fuzzy systems - possibility theoryfuzzy.cs.ovgu.de/.../lehre.fs0910/fs0910lecture07.pdf ·...

61
Fuzzy Systems Possibility Theory Rudolf Kruse Christian Moewes {kruse,cmoewes}@iws.cs.uni-magdeburg.de Otto-von-Guericke University of Magdeburg Faculty of Computer Science Department of Knowledge Processing and Language Engineering R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 1 / 61

Upload: others

Post on 08-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Fuzzy SystemsPossibility Theory

Rudolf Kruse Christian Moeweskruse,[email protected]

Otto-von-Guericke University of MagdeburgFaculty of Computer Science

Department of Knowledge Processing and Language Engineering

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 1 / 61

Page 2: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Outline

1. Introduction

Problems with Probability TheoryUpper and Lower Probabilities

2. Dempster-Shafer Theory of Evidence

3. Possibility Theory

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 2 / 61

Page 3: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Problems with Probability TheoryRepresentation of Ignorance

• standard method to handle uncertainty is probability theory

• however, it has some problems regarding ignorance

• we are given one die with faces 1, . . . , 6

• what is the certainty of showing up face i?

1. conduct statistical survey (roll the die 10,000 times) andestimate relative frequency: P(i) = 1

62. use subjective probabilities (i.e., often normal case):

we don’t know anything (especially and explicitly we don’t haveany reason to assign unequal probabilities),so most plausible distribution is uniform

⇒ problem: uniform distribution because of ignorance or extensivestatistical tests

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 3 / 61

Page 4: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Problems with Probability TheoryRepresentation of Ignorance

• experts analyze aircraft shapes• 3 aircraft types A, B, C

• “It is type A or B with 90% certainty.”

• “About C , I don’t have any clue and I don’t want to commitmyself.”

• “No preferences for A or B.”

⇒ problem: propositions hard to handle with Bayesian theory

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 4 / 61

Page 5: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Upper and Lower Probabilities

• a way out to solve this problem are upper and lower probabilities• “It is type A or B with 90% certainty.”• “About C , I don’t have any clue and I don’t want to commit

myself.”• “No preferences for A or B.”

A ∅ A B C A, B A, C A, B, CP∗(A) 0 0 0 0 .9 0 1P∗(A) 0 1 1 .1 1 1 1

• consider P∗(A) and P∗(A) as lower and upper probability bounds

• total ignorance: P∗(A) = 0 and P∗(A) = 1 for A 6= ∅, A 6= Ω

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 5 / 61

Page 6: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Properties of Upper and Lower Probabilities I

1. P∗ : P(Ω) → [0, 1]

2. 0 ≤ P∗ ≤ P∗ ≤ 1, P∗(∅) = P∗(∅) = 0, P∗(Ω) = P∗(Ω) = 1

3. A ⊆ B ⇒ P∗(A) ≤ P∗(B) and P∗(A) ≤ P∗(B)

4. A ∩ B = ∅ 6⇒ P∗(A) + P∗(B) = P∗(A ∪ B)

5. P∗(A ∪ B) ≥ P∗(A) + P∗(B) − P∗(A ∩ B)

6. P∗(A ∪ B) ≤ P∗(A) + P∗(B) − P∗(A ∩ B)

7. P∗(A) = 1 − P∗(Ω\A)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 6 / 61

Page 7: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Properties of Upper and Lower Probabilities II

• one can prove the following generalized equation

P∗(n

i=1

Ai) ≥∑

∅6=I:I⊆1,...,n

(−1)|I|+1 · P∗(⋂

i∈I

Ai)

• these set functions play important role in theoretical physics (i.e.,capacities [Choquet, 1954])

• thoughts have been generalized and developed into theory ofbelief functions [Shafer, 1976]

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 7 / 61

Page 8: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Outline

1. Introduction

2. Dempster-Shafer Theory of Evidence

Belief MeasurePlausibility MeasureBasic Belief AssignmentFocal ElementTotal IgnoranceDempster’s Rule of CombinationExampleConversion Formulas

3. Possibility Theory

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 8 / 61

Page 9: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Dempster-Shafer Theory of Evidence

• like Bayesian approaches, the Dempster-Shafer theory ofevidence [Shafer, 1976] wants to model and quantify uncertaintyby degrees of belief

• Bayesian approaches assign degrees of belief to single hypothesis

• theory of evidence assigns degrees of belief to sets of hypotheses

• let X be finite “frame of discernment”

⇒ each proposition “x is x0” can be represented by A ⊂ X

• sets including one element correspond to elementary propositions

• proposition is true if it contains the true elementary proposition

• belief in “x is x0” related to A ⊂ X is measured by Bel(A) ∈ [0, 1]

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 9 / 61

Page 10: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Belief Measure

Definition

Given a finite universe X , a belief measure is a function

Bel : P(X ) → [0, 1]

s.t. Bel(∅) = 0, Bel(X ) = 1 and

Bel(A1 ∪ A2 ∪ . . . ∪ An) ≥n

i=1

Bel(Aj) −∑

i ,j:1≤i<j≤n

Bel(Ai ∩ Aj)

+ . . . + (−1)n+1 Bel(A1 ∩ A2 ∩ . . . ∩ An)

for all possible families of subsets of X .

• e.g., Bel(A1 ∪ A2) ≥ Bel(A1) + Bel(A2) − Bel(A1 ∩ A2)

• difficult to test ⇒ basic belief assignment is introduced

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 10 / 61

Page 11: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Belief Measure

• for each A ∈ P(X ), Bel(A) is interpreted as degree of belief

(based on evidence) that given x ∈ X belongs to A

• if sets A1, . . . , An are pairwise disjoint, then

Bel(A1, . . . , An) ≥n

i=1

Bel(Ai)

⇒ weaker version of additivity property of probability measures

⇒ probability measures are special case of belief measures whereequality in definition of Bel is always satisfied

• fundamental property: let n = 2, A1 = A and A2 = AC , then

Bel(A) + Bel(AC ) ≤ 1

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 11 / 61

Page 12: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Plausibility Measure

• associated with each Bel is plausibility measure Pl defined by

Pl(A) = 1 − Bel(AC ), ∀A ∈ P(X )

Definition

Given a finite universe X , a plausibility measure is a function

Pl : P → [0, 1]

s.t. Pl(∅) = 0, Pl(X ) = 1 and

Pl(A1 ∩ A2 ∩ . . . ∩ An) ≤∑

j

Pl(Aj) −∑

j<k

Pl(Aj ∪ Ak)

+ . . . + (−1)n+1 Pl(A1 ∪ A2 ∪ . . . ∪ An)

for all possible families of subsets of X .

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 12 / 61

Page 13: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Basic Belief Assignment

• fundamental property: let n = 2, A1 = A and A2 = AC , then

Pl(A) + Pl(AC ) ≥ 1

• Bel and Pl can be characterized by basic belief assignment

m : P(X ) → [0, 1]

s.t. m(∅) = 0 and∑

A∈P(X)

m(A) = 1

• for each A ∈ P(X ), value m(A) expresses proportion to which allevidence supports “particular element of X belongs to A”

• m(A) does not imply any claim regarding subsets of A• definition of m resembles probability distribution function, but

• probability distribution functions are defined on X• basic belief assignments are defined on P(X)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 13 / 61

Page 14: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Basic Belief Assignment

• observe the following facts

1. m(X) need not to be 12. if A ⊆ B, then m(A) need not to be less or equal m(B)3. there does not need to be any relation between m(A) and m(AC )

• given m, Bel and Pl are uniquely determined for all A ∈ P(X ) by

Bel(A) =∑

B|B⊆A

m(B) and Pl(A) =∑

B|A∩B 6=∅

m(B)

• note that inverse is also possible

• e.g., m(A) =∑

B|B⊆A(−1)|A\B| Bel(B)

• each of three functions is sufficient to determine other two

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 14 / 61

Page 15: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Basic Belief Assignment

• Bel(A) =∑

B|B⊆A m(B) means the following:

• m(A) characterizes degree of evidence that x belongs only to A• Bel(A) represents evidence that x belongs to A and subsets of A

• Pl(A) =∑

B|A∩B 6=∅ m(B) represents both kinds of evidence:• total evidence that x belongs to A or to any subset of A

• evidence associated with sets that overlap with A

⇒ for all A ∈ P(X ),Pl(A) ≥ Bel(A)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 15 / 61

Page 16: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Focal Element

• every A ∈ P(X ) for which

m(A) > 0

is called focal element of m

• they are subsets of X on which available evidence focuses

• X is finite ⇒ m can be fully described by list F of its focalelements

• pair 〈F , m〉 is called body of evidence

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 16 / 61

Page 17: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Total Ignorance

• total ignorance is expressed by

m(X ) = 1 and m(A) = 0 ∀A 6= X

• i.e., we know that element is in universal set but we have noevidence about is location in any A ⊂ X

• in terms of belief it is exactly the same:

Bel(X ) = 1 and Bel(A) = 0 ∀A 6= X

• in terms of plausibility it is different:

Pl(∅) = 0 and Pl(A) = 1 ∀A 6= ∅

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 17 / 61

Page 18: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Dempster’s Rule of Combination

• now, consider two independent sources of evidence (e.g., fromtwo experts) expressed by m1 and m2 on some P(X )

• they must be appropriately combined to obtain joint basicassignment m1,2

• generally, evidence can be combined in many ways

• standard way is expressed by Dempster’s rule of combination

m1,2(A) =

B∩C=A m1(B) · m2(C)

1 − K

for all A 6= ∅ and m1,2(∅) = 0 where

K =∑

B∩C=∅

m1(B) · m2(C)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 18 / 61

Page 19: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Dempster’s Rule of Combination

• m1(B) (focusing on B ∈ P(X )) and m2(C) (focusing onC ∈ P(X )) are combined by m1(B) · m2(C) (focusing on B ∩ C)

⇒ exactly the same for joint probability distribution which iscalculated from two independent marginal distributions

• since some intersections of focal elements may result in A, wemust add corresponding product to obtain m1,2

• also, some intersections may be empty

• since, m1,2(∅) = 0, K is thus not included in definition of m1,2

• to ensure∑

A∈P(X) m(A) = 1, each product is divided by 1 − K

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 19 / 61

Page 20: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• assume that an old code fragment was discovered by Rudolf Kruse

• it strongly resembles code fragments of Christian Moewes

• several questions can come up:

1. Is the discovered code fragment a real fragment of Christian?2. Is the discovered code fragment a product of one of Christian’s

students?3. Is the discovered code fragment a fake?

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 20 / 61

Page 21: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• let C , S, F denote subsets of universe X of all code fragments

• C denotes set of all code fragments by Christian

• S represents set of all code fragments by Christian’s students

• F represents set of all fakes of Christian’s code fragments

• two experts (named Georg and Matthias) performed carefulexaminations of the code fragment

• afterwards, they provided Rudolf Kruse with basic assignments m1

and m2 as follows

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 21 / 61

Page 22: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Two Independent Sources

experts Georg Matthias

focalelements m1 Bel1 m2 Bel2

C .05 .05 .15 .15S 0 0 0 0F .05 .05 .05 .05

C ∪ S .15 .2 .05 .2C ∪ F .1 .2 .2 .4S ∪ F .05 .1 .05 .1

C ∪ S ∪ F .6 1 .5 1

• degrees of evidence obtained by examination and supported bydifferent claims that code fragment belongs to one of the sets

• given m1 and m2, we can easily compute Bel1 and Bel2, resp.

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 22 / 61

Page 23: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Calculation of Normalization Factor

K = m1(C) · m2(S) + m1(C) · m2(F ) + m1(C) · m2(S ∪ F )

+ m1(S) · m2(C) + m1(S) · m2(F ) + m1(S) · m2(C ∪ F )

+ m1(F ) · m2(C) + m1(F ) · m2(S) + m1(F ) · m2(C ∪ S)

+ m1(C ∪ S) · m2(F ) + m1(C ∪ F ) · m2(S) + m1(S ∪ F ) · m2(C)

= .03

• so, the normalization factor is then 1 − K = .97

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 23 / 61

Page 24: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Calculation of Joint Basic Assignment

m1,2(C) = [m1(C) · m2(C) + m1(C) · m2(C ∪ S)

+ m1(C) · m2(C ∪ F ) + m1(C) · m2(C ∪ S ∪ F )

+ m1(C ∪ S) · m2(C) + m1(C ∪ S) · m2(C ∪ F )

+ m1(C ∪ F ) · m2(C) + m1(C ∪ F ) · m2(C ∪ S)

+ m1(C ∪ S ∪ F ) · m2(C)]/.97

= .21

m1,2(S) = [m1(S) · m2(S) + m1(S) · m2(C ∪ S) + m1(S) · m2(S ∪ F )

+ m1(S) · m2(C ∪ S ∪ F ) + m1(C ∪ S) · m2(S)

+ m1(C ∪ S) · m2(S ∪ F ) + m1(S ∪ F ) · m2(S)

+ m1(S ∪ F ) · m2(C ∪ S) + m1(C ∪ S ∪ F ) · m2(S)]/.97

= .01

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 24 / 61

Page 25: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Calculation of Joint Basic Assignment

m1,2(C ∪ F ) = [m1(C ∪ F ) · m2(C ∪ F )

+ m1(C ∪ F ) · m2(C ∪ S ∪ F )

+ m1(C ∪ S ∪ F ) · m2(C ∪ F )]/.97

= .2

m1,2(C ∪ S ∪ F ) = [m1(C ∪ S ∪ F ) · m2(C ∪ S ∪ F )]/.97

= .31

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 25 / 61

Page 26: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Joint Basic Assignment

combinedGeorg Matthias evidence

focalelements m1 Bel1 m2 Bel2 m1,2 Bel1,2

C .05 .05 .15 .15 .21 .21S 0 0 0 0 .01 .01F .05 .05 .05 .05 .09 .09

C ∪ S .15 .2 .05 .2 .12 .34C ∪ F .1 .2 .2 .4 .2 .5S ∪ F .05 .1 .05 .1 .06 .16

C ∪ S ∪ F .6 1 .5 1 .31 1

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 26 / 61

Page 27: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Projection

• consider now basic assignment m : P(X × Y ) → [0, 1]

• each focal element of m is binary relation R on X × Y

• let RX denote projection of R on X , then

RX = x ∈ X | (x , y) ∈ R for some y ∈ Y

• projection mX of m on X is

mX (A) =∑

R|A=RX

m(R) for all A ∈ P(X )

• adding values of m(R) for all focal elements R whose RX is A

• similarly, computation can be done for RY and mY , resp.

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 27 / 61

Page 28: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Marginal Bodies of Evidence

• mX and mY are also called marginal basic assignments

• 〈FX , mX 〉 and 〈FY , mY 〉 are called marginal bodies of evidence

• 〈FX , mX 〉 and 〈FY , mY 〉 are called noninteractive if and only iffor all A ∈ FX and B ∈ FY

m(A × B) = mX (A) · mY (B)

andm(R) = 0 for all R 6= A × B

• i.e., joint basic assignment m is uniquely determined by marginalbasic assignments

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 28 / 61

Page 29: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• consider body of evidence given in table• focal elements are subsets of X × Y = 1, 2, 3 × a, b, c

focal X × Y

element 1a 1b 1c 2a 2b 2c 3a 3b 3c m(Ri)R1 = 0 0 0 0 1 1 0 1 1 .0625R2 = 0 0 0 1 0 0 1 0 0 .0625R3 = 0 0 0 1 1 1 1 1 1 .125R4 = 0 1 1 0 0 0 0 1 1 .0375R5 = 0 1 1 0 1 1 0 0 0 .075R6 = 0 1 1 0 1 1 0 1 1 .075R7 = 1 0 0 0 0 0 1 0 0 .375R8 = 1 0 0 1 0 0 0 0 0 .075R9 = 1 0 0 1 0 0 1 0 0 .075

R10 = 1 1 1 0 0 0 1 1 1 .075R11 = 1 1 1 1 1 1 0 0 0 .15R12 = 1 1 1 1 1 1 1 1 1 .15

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 29 / 61

Page 30: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• margin bodies of evidence are obtained as follows

mX (2, 3) = m(R1) + m(R2) + m(R3) = .25

mX (1, 2) = m(R5) + m(R8) + m(R11) = .3

mY (a) = m(R2) + m(R7) + m(R8) + m(R9) = .25

mY (a, b, c) = m(R4) + m(R10) + m(R11) + m(R12) = .5

X

1 2 3 mX (A)

A = 0 1 1 .251 0 1 .151 1 0 .31 1 1 .3

Y

a b c mY (B)

B = 0 1 1 .251 0 0 .251 1 1 .5

• here, marginal bodies of evidence are noninteractive

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 30 / 61

Page 31: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Conversion Formulas

given m(A) = Bel(A) = Pl(A) =

m m(A)∑

B⊆A

m(B)∑

B∩A6=∅m(B)

Bel∑

B⊆A

(−1)|A\B| Bel(B) Bel(A) 1 − Bel(AC )

Pl∑

B⊆A

(−1)|A\B|(1 − Pl(BC )) 1 − Pl(AC ) Pl(A)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 31 / 61

Page 32: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Outline

1. Introduction

2. Dempster-Shafer Theory of Evidence

3. Possibility Theory

Necessity and PossibilityPossibility DistributionBasic DistributionMathematical PropertiesFuzzy Sets and Possibility TheoryPossibility versus ProbabilityComparison of both TheoriesInterpretations

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 32 / 61

Page 33: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibility Theory

• possibility theory is special kind of evidence theory

• bodies of evidence contain nested focal elements

• belief and plausibility are called necessity and possibility, resp.

Theorem

Let a given finite body of evidence 〈F , m〉 be nested. Then, the

associated belief and plausibility measures have the following

properties for all A, B ∈ P(X )

Bel(A ∩ B) = minBel(A), Bel(B)

Pl(A ∪ B) = maxPl(A), Pl(B)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 33 / 61

Page 34: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Necessity and Possibility

• let necessity and plausibility be denoted by Nec and Pos, resp.

• last theorem provides us with basic equation of possibility theory

Nec(A ∩ B) = minNec(A), Nec(B)

Pos(A ∪ B) = maxPos(A), Pos(B)

• can be also defined for arbitrary universal set (e.g., infinite ones)

Nec

k∈K

Ak

= infk∈K

Nec(Ak)

Pos

k∈K

Ak

= supk∈K

Pos(Ak)

where K is arbitrary index set

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 34 / 61

Page 35: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Necessity and Possibility

• since necessity and possibility are special belief and plausibilitymeasures, resp., they satisfy fundamental properties

Nec(A) + Nec(AC ) ≤ 1

Pos(A) + Pos(AC ) ≥ 1

1 − Pos(AC ) = Nec(A)

• from the definitions of Nec and Pos, it follows that

min

Nec(A), Nec(AC )

= 0

max

Pos(A), Pos(AC )

= 1

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 35 / 61

Page 36: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Necessity and Possibility

• necessity and possibility measures constrain each other

Theorem

For every A ∈ P(X ), any necessity measure Nec on P(X ) and the

associated possibility measure Pos satisfy the following implications:

Nec(A) > 0 ⇒ Pos(A) = 1,

Pos(A) < 1 ⇒ Nec(A) = 0.

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 36 / 61

Page 37: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibility Distribution Function

• given possibility measure Pos on P(X )• let function r : X → [0, 1] s.t. ∀x ∈ X : r(x) = Pos(x) be called

possibility distribution function associated with Pos

Theorem

Every possibility measure Pos on a finite power set P(X ) is uniquely

determined by a possibility distribution function

r : X → [0, 1]

via the formula

Pos(A) = maxx∈A

r(x)

for each A ∈ P(X ).

• when X is infinite, then Pos(A) = supx∈A r(x)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 37 / 61

Page 38: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibility Distribution

• let possibility distribution function r be defined onX = x1, . . . , xn

⇒ possibility distribution associated with r is n-tuple

r = (r1, . . . , rn)

where ri = r(xi ) for all xi ∈ X

• it is useful to order r s.t. ri ≥ rj when i < j

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 38 / 61

Page 39: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• now, let Pos be defined on P(X ) by m

• w.l.o.g., focal elements are some or all of subsets in completesequence of nested subsets

A1 ⊂ A2 ⊂ . . . ⊂ An = X

where Ai = x1, . . . , xi i ∈ INn

• i.e., m(A) = 0 for each A 6= Ai (i ∈ IN) and∑n

i=1 m(Ai) = 1

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 39 / 61

Page 40: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• complete sequence of nested focal elements of Pos on P(X )where X = x1, . . . , xn

x1 x2 x3 x4 . . . xn

m(A1)

m(A2)

m(A3)m(A4) . . . m(An)

r(x1)

r(x2)r(x3) r(x4)

. . . r(xn)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 40 / 61

Page 41: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Basic Distribution

• every Pos on finite set can be represented by n-tuple

m = (m1, . . . , mn)

for some finite n ∈ IN where mi = m(Ai) for all i ∈ INn

• clearly,∑n

i=1 mi = 1 and mi ∈ [0, 1] for all i ∈ INn

• m is called basic distribution

• each m represents exactly one possibility distribution r

∀xi ∈ X : ri = r(xi) = Pos(xi ) = Pl(xi )

∀i ∈ INn : ri = Pl(xi) =n

k=i

m(Ak) =n

k=i

mk

• solving all n equations for mi (i ∈ INn) leads to

mi = ri − ri+1

• one-two-one correspondence between r and m

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 41 / 61

Page 42: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• one possibility measure defined on X with A1 ⊂ A2 ⊂ . . . ⊂ A7

x1 x2 x3 x4 x5 x6 x7

m(A2) = .3

m(A3) = .4

m(A6) = .1

m(A7) = .2

r(x1) = 1

r(x2) = 1

r(x3) = .7

r(x4) = .3

r(x5) = .3

r(x6) = .3

r(x7) = .2

∑ni=1 m(Ai ) = 1 is satisfied, m(A1) = m(A4) = m(A5) = 0

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 42 / 61

Page 43: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Perfect Evidence and Total Ignorance

• smallest possibility distribution r has form

r = (1, 0, 0, . . . , 0)

• r represents perfect evidence with no uncertainty involved

• largest possibility distribution r can be written as

r = (0, 0, . . . , 0, 1)

• r denotes total ignorance, i.e., no relevant evidence is available

• the larger r , the less specific the evidence is (the more ignorant)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 43 / 61

Page 44: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Joint Possibility Distribution

• consider joint possibility distributions r defined on X × Y

• projections rX , rY are called marginal possibility distribution

∀x ∈ X : rX (x) = maxy∈Y

r(x , y)

∀y ∈ Y : rY (y) = maxx∈X

r(x , y)

• formula for rX (x) follows from

PosX (x) = Pos((x , y) | x ∈ Y )

PosX (x) = rX (x)

Pos((x , y) | x ∈ Y ) = maxy∈Y

r(x , y)

• same argumentation can be done for rY

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 44 / 61

Page 45: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibilistic Noninteraction

Definition

Nested bodies of evidence X and Y represented by rX and rY , resp.,are called noninteractive if and only if

r(x , y) = minrX (x), rY (y)

for all x ∈ X and all y ∈ Y

• recall product rule m(A × B) = mX (A) · mY (B)

• this definition is not based upon product rule of evidence theory

⇒ does not conform to general noninteractive bodies of evidence

• here, product rule would not preserve nested structure

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 45 / 61

Page 46: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibilistic Noninteraction

• let rX and rY be marginal poss. dist. functions on X and Y , resp.

• let r be joint poss. dist. function on X × Y defined by rX and rY

• assume PosX , PosY and Pos correspond to rX , rY and r , then

Pos(A × B) = minPosX (A), PosY (B)

for all A ∈ P(X ) and all B ∈ P(Y ) where

PosX (A) = maxx∈A

rX (x),

PosY (B) = maxy∈B

rY (y),

Pos(A × B) = maxx∈A,y∈B

r(x , y)

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 46 / 61

Page 47: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibilistic Independence

• two marginal possibilistic bodies of evidence are independent ifand only if conditional possibilities equal marginal possibilities,i.e., for all x ∈ X and all y ∈ Y

rX |Y (x |y) = rX (x),

rY |X (y |x) = rY (y)

• rX |Y (x |y) and rY |X (y |x) are conditional possibilities on X × Y

• possibilistic independence is stronger than poss. noninteraction

• possibilistic independence ⇒ possibilistic noninteraction

• possibilistic noninteraction 6⇒ possibilistic independence

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 47 / 61

Page 48: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Fuzzy Sets and Possibility Theory

• possibility theory can be also formulated using fuzzy sets insteadof nested bodies of evidence [Dubois and Prade, 1993]

• fuzzy sets (similar to bodies of evidence) are based on families ofnested sets, i.e., α-cuts

• Pos measures are connected with fuzzy sets via possibilitydistribution function

• e.g., let X denote variable taking values in set X

• X = x describes “X is x” where x ∈ X

• fuzzy set F on X expresses constraint on values assigned to X

• given value x ∈ X , F (x) is degree of compatibility of x

described by F

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 48 / 61

Page 49: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Fuzzy Sets and Possibility Theory

• given “X is F”, F (x) is degree of possibility that X = x

• i.e., given F on X and “X is F”,

∀x ∈ X : rF (x) = F (x)

• given rF , associated possibility measure is

∀A ∈ P(X ) : PosF (A) = supx∈A

rF (x)

• for normal fuzzy sets, NecF can be calculated by

∀A ∈ P(X ) : NecF (A) = 1 − PosF (AC )

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 49 / 61

Page 50: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Example

• let variable T be temperature in C (only integers)

• actual value is given in terms of “T is around 21C”

0

0.25

0.50

0.75

1.00

17 18 19 20 21 22 23 24 25

F = rF

b b

b

b

b

b

b

b b b

19 20 21 22 23

rF : 1/3 2/3 1 2/3 1/3

• incomplete information induces possibility distribution function rF

• rF is numerically identical with membership function F

• nested α-cuts play same role as focal elements

• Question: Which values do Nec and m take here?

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 50 / 61

Page 51: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Summary

• if rF is derived from normal fuzzy set F , then both formulationsof possibility theory are equivalent

• full equivalence breaks when F is not normal

⇒ basic probability assignment function m is not directly applicable

• all other properties remain equivalent in both formulations

• possibility theory is measure-theoretic counterpart of fuzzy settheory based upon standard fuzzy operations

• it provides tools for processing incomplete information expressedby fuzzy propositions

• it plays major role in fuzzy logic and approximate reasoning

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 51 / 61

Page 52: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Possibility versus Probability

• probability theory and possibility theory are distinct

• to show that, let us examine both theories by evidence theory

• probability theory and possibility theory are special branches of it

⇒ introduce probabilistic counterparts of possibilistic characteristics

• probability measure Pro must satisfy

Pro(A ∪ B) = Pro(A) + Pro(B)

for all A, B ∈ P(X ) s.t. A ∩ B = ∅

• Pro is special type of belief measure

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 52 / 61

Page 53: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Probability Distribution Function

Theorem

A belief measure Bel on a finite power set P(X ) is a probability

measure if and only if the associated basic probability assignment

function m is given by m(x) = Bel(x) and m(A) = 0 for all

subsets of X that are no singletons.

⇒ probability measures on finite sets are thus fully represented by

p : X → [0, 1]

s.t. p(x) = m(x)

• p is called probability distribution function

• let p = (p(x) | x ∈ X ) be probability distribution on X

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 53 / 61

Page 54: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Probability Measure

• recall definition of belief and plausibility

Bel(A) =∑

B|B⊆A

m(B) and Pl(A) =∑

B|A∩B 6=∅

m(B)

• if basic probability assignment focuses only on singletons, then

∀A ∈ P(X ) : Bel(A) = Pl(A) =∑

x∈A

m(x)

• Bel and Pos are equal ⇒ it is convenient to use one symbol Pro

∀A ∈ P(X ) : Pro(A) =∑

x∈A

p(x)

• Pro is called probability measure

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 54 / 61

Page 55: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Comparison of both Theories

Property Probability Theory Possibility Theory

measures probability Pro possibility Pos, necessity Nec

body of evidence singletons family of nested subsets

representation p : X → [0, 1] via r : X → [0, 1] viaPro(A) =

x∈Ap(x) Pos(A) = maxx∈A r(x)

normalization∑

x∈Xp(x) = 1 maxx∈X r(x) = 1

total ignorance ∀x ∈ X : p(x) = 1|X |

∀x ∈ X : r(x) = 1

noninteraction p(x , y) = pX (x) · pY (y) r(x , y) = minrX (x), rY (y)

independence pX |Y (x |y) = pX (x) rX |Y (x |y) = rX (x)pY |X (y |x) = pY (y) rY |X (y |x) = rY (y)

independence ⇔ noninteraction independence ⇒ noninteraction

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 55 / 61

Page 56: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Comparison of both Theories

Probability Theory Possibility Theory

Pro(A ∪ B) = Pro(A) + Pro(B) − Pro(A ∩ B) Pos(A ∪ B) = maxPos(A), Pos(B)Nec(A ∩ B) = minNec(A), Nec(B)

not applicable Nec(A) = 1 − Pos(AC )Pos(A) < 1 ⇒ Nec(A) = 0Nec(A) > 0 ⇒ Pos(A) = 1

Pro(A) + Pro(AC ) = 1 Pos(A) + Pos(AC ) ≥ 1Nec(A) + Nec(AC ) ≤ 1maxPos(A), Pos(AC ) = 1minNec(A), Nec(AC ) = 0

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 56 / 61

Page 57: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

FUZZY MEASURES: monotonic and continuousor semi continuous

BELIEF

MEASURES:

superadditiveand continuous

from above

PLAUSIBILITY

MEASURES:

subadditiveand continuous

from below

PROBABILITY

MEASURES:

additive

NEC.

MEASURES

POS.

MEASURES

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 57 / 61

Page 58: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Discussion

• possibility, necessity and probability measures do not overlap

• except for one special measure characterized by only one focalelement (i.e., singleton)

⇒ distribution functions of possibility and probability become equal

• one element of universal set is assigned 1, all others assigned 0

⇒ this measure represents perfect evidence

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 58 / 61

Page 59: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Discussion

• differences in mathematical properties make each theory suitablefor modeling certain types of uncertainty

• probability theory is ideal for formalizing uncertainty in situations• where class frequencies are known or• evidence is based on outcomes of many independent random

experiments

• possibility theory is ideal for formalizing incomplete informationexpressed by fuzzy propositions

• necessity and possibility can be seen as lower and upperprobabilities [Dempster, 1967]

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 59 / 61

Page 60: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Interpretations of Possibility Theory

based on similarity

• possibility r(x) reflects degree of similarity between x andprototype xi having r(xi) = 1

• i.e., r(x) is expressed by distance between x and xi

• the closer x to xi , the more possible its interpretation

• closeness comes from measurement or subjective judgement

based on preference

• due to ordering ≤Pos defined on P(X )

• ∀A, B ∈ P(X ), A ≤Pos B means B is at least as possible as A

• A ≤Pos B ⇔ AC ≤Nec BC

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 60 / 61

Page 61: Fuzzy Systems - Possibility Theoryfuzzy.cs.ovgu.de/.../Lehre.FS0910/fs0910lecture07.pdf · 2009-12-15 · 1. Introduction 2. Dempster-Shafer Theory of Evidence Belief Measure Plausibility

Literature for the Lecture

Choquet, G. (1954).Theory of capacities.Annales de l’Institut Fourier, 5:131–295.

Dempster, A. P. (1967).Upper and lower probabilities induced by a multivalued mapping.The Annals of Mathematical Statistics, 38(2):325–339.

Dubois, D. and Prade, H. (1993).Fuzzy sets and probability: Misunderstanding, bridges and gaps.In Proceedings of the Second IEEE International Conference on Fuzzy Systems, pages1059–1068, San Francisco, CA, USA.

Shafer, G. (1976).A Mathematical Theory of Evidence.Princeton University Press.

R. Kruse, C. Moewes Fuzzy Systems – Possibility Theory 2009/12/16 61 / 61