempster's rule (Eq. (***)) to m1 and m2, we obtain the joint basic assignment m1,2, which is also shown in thetable. To determinethevalues of m1,2, wecalculatethenormalization factor 1-Kfirst:  K m B m C B C     1 2 ( ) ( ) K=m1(R)m2(D)+m1(R)m2(C)+m1(R)m2(DC)+m1(D)m2(R)+m1(D)m2(C)+m1(D)m2(RC)+m1(C)m2(R)+  +m1(C)m2(D)+m1(C)m2(RD)+m1(RD)m2(C)+m1(RC)m2(D)+m1(DC)m2(R)=0.03 Thenormalization factor is then 1-K=0.97. Values of m1,2 arecalculated byEq. (***). For example: m1,2(R)=[m1(R)m2(R)+m1(R)m2(RD)+m1(R)m2(RC)+m1(R)m2(RDC)+m1(RD)m2(R)+  +m1(RD)m2(RC)+m1(RC)m2(R)+m1(RC)m2(RD)+m1(RDC)m2(R)]/0.97=0.21 m1,2(D)=[m1(D)m2(D)+m1(D)m2(RD)+m1(D)m2(DC)+m1(D)m2(RDC)+m1(RD)m2(D)+  +m1(RD)m2(DC)+m1(DC)m2(D)+m1(DC)m2(RD)+m1(RDC)m2(D)]/0.97=0.01 m1,2(RC)=[m1(RC)m2(RC)+m1(RC)m2(RDC)+m1(RDC)m2(RC)]/0.97=0.2 m1,2(RDC)=[m1(RDC)m2(RDC)]/0.97=0.31 and similarly for the remaining focal elements C, RD and DC. The joint basic assignment can now be used to calculate the joint Belief Bel1,2 (see table) and joint plausibilityPl1,2.  Expert 1 Expert 2 Combined evidence Focal elements m1 Bel1 m2 Bel2 m1,2 Bel1,2 R 0.05 0.05 0.15 0.15 0.21 0.21 D 0 0 0 0 0.01 0.01 C 0.05 0.05 0.05 0.05 0.09 0.09 RD 0.15 0.2 0.05 0.2 0.12 0.34 RC 0.1 0.2 0.2 0.4 0.2 0.5 DC 0.05 0.1 0.05 0.1 0.06 0.16 RDC 0.6 1 0.5 1 0.31 1 Combination of degrees of evidence fromtwo independent sources 4 / 9
  • 10. • Marginal basic assignment: Given m: P(X×Y)  [0,1], Focal elements are binary relations R on X×Y If RX is RX and RY is RY, the projections of m on X and Y, resp. are the marginal basic assignments: " AP(X) " AP(X) (****) • Marginal bodies of evidence: <FX,mX>, <FY,mY> • Noninteractive marginal basic assignment: m(A×B)=mX(A)mY(B) " AFX, " BFY and m(R)=0 for RA×B m A m R X R A RX ( ) ( ) :    m B m R Y R B RY ( ) ( ) :    4 / 10
  • 11.  Example: body of evidence (only focal elements are given) This is also a case of noninteractive marginal basic assignments E.g.: m(R1)=0.0625 m(R1)=mX({2,3})*mY({b,c})=0.25*0.25 {2,3}x{b,c}={2b,2c,3b,3c}=R1 Or m(R7)=0.0375 m(R7)=mX({1,3})*mY({a})=0.15*0.25 XxY 1a 1b 1c 2a 2b 2c 3a 3b 3c m(Ri) R1 = 0 0 0 0 1 1 0 1 1 0.0625 R2 = 0 0 0 1 0 0 1 0 0 0.0625 R3 = 0 0 0 1 1 1 1 1 1 0.125 R4 = 0 1 1 0 0 0 0 1 1 0.0375 R5 = 0 1 1 0 1 1 0 0 0 0.075 R6 = 0 1 1 0 1 1 0 1 1 0.075 R7 = 1 0 0 0 0 0 1 0 0 0.0375 R8 = 1 0 0 1 0 0 0 0 0 0.075 R9 = 1 0 0 1 0 0 1 0 0 0.075 R10 = 1 1 1 0 0 0 1 1 1 0.075 R11 = 1 1 1 1 1 1 0 0 0 0.15 R12 = 1 1 1 1 1 1 1 1 1 0.15 m: P(XxY)  [0,1] By (****) we obtain: X 1 2 3 mX(A ) A = 0 1 1 0.25 1 0 1 0.15 1 1 0 0.3 1 1 1 0.3 mx: P(X)  [0,1] Y a b c mY(B) B = 0 1 1 0.25 1 0 0 0.25 1 1 1 0.5 mY : P(Y)  [0,1] 4 / 11
  • 12. • Probability measures If g4Bel is replaced by g4P: Bel(AB)=Bel(A)+Bel(B), if AB= (additivity) we obtain a special class of belief measures: probability measures (Bayesian belief measures)  Bel is P iff the basic assignment: m({x})=Bel({x}) and m(A)=0 if A is not singleton  Probability measures are fully represented by p: x  [0,1] probability distribution function (as focal elements are only singletons)  " AP(X) i.e. • Total ignorance: • Joint and marginal probability distributions - similar way  Very Broad field Bel A Pl A m x x A ( ) ( ) ({ })     Bel A Pl A p x x A ( ) ( ) ( )       p x X m x ( ) ({ })   1 4 / 12
  • 13. • Nested subsets: • Consonant Bel / Pl The family of focal elements is nested (consonant body of evidence <F, m> )  Given a consonant <F, m> then " A,B P(X) • Necessity measure = consonant Bel Possibility measure = consonant Pl (very typical – ‘axiomatic’ – prop’s)  Every possibility measure can be uniquely determined by a possibility distribution f’n So that 4 / 13 X A A A A i n      2 1 )) ( ), ( max( ) ( )) ( ), ( min( ) ( B Pl A Pl B A Pl B Bel A Bel B A Bel     )) ( ), ( max( ) ( )) ( ), ( min( ) ( B A B A B A B A           ) ( 1 ) ( A A     ] 1 , 0 [ :  X r ) ( max π(A) x r A x 
  • 14. • Let where (possibility distribution) length of r: n • Let when i < j is the set of all possibility distributions of length n • Given two possibility distributions (length n)  is a lattice: lattice of possibility distributions of length n n = 3 4 / 14 } , , , { 2 1 n x x x X   ) , , , ( 2 1 n r      ) ( i i x r   j i    R n R R N n n    )) , min( , ), , (min( )) , max( , ), , (max( ) , , , ( ) , , , ( 1 1 1 1 2 1 2 1 2 2 2 1 2 2 1 2 1 1 1 1 n j n i j i j i n j n i j i j i n i i n n n n r r r r N i IFF r r R r R r                          "        , R n     3 3 3 r r r i ) 1 , 1 , 1 ( 3   r ) , , ( 3 2 1 3    i i i i r  ) 0 , 0 , 1 ( 3   r Total ignorance Perfect evidence
  • 15. • ,  is defined on P(X) in terms of its m. So all focal elements are nested subsets: 4 / 15 } , , , { 2 1 n x x x X   ) ( 2 1 X A A A n      i n i i A A if A m N i x x A     0 ) ( } , , { 1  x1 x2 x3 x4 xn m(A1) m(A2) m(A3) m(A4) m(An) r(x1)=1 r(x2)=2 r(x3)=3 r(x4)=4 r(xn)=n … … … Complete sequence of nested subsets of X
  • 16. • Example 4 / 16 x1 x2 x3 x4 m(A2) = 0.3 m(A3) = 0.4 m(A6) = 0.1 r(x1)=1=1 r(x2)=2=1 r(x3)=3=0.7 r(x4)=4=0.3 r(x6)=6=0.3 x5 x6 x7 m(A7) = 0.2 r(x5)=5=0.3 r(x7)=7=0.2 i m(Ai) 1 0 2 0.3 3 0.4 4 0 5 0 6 0.1 7 0.2 A possibility measure defined on X
  • 17. • Basic distribution m • The set of all basic distributions  m and r represent each other unambiguously • (one-to-one correspondance) 4 / 17 1 , ) ( ) , , , ( 1 2 1      n i i i i n A m m       ) 0 ( (*) : . . ) ( }) ({ 1 1 2 2 2 1 1 convention by i E I N i A m x Pl n i i i n n n n n n i k k n i k k i i  "            "                            M M M N n n n    (length n) M R t  : ) ( ) ( (*) ) ( 2 1 1 1 2 1 m t m t IFF m m satisfied is IFF m r t      Partial ordering defined on according to on M n  R n
  • 18. • Let’s calculate r from m in the example on 4/16 • Possibility measure: • Marginal possibility distributions rX and rY of joint possibility distribution r • Noninteractive sets (possibilistic sense) if 4 / 18       i j j i i i i 1 1      7 . 0 ) , , max( }) , , ({ 5 4 3 5 4 3       x x x )) 1 , , 1 , 1 ( ( ) 1 , , 0 , 0 , 0 ( ) ( )) 0 , , 0 , 1 ( ( ) 0 , , 0 , 0 , 1 ( ) (             n n n n r r t r r t )) , ( ( max ) ( )) , ( ( max ) ( y x r y r y x r x r X x Y Y y X     Y y X x y r x r y x r Y X  "  "  )) ( ), ( min( ) , (
  • 19. • Example for noninteractive consonant bodies of evidence 4 / 19 y1 y2 x1 x2 x3 x4 y3 1,1 1,2 2,1 2,2 3,1 3,2 4,1 4,2 1,3 2,3 3,3 4,3 X Y X×Y 1=0.4 3=0.1 4=0.5 12=0.4 32=0.1 42=0.2 43=0.3 2´=0.7 3´=0.3 1=1 2=0.6 3=0.6 4=0.5 11=1 12=1 21=0.6 22=0.6 31=0.6 32=0.6 41=0.5 42=0.5 13=0.3 23=0.3 33=0.3 43=0.3 1´=1 2´=1 3´=0.3 Marginal consonant bodies of evidence Joint consonant body of evidence ) , min( ' j i ij     This definition of noninteraction does not conform with Dempster’s rule (4/8)
  • 20. • Example: two consonant bodies of evidence 4 / 20 1 2 .x1 .x2 .y1 .y2 .x1y1 .x2y1 .x1y2 .x2y2 .x1y2 .x1y1 .x2y1 .x2y2 1 0.8 X Y X×Y 1 0.6 1´ 2´ 0.2 0.8 1 2 0.4 0.6 1´ 2´ 0.2 0.2 0.6 1 0.8 0.6 0.6 0.12 0.08 0.32 0.48 Marginal possibility distributions Marginal basic assignment )) ( ), ( min( ) , ( y r x r y x r Y X  K C m B m y x m Y A C B X     1 ) ( ) ( ) , ( Joint basic assignment Joint possibility distribution Joint basic assigmnet m r m’ Largest joint poss. distr. that satisfies the marginal distr’s Not consonant! Not a subject of possibility theory X×Y
  • 21. • Classification of fuzzy measures 4 / 21 Fuzzy measures Plausibility m’s Belief m’s Nec. m’s crisp Poss. m’s crisp Probability m’s (additivity) Nested focal elements Focal element is singleton Mathematical theory of evidence