Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Generalizations
of Fuzzy
Information
Measures
Generalizations of Fuzzy Information Measures
Anshu Ohlan Ramphul Ohlan
Generalizations of Fuzzy
Information Measures
123
Anshu Ohlan Ramphul Ohlan
Department of Mathematics Institute of Management Studies and
Deenbandhu Chhotu Ram University of Research
Science and Technology Maharshi Dayanand University
Murthal, Haryana Rohtak, Haryana
India India
v
Contents
vii
viii Contents
Fuzzy set theory was rst proposed by Zadeh [94] as a generalization of classical
(crisp) set theory. It is based on the idea that in the real world a precise description
of many situations is nearly impossible, and that imprecisely dened classes play an
important role in human thinking and natural language. In natural language, many
words have ambiguous meanings that are properly quantied and analyzed with the
help of fuzzy set theory. Over the last six decades, research on fuzzy set theory and
its application to different areas have been extended evenly. The theory of fuzzy sets
has been proven useful in the context of decision theory, pattern recognition,
medical diagnosis, control theory, etc.
In our life we mostly use the crisp sets. A crisp set is a collection of things that
belongs to a denition. In a crisp set any item either belongs or does not belong to
the set. For a crisp set A, the characteristic function is dened as vA : X ! f0; 1g
1; if x 2 A
vA
0; if x 62 A
The set A fx; vA xg is said to be a crisp set, i.e. the characteristic function
having values either of 0 or 1 in the classical set. Thus, the crisp set is also known as
the classical (or well-dened sharp) set. Finally, a classical set A is the collection of
all those members of universe of discourse X for which vA x 1.
A fhx; lA xi=x 2 X g;
The key difference between a crisp set and a fuzzy set is their membership function.
A crisp set has unique membership function, whereas a fuzzy set can have an innite
number of membership functions to represent it. For example, we can dene a
possible membership function for the set of real numbers close to 0 as follows [48]:
4 1 Fundamentals of Fuzzy Information Measures
1
lA x ; x 2 R:
1 10x2
Here the number 3 is assigned a grade of 0.01, the number 1 is assigned a grade
of 0.09 and the number 0 is assigned a grade of 1. Thus, the concepts of crisp sets
are incorporated by fuzzy sets as a special case.
Most commonly used crisp operations are extended up by the operations on fuzzy
sets. This extension imposes a key condition that all the fuzzy operations, which are
just the extensions of crisp concepts must reduce to their natural meaning where the
fuzzy sets have only 1 and 0 as the membership values. For dening the following
operations, it is assumed that A and B are two fuzzy subsets of universe of discourse
X fx1 ; x2 ; . . .; xn g with the membership functions lA x and lB x; x denotes an
arbitrary element of X.
Zadeh [94] gave some notions related to fuzzy sets which are as follows:
(1) Containment; A B , lA x lB x for all x 2 X.
(2) Equality; A B , lA x lB x for all x 2 X.
(3) Complement; A Complement of A , lA x 1 lA x for all x 2 X.
(4) Union; A [ B Union of A and B , lA [ B x maxflA x; lB xg for all
x 2 X.
(5) Intersection; A \ B Intersection of A and B , lA \ B x minflA x;
lB xg for all x 2 X.
(6) Product; AB Product of A and B , lAB x lA xlB x for all x 2 X.
(7) Sum; A B Sum of A and B , lAB x lA x lB x lA xlB x for
all x 2 X.
A \ A A; A [ A A; A A:
A \ B B \ A; A [ B B [ A:
1.1 Introduction to Fundamental Concepts of Fuzzy Set Theory 5
A \ B \ C A \ B \ C; A [ B [ C A [ B [ C:
A \ B [ C A [ C \ B [ C; A [ B \ C A \ C [ B \ C:
A [ A \ B A; A \ A [ B A:
A [ U U; A \ / /:
A [ / A; A \ U A:
A [ B A \ B; A \ B A [ B:
Some important basic concepts related to fuzzy set theory are dened as given
below
(i) Equality of Two Fuzzy Sets
Two fuzzy sets are said to be equal if and only if lA xi lB xi ; 8xi 2 X.
(ii) Standard Fuzzy Sets
Fuzzy sets are said to be standard if lA xi 0:5; 8xi 2 X.
(iii) Support of a Fuzzy Set
Given a fuzzy set A which is a subset of the universal set X fx1 ; x2 ; . . .; xn g,
the support of A, denoted by SuppA, is an ordinary set dened as the set of
elements whose degree of membership in A is greater than 0, i.e.
SuppA fxi 2 X=lA xi [ 0g.
6 1 Fundamentals of Fuzzy Information Measures
There have been several attempts to quantify the uncertainty associated with fuzzy
sets. Zadeh [95] rst introduced the concept of fuzzy entropy as a measure of
fuzziness. Fuzzy entropy is an important concept for measuring fuzzy information.
A measure of the fuzzy entropy of a fuzzy set is a measure of the fuzziness of the
set. Fuzzy entropy have been widely applied in many elds such as pattern
recognition, image processing, speech recognition, bioinformatics, fuzzy aircraft
control, feature selection, etc.
Since lA xi and 1 lA xi for all i 1; 2; . . .; n give the same degree of
fuzziness, therefore, De Luca and Termini [21] introduced an axiomatic structure of
the measure of fuzzy entropy corresponding to Shannon [74] entropy as
X
n
HA lA xi log lA xi 1 lA xi log1 lA xi 1:1
i1
lA xi lA xi ; if lA xi 0:5; 8i
and lA xi lA xi ; if lA xi 0:5; 8i:
l xi
UA xi Pn A ; i 1; 2; . . .n is a probability distribution:
i1 lA xi
1.2 Fuzzy Entropy Measures and Their Generalizations 7
1 X n
HA UA xi log UA xi 1:2
log n i1
Bhandari and Pal [8] extended the probabilistic exponential entropy idea of Pal
and Pal [66] to the fuzzy phenomenon and dened the exponential fuzzy entropy
for a fuzzy set A as
1 X n
He A p l xi e1lA xi 1lA xi elA xi 1 1:3
n e 1 i1 A
Kapur [46] argues that fuzzy entropy measures uncertainty due to fuzziness of
information, while probabilistic entropy measures uncertainty due to the informa-
tion being available in terms of a probability distribution only. The advantage of
making probability theory work in fuzzy environment is to deal with a different
kind of uncertainties that may arise within the same problem.
In fuzzy context, several measures have been proposed in order to measure the
degree of difference between two fuzzy sets. Measure of fuzzy divergence between
two fuzzy sets gives the difference between two fuzzy sets and this measure of
difference between two fuzzy sets is called the fuzzy divergence measure.
Analogous to Kullback and Leibler [51] measure of divergence, Bhandari and Pal
[8] introduced divergence measure between two fuzzy sets A and B of universe of
discourse X fx1 ; x2 ; . . .; xn g having the membership values lA xi and lB xi , as
n
X
lA xi 1 lA xi
IA : B lA xi log 1 lA xi log 1:4
i1
lB xi 1 lB xi
Xn
lA xi 1 lB xi
JA : B IA : B IB : A lA xi lB xi log
i1
lB xi 1 lA xi
1:5
8 1 Fundamentals of Fuzzy Information Measures
1:6
Fan and Xie [27] proposed the fuzzy information of discrimination of A against
B corresponding to the exponential fuzzy entropy of Pal and Pal [66] and is given
by
n h
X i
IA; B 1 1 lA xi elA xi lB xi lA xi elB xi lA xi 1:7
i1
d1 : DA : B DB : A
d2 : DA : A 0
d3 : maxfDA [ C; B [ C; DA \ C; B \ Cg DA; B
1 X n
Ia A; B la xi lB1a xi 1 lA xi a 1 lB xi 1a 1;
a 1 i1 A 1:8
a 6 1; a [ 0:
Ia A : B
n
1X 1 alA xi 1 a1 lA xi
1 alA xi log f1 a1 lA xi g log ; a[0
a i1 1 alB xi 1 a1 lB xi
1:9
Corresponding to Renyi [70] and Sharma and Mittal [75] generalized measure of
divergence Bajaj and Hooda [5] provided the generalized fuzzy divergence mea-
sures which are given by
1 X n
a
Da A; B loglaA xi l1a
B xi 1 lA xi 1 lB xi
1a
;
a 1 i1 1:11
a 6 1; a [ 0:
and
n h
X i
1 a 1a b1
Da;b A; B laA xi l1a
B xi 1 lA xi 1 lB xi a1 1 ;
21b 1 i1
a 6 1; a [ 0; b 6 1; b [ 0:
1:12
Singh and Tomar [77], Tomar and Ohlan [84] have dened and studied some of
symmetric and non-symmetric fuzzy divergence measures analogous to proba-
bilistic divergence measures and inequalities among them. They also presented a
number of renement of inequalities among fuzzy divergence measures. Bhatia and
Singh [9] proposed three families of fuzzy divergence
(i) The fuzzy divergence measure corresponding to Taneja [82]
Arithmetic-Geometric divergence measure is given by
2 3
lA xi lB xi lA xi lB xi
n 6
log p 7
X 6 2 2 lA xi lB xi 7
TA; B 6 7
4 2 l x i l x i 2 l x i l x i 5
i1 A B
log p
A B
2 2 1 lA xi 1 lB xi
1:13
10 1 Fundamentals of Fuzzy Information Measures
(iii) A a; b class of measures of fuzzy divergence for two arbitrary fuzzy sets
A and B corresponding to Taneja [82]
n h i
1 X a 1a b1
Dba A; B laA xi l1a
B xi 1 lA xi 1 lB xi a1 1 ;
b 1 i1
a 6 1; a [ 0; b 6 1; b [ 0:
1:15
Atanassovs intuitionistic fuzzy set is expressed with the ordered pair of real
numbers lx; mx and 1 lx mx is called the degree of hesitancy.
Moreover, Gau and Buehrer [31] introduced the notion of vague sets. But, Bustince
and Burillo [14] presented that the notion of vague sets was equivalent to that of
Atanassov IFSs. The investigation on the IFS theory has been extended exponen-
tially in last decades and applied in many areas as decision-making, medical
diagnosis, pattern recognition, linguistic variables, etc.
Atanassov [4] further dened set operations on intuitionistic fuzzy sets as follows:
Let A; B 2 IFSX given by
(i) A
B iff lA xi lB xi and mA xi mB xi 8xi 2 X.
(ii) A B iff A
B and B
A.
(iii) Ac fhxi ; mA xi ; lA xi i=xi 2 Xg.
(iv) A [ B fhxi ; maxlA xi ; lB xi ; minmA xi ; mB xi i=xi 2 X g.
(v) A \ B fhxi ; minlA xi ; lB xi ; maxmA xi ; mB xi i=xi 2 X g.
12 1 Fundamentals of Fuzzy Information Measures
Hung and Yang [40] dened an axiomatic structure of the divergence measure
between IFSs using the Hausdorff distance as follows:
1
dH A; B HIA xi ; IB xi ; 1:17
n
where
" #
X
n
lA xi mA xi
IIFS A; B lA xi In 1 mA xi In 1 ;
i1 2 lA xi lB xi 2 mA xi mB xi
where I l A; B In llA x i
xi is the amount of discrimination of lA xi from lB xi .
B
1.5 Intuitionistic Fuzzy Divergence Measures 13
Where
2 3
lA xi 1 mA xi lA xi 1 mA xi =2
n 6
log 7
X 6 2 21
4 flA xi 1 mA xi lB xi 1 mB xi g 7
DA; B 6 7
4 1 l xi mA xi 1 l x i m A xi =2 5
i1 A
log2 1 A
2 4 f 1 l A xi m A x i 1 l B xi mB xi g
Li [54] also introduced the intuitionistic fuzzy dissimilarity measure between IFSs.
Hung and Yang [41] constructed J-divergence between IFSs and applied it to pattern
recognition. Further, Papakostas et al. [67] provided a comparative analysis of dis-
tance and similarity measures between IFSs from a pattern recognition point of view.
Divergence measures of IFSs have been widely applied to many elds such as
pattern recognition [33, 41, 67, 86], linguistic variables [40], medical diagnosis [22,
100], logical reasoning [43] and decision-making [55, 86, 87, 101]. Since the
divergence measures of IFSs have been applied to many real-world situations, it is
expected to have an efcient divergence measure which deals with the aspect of
uncertainty, i.e. fuzziness and non specicity or lack of knowledge.
In Chap. 7 we introduce an exponential methodology for measuring the degree
of divergence between two intuitionistic fuzzy sets. For this an intuitionistic fuzzy
exponential divergence measure is proposed and its important properties are dis-
cussed axiomatically. In addition, the applicability and efciency of the proposed
intuitionistic fuzzy exponential divergence measure have been demonstrated by
comparing it with existing intuitionistic fuzzy divergence measures using a
numerical example in the framework of pattern recognition.
1.6 Decision-Making
As human beings daily face situations in which they should choose among the
alternatives, thus decision-making is inherent to mankind. Decision problems have
been classied in decision theory attending to their framework and elements
alternatives, or ranking alternatives, from all feasible alternatives (options) based on
a set of attributes or criteria [25]. It can be seen as a process composed of different
phases such as information gathering, analysis and selection based on different
rational and reasoning processes that lead to choose a suitable alternative among a
set of possible alternatives [19, 26, 58, 60]. Thus, decision-making deals with the
problem of choosing the best alternative, that is, the one with the highest degree of
satisfaction for all the appropriate criteria or goals.
14 1 Fundamentals of Fuzzy Information Measures
There are the some important reasons why the fuzzy set theory is applicable in
different areas of research:
Uncertainty and vagueness of such problems
Mostly, the information that is required for development of a new model to
solve the problem is not complete and measureable
The qualitative and quantitative information is reduced due to the inexact
information and biased estimation of the people concerned.
In view of importance of fuzzy set theory we develop the different methods to
solve decision-making problems in fuzzy environment utilizing the proposed
generalized fuzzy divergence measures. Although a lot of multi-criteria
decision-making (MCDM) methods are now available to deal with varying eval-
uation and selection problems, in the present study we explore the applicability of
new MCDM methods that are simple, efcient and computationally easy. For this, a
16 1 Fundamentals of Fuzzy Information Measures
Pattern recognition is one of the problems that need to recognize that the given
pattern belongs to what class, among several available classes. The process of the
pattern recognition in fuzzy environment by use of an algorithm is given below.
We here present an algorithm to recognize and classify the given pattern in the
pattern recognition problems given in Vlachos and Sergiadis [88].
Suppose that we are given m known patterns P1 ; P2 ; P3 ; . . .; Pm which have
classications C1 ; C2 ; C3 ; . . .; Cm , respectively. The patterns are represented by the
following fuzzy sets in the universe of discourse X fx1 ; x2 ; . . .; xn g:
Pi xj ; lPi xj xj 2 X ;
where i 1; 2; . . .; m and j 1; 2; . . .; n.
Given an unknown pattern Q, represented by the fuzzy set
Qi xj ; lQi xj xj 2 X :
According to this algorithm, the given pattern can be recognized so that the best
class can be selected. It is a practical application of minimum divergence measure
principle of Shore and Gray [76] to pattern recognition.
This algorithm is used by us in Chap. 4 to show that Q has been classied to
same class correctly by using all of the proposed fuzzy mean difference divergence
1.7 Pattern Recognition 17
measures. This algorithm is also applied in Chaps. 5 and 6 to illustrate that the
proposed generalized fuzzy divergence measures are applicable in pattern recog-
nition and medical diagnosis problems. Moreover, the above algorithm is used in
Chap. 7 to demonstrate the efciency and applicability of the proposed intuitionistic
fuzzy exponential divergence measure by comparing it with the existing intu-
itionistic fuzzy divergence measures.
In this chapter, we have laid down the foundations for this books motivation, and
have given a high-level overview of its main aim: to present the applications of the
fresh generalizations of fuzzy information and divergence measures. First, the
18 1 Fundamentals of Fuzzy Information Measures
extant literature offers several fuzzy information and divergence measures. But,
there is still much scope for development of the better information-theoretic mea-
sures which will deal with the real-world problems in pattern recognition, image
processing, speech recognition, bioinformatics, fuzzy aircraft control, feature
selection, etc. Second, the generalization of information measures improves their
flexibility from the application point of view. Third, generalized measures have
capability to properly and efciently handle the information which deals with the
different aspects of uncertainty.
The rest of this book is devoted to fleshing out the generalizations of fuzzy
information measures proposed, and to illustrate these with a number of illustrative
examples. Chapter 2 provides two parametric generalizations of Rnorm infor-
mation and divergence measures and establishes a relation between them.
A generalized exponential measure of fuzzy divergence and a method to solve the
problem related to strategic decision-making are given in Chap. 3. Thereafter, a
sequence of fuzzy divergence measures with their applicability to deal with lin-
guistic variables and dynamic pattern recognition is presented with discussion in
Chap. 4. In what follows, Chap. 5 relates a newly generalized fuzzy divergence
measure and method to solve the multi-criteria decision-making problems with the
help of illustrative examples.
In this way, generalization of Hellingers fuzzy divergence measure with its
application to multi-criteria decision-making and medical diagnosis is studied in
Chap. 6. An efcient and novel generalized exponential intuitionistic fuzzy diver-
gence measure and a method to solve multi-attribute decision-making are intro-
duced in Chap. 7. Finally, this book concludes by discussing outlook of other
application domains of proposed fuzzy generalized measures and some possible
further developments of this line of research.
References
1. Amiri-Aref M, Javadian N, Kazemi M (2012) A new fuzzy positive and negative ideal
solution for fuzzy TOPSIS. Wseas Trans Circ Syst 11(3):92103
2. Andreica ME, Dobre I, Andreica MI, Resteanu C (2010) A new portfolio selection method
based on interval data. Stud Inform Control 19(3):253262
3. Atanassov KT (1986) Intuitionistic fuzzy sets. Fuzzy Sets Syst 20:8796
4. Atanassov KT (1994) New operations dened over the intuitionistic fuzzy sets. Fuzzy Sets
Syst 61:137142
5. Bajaj RK, Hooda DS (2010) On some new generalized measures of fuzzy information.
World Acad Sci Eng Technol 62:747753
6. Bellman RE, Zadeh LA (1970) Decision-making in a fuzzy environment. Manage Sci
17:141164
7. Beynon M, Cosker D, Marshall D (2001) An expert system for multi-criteria decision
making using dempster shafer theory. Expert Syst Appl 20(4):357367
8. Bhandari D, Pal NR (1993) Some new information measures for fuzzy sets. Inf Sci 67
(3):209228
References 19
9. Bhatia PK, Singh S (2012) Three families of generalized fuzzy directed divergence.
AMO-Adv Model Optim 14(3):599614
10. Bouchon-Meunier B, Rifqi M, Bothorel S (1996) Towards general measures of comparison
of objects. Fuzzy Sets Syst 84:143153
11. Brans JP, Vincke P (1985) A preference ranking organization method: The PROMETHEE
method for MCDM. Manage Sci 31(6):647656
12. Brauers WKM, Zavadskas EK (2006) The MOORA method and its application to
privatization in transition economy. Control Cybern 35(2):443468
13. Brauers WKM, Zavadskas EK (2010) Project management by MULTIMOORA as an
instrument for transition economies. Tech Econ Dev Econ 16(1):524
14. Bustince H, Burillo P (1996) Vague sets are intuitionisic fuzzy sets. Fuzzy Sets Syst 79:
403405
15. Chen SJ, Chen SM (2005) A prioritized information fusion method for handling fuzzy
decision-making problems. Appl Intel 22(3):219232
16. Chen SJ, Hwang CL (1992) Fuzzy multiple attribute decision making-methods and
applications. Springer, Berlin
17. Chen SM, Lee LW (2010) Fuzzy multiple attributes group decision-making based on the
interval type-2 TOPSIS method. Expert Syst Appl 37(4):27902798
18. Chen SM, Wang CH (2009) A generalized model for prioritized multi-criteria decision
making systems. Expert Syst Appl 36(3):47734783
19. Clement RT (1996) Making hard decisions. An introduction to decision analysis. Duxbury
Press, Pacic Grove
20. Couso I, Janis V, Montes S (2000) Fuzzy divergence measures. Acta Univ M Belii 8:2126
21. De Luca A, Termini S (1972) A denition of non-probabilistic entropy in the setting of fuzzy
set theory. Inf Control 20(4):301312
22. De SK, Biswas R, Roy AR (2001) An application of intuitionistic fuzzy sets in medical
diagnosis. Fuzzy Sets Syst 117(2):209213
23. Deng JL (1989) Introduction to grey system. J Grey Syst 1(1):124
24. Devi K, Yadav SP, Kumar SP (2009) Extension of fuzzy TOPSIS method based on vague
sets. Comput Cogn 7:5862
25. Duncan R, Raiffa H (1985) Games and decision. Introduction and critical survey. Dover
Publications
26. Evangelos T (2000) Multi-criteria decision making methods: a comparative study. Kluwer
Academic Publishers, Dordrecht
27. Fan J, Xie W (1999) Distance measures and induced fuzzy entropy. Fuzzy Sets Syst 104
(2):305314
28. Flix P, Barroa S, Marn R (2003) Fuzzy constraint networks for signal pattern recognition.
Artif Intell 148:103140
29. Ferreri C (1980) Hyperentropy and related heterogeneity divergence and information
measures. Statistica 40(2):155168
30. Fu G (2008) A fuzzy optimization method for multi-criteria decision making: an application
to reservoir flood control operation. Expert Syst Appl 34(1):145149
31. Gau WL, Buehrer DJ (1993) Vague sets. IEEE Trans Syst Man Cybern 23:610614
32. Hartley RVL (1928) Transmission of information. Bell Syst Tech J 7(3):535563
33. Hatzimichailidis AG, Papakostas GA, Kaburlasos VG (2012) A novel distance measure of
intuitionistic fuzzy sets and its application to pattern recognition problems. Int J Intel Syst
27(4):396409
34. Havrada JH, Charvat F (1967) Quantication methods of classication processes: concept of
structural a-entropy. Kybernetika 3(1):3035
35. Herrera F, Herrera-Viedma E (1996) A model of consensus in group decision making under
linguistic assessments. Fuzzy Sets Syst 78(1):7387
36. Herrera F, Herrera-Viedma E (2000) Linguistic decision analysis: steps for solving decision
problems under linguistic information. Fuzzy Sets Syst 11(1):6782
37. Hooda DS (2004) On generalized measures of fuzzy entropy. Math Slovaca 54:315325
20 1 Fundamentals of Fuzzy Information Measures
38. Hooda DS, Jain D (2012) The generalized fuzzy measures of directed divergence, total
ambiguity and information improvement. Investig Math Sci. 2:239260
39. Hua Z, Gong B, Xu X (2008) A DSAHP approach for multi-attribute decision making
problem with incomplete information. Expert Syst Appl 34(3):22212227
40. Hung WL, Yang MS (2004) Similarity measures of intuitionistic fuzzy sets based on
Hausdorff distance. Pattern Recogn Lett 25:16031611
41. Hung WL, Yang MS (2008) On the J-divergence of intuitionistic fuzzy sets and its
application to pattern recognition. Inf Sci 178(6):16411650
42. Hwang CL, Yoon K (1981) Multiple attribute decision makingmethods and applications.
Springer, New York
43. Jiang YC, Tang Y, Wang J, Tang S (2009) Reasoning within intuitionistic fuzzy rough
description logics. Inf Sci 179:23622378
44. Joshi D, Kumar S (2014) Intuitionistic fuzzy entropy and distance measure based TOPSIS
method for multi-criteria decision making. Egypt Inf J 15:97104
45. Kahraman C, Cebi S (2009) A new multi-attribute decision making method: hierarchical
fuzzy axiomatic design. Expert Syst Appl 36(3):48484861
46. Kapur JN (1997) Measures of fuzzy information. Math Sci Trust Soc, New Delhi
47. Kaufmann A (1980) Fuzzy subsets: fundamental theoretical elements. Academic Press, New
York 3
48. Klir GJ, Folger TA (2009) Fuzzy sets, uncertainty and information. Prentice Hall, New York
49. Kosko B (1990) Fuzziness vs. probability. Int J Gen Syst 17:211240
50. Kulak O (2005) A decision support system for fuzzy multi-attribute selection of material
handling equipments. Expert Syst Appl 29(2):310319
51. Kullback S, Leibler RA (1951) On information and sufciency. Ann Math Stat 22(1):7986
52. Kuo MS, Tzeng GH, Huang WC (2007) Group decision-making based on concepts of ideal
and anti-ideal points in a fuzzy environment. Math Comput Model 45:324339
53. Kwon O, Kim M (2004) My message: case-based reasoning and multi-criteria decision
making techniques for intelligent context-aware message ltering. Expert Syst Appl 27(3):
467480
54. Li DF (2004) Some measures of dissimilarity in intuitionistic fuzzy structures. J Comput
Syst Sci 68(1):115122
55. Li DF (2005) Multi-attribute decision-making models and methods using intuitionistic fuzzy
sets. J Comput Syst Sci 70(1):7385
56. Lin J (1991) Divergence measure based on Shannon entropy. IEEE Trans Inf Theory 37(1):
145151
57. Lin HY, Hsu PY, Sheen GJ (2007) A fuzzy-based decision-making procedure for data
warehouse system selection. Expert Syst Appl 32(3):939953
58. Lu J, Zhang G, Ruan D (2007) Multi-objective group decision making. Methods, software
and applications with fuzzy set techniques. Imperial College Press
59. MacCrimon KR(1968) Decision marking among multiple-attribute alternatives: a survey and
consolidated approach. RAND memorandum, RM-4823-ARPA. The Rand Corporation,
S Martnez anta Monica, California
60. Martnez L (2010) Computing with words in linguistic decision making: Analysis of
Linguistic Computing Models. IEEE 58
61. Mikhailov L (2003) Deriving priorities from fuzzy pair wise comparison judgments. Fuzzy
Sets Syst 134:365385
62. Montes S, Couso I, Gil P, Bertoluzza C (2002) Divergence measure between fuzzy sets. Int J
Approximate Reasoning 30:91105
63. Nyquist H (1924) Certain factors affecting telegraph speed. Bell Syst Tech J 3(2):324346
64. Nyquist H (1928) Certain topics in telegraph transmission theory. A.I.E.E. Trans 47:
617644
65. Opricovic S (1998) Visekriterijumska optimizacija u graevinarstvumulti-criteria opti-
mization of civil engineering systems. Faculty of Civil Engineering, Belgrade (Serbian)
References 21
66. Pal NR, Pal SK (1989) Object background segmentation using new denition of entropy.
IEE Proc Comput Digital Tech 136(4):248295
67. Papakostas GA, Hatzimichailidis AG, Kaburlasos VG (2013) Distance and similarity
measures between intuitionistic fuzzy sets: a comparative analysis from a pattern recognition
point of view. Pattern Recogn Lett 34(14):16091622
68. Parkash O, Sharma PK, Kumar S (2006) Two new measures of fuzzy divergence and their
properties. SQU J Sci 11:6977
69. Pirdashti M, Ghadi A, Mohammadi M, Shojatalab G (2009) Multi-criteria decision-making
selection model with application to chemical engineering management decisions. World
Academy Sci Eng Technol 3(1):2732
70. Renyi A (1961) On measures of entropy and information. In: Proceeding of Fourth Berkeley
Symposium on Mathematics, Statistics and Probability, vol 1, 547561
71. Roy B (1991) The outranking approach and the foundation of ELECTRE methods. Theor
Decis 31:4973
72. Saaty TL (1980) The analytic hierarchy process: planning, priority setting, resource
allocation. McGraw-Hill, New York
73. Shang X, Jiang G (1997) A note on fuzzy information measures. Pattern Recogn Lett 18
(5):425432
74. Shannon CE (1948) The mathematical theory of communication. Bell Syst Tech J 27(3):
379423
75. Sharma BD, Mittal DP (1977) New non-additive measures of relative information. J Comb
Inf Syst Sci 2:122133
76. Shore JE, Gray RM (1982) Minimization cross-entropy pattern classication and cluster
analysis. IEEE Trans Pattern Anal Mach Intel 4(1):1117
77. Singh RP, Tomar VP (2010) Renement of inequalities among fuzzy divergence measures.
Adv Appl Res 2(2):142156
78. Singpurwalla ND, Booker JM (2004) Membership functions and probability measures of
fuzzy sets. J Am Stat Assoc 99:867877
79. Stanujkic D, orevi B, orevi M (2013) Comparative analysis of some prominent
mcdm methods: a case of ranking serbian banks. Serbian J Manag 8(2):211241
80. Szmidt E, Kacprzyk J (2000) Distances between intuitionistic fuzzy sets. Fuzzy Sets Syst
114(3):505518
81. Tacker EC, Silvia MT (1991) Decision making in complex environments under conditions
of high cognitive loading: a personal expert systems approach. Expert Syst Appl 2(23):
121127
82. Taneja IJ (2008) On mean divergence measures. In: Barnett NS, Dragomir SS (eds) Advances
in Inequalities from probability theory and statistics. Nova, USA, pp 169186
83. Tomar VP, Ohlan A (2014a) New parametric generalized exponential fuzzy divergence
measure. J Uncertain Anal Appl 2(1):114
84. Tomar VP, Ohlan A (2014b) Sequence of fuzzy divergence measures and inequalities. Adv
Model Optim 16(2):439452
85. Triantaphyllou E, Shu B, Nieto Sanchez S, Ray T (1998) Encyclopedia of electrical and
electronics engineering, Webster JG (ed), John Wiley and Sons, New York, NY, 15, 175186
86. Verma R, Sharma BD (2011) On generalized exponential fuzzy entropy. World Acad Sci
Eng Technol 69:14021405
87. Verma R, Sharma BD (2012) On generalized intuitionistic fuzzy divergence (relative
information) and their properties. J Uncertain Syst 6(4):308320
88. Vlachos IK, Sergiadis GD (2007) Intuitionistic fuzzy information-application to pattern
recognition. Pattern Recognit Lett 28(2):197206
89. Wang WQ, Xin XL (2005) Distance measures between intuitionistic fuzzy sets. Pattern
Recogn Lett 26:20632069
90. Xu L, Yang JB (2001) Introduction to multi-criteria decision making and the evidential
reasoning approach. University of Manchester, Institute of Science and Technology,
Manchester School of Management. ISBN 1 86115 111 X
22 1 Fundamentals of Fuzzy Information Measures
91. Yager RR (1991) Non-monotonic set theoretic operations. Fuzzy Sets Syst 42(2):173190
92. Yager RR (1992) Second order structures in multi-criteria decision making. Int J Man Mach
Stud 36(6):553570
93. Yu PL (1973) A class of solutions for group decision problems. Manage Sci 19(8):936946
94. Zadeh LA (1965) Fuzzy sets. Inf Control 8:338353
95. Zadeh LA (1968) Probability measures of fuzzy events. J Math Anal Appl 23:421427
96. Zadeh LA (1987) Fuzzy sets and applications: Selected Papers by L.A. Zadeh. Wiley, New
York
97. Zadeh LA (1995) Discussion: probability theory and fuzzy logic are complementary rather
than competitive. Technometrics 37(3):271276
98. Zavadskas EK, Kaklauskas A, Sarka V (1994) The new method of multicriteria complex
proportional assessment of projects. Technol Econ Dev Econ 1(3):131139
99. Zeleny M (1973) Compromise programming. In: Cochrane JL, Zeleny M (eds) Multiple
criteria decision making. University of South Carolina Press, Columbia, SC, 262301
100. Zhang QS, Jiang SY (2008) A note on information entropy measures for vague sets and its
applications. Inf Sci 178:41844191
101. Zhang SF, Liu SY (2011) A GRA-based intuitionistic multi-criteria decision making method
for personnel selection. Expert Syst Appl 38(9):1140111405
102. Zimmermann HJ (1991) Fuzzy set theory and its applications. Kluwer Academic Publishers,
Second Edition, Boston, MA
Chapter 2
Parametric Generalized R-norm Fuzzy
Information and Divergence Measures
This chapter deals with two new generalized R-norm fuzzy entropy measures
and some of their interesting properties. In addition, we propose two new gener-
alized R-norm fuzzy divergence measures. In order to check the validity of pro-
posed measures, their essential properties are studied.
Entropy is very important for measuring uncertain information rst introduced
by Shannon [14] to measure the uncertain degree of the randomness in a probability
distribution. Let X is a discrete random variable with probability distribution P
p1 ; p2 ; . . .; Pn in an experiment. The information contained in this experiment is
given by
X
n
HP pi log pi 2:1
i1
measure, a numerical example is given in Sect. 2.3. In Sect. 2.4 we review some
existing R-norm fuzzy measures of divergence. Section 2.5 introduces and validates
two new parametric generalized R-norm fuzzy divergence measures. Some of
interesting properties of these two proposed fuzzy divergence measures are estab-
lished and proved in Sect. 2.6. A relation between the new generalized R-norm
measures of fuzzy information and divergence is established in Sect. 2.7. The
concluding remarks are drawn in the last section.
X
n
HA lA xi log lA xi 1 lA xi log1 lA xi 2:2
i1
X n h i
1
EA p lA xi e1lA xi 1 lA xi elA xi 1 2:3
n e 1 i1
Boekee and Lubbe [3] dened and studied R-norm information measure, which
has been mentioned by Arimoto [1] as an example of a generalized class of
information measures of the distribution P for R 2 R as given by
" #
R Xn
R R1
HR P 1 pi ; R [ 0; R 6 1: 2:4
R1 i1
2.1 Generalized R-norm Fuzzy Information Measures 25
Further, from the signicant studies it is noted that Hooda and Bajaj [6] and
Hooda and Jain [7] provide the generalization of R-norm fuzzy information mea-
sure (2.5) corresponding to generalized R-norm information measures proposed by
Hooda and Ram [8] and Hooda and Sharma [9] respectively.
Kumar [11] generalized the measure (2.4) and gave R-norm measure of infor-
mation of order a which is
2 !Ra 3
R X
n
41 pi 5 ;
R
HRa P a
0\a 1; R [ 0 6 1: 2:6
Ra i1
In addition, Kumar and Choudhary [12] generalized the measure (2.4) and gave
the R-norm information measure of degree m as
2 !Rm1 1 3
R m14 X
n
HRm P 1 pRm 1 5;
Rm i1
i
2:7
R m 1 [ 0; R 6 m; R; m [ 06 1:
n n R oa
R X
HRa A
R R
1 lAa xi 1 lA xi a ; 0\a 1; R [ 0 6 1:
R a i1
2:8
n n o 1
R m1X Rm 1 Rm 1 Rm 1
HRm A 1 lA xi 1 lA xi ;
R m i1 2:9
R m 1 [ 0; R 6 m; R; m [ 06 1:
Theorem 2.1 The generalized R-norm fuzzy information measures HRa A and
HRm A are valid measures of fuzzy information.
Proof P1 (Sharpness): The measures (2.8) and (2.9) clearly satisfy the property
P1, i.e. HRa A 0 and HRm A 0 if and only if A is non-fuzzy set or crisp set.
P2 (Maximality): To verify that the proposed measure (2.8) is concave; the
values of HRa A are computed rst for a xed value of R and different values of a,
second for a xed value of a and different values of R.
Case 1: Let us assume a particular value of R 0:5 and different values of a.
The computed values of HRa A using (2.8) with different values of a are given in
Table 2.1 which show the concavity of HRa A with respect to a.
A glance at Fig. 2.1 more precisely shows the concave behaviour of HRa A with
respect to a.
Case 2: Let us assume a particular value of a 0:7 and different values of
R. The computed values of HRa A for a 0:7 using (2.8) and different values of
R are given in Table 2.2.
Table 2.2 and Fig. 2.2 clearly depict the concavity of HRa A with respect to
R using (2.8).
Thus HRa A is a concave function with respect to a and R and its maximum
value exists at lA xi 0:5.
Now to verify that the proposed measure HRm A in (2.9) is concave; the values
of HRm A rst for a xed value of R and different values of m, second for a xed
value of m and different values of R are computed.
Case 1: Let us assume a particular value of R 0:5 and different values of
m. The computed values of HRm A using (2.9) with different values of m are given
in Table 2.3 and Fig. 2.3 which depict the concavity of HRm A using (2.9) with
respect to m.
Case 2: Let us assume a particular value of m 0:7 and different values of
R. The computed values of HRm A for m 0:7 using (2.9) and different values of
R are given in Table 2.4.
It may be seen from the results presented in Table 2.4 and Fig. 2.4 that beha-
viour of HRm A with respect to R is concave.
Table 2.1 Different values of HRa A for R 0:5 and for different values of a
a lA xi 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.1 0:1 0 0.1250 0.2498 0.3725 0.4812 0.7179 0.4812 0.3725 0.2498 0.1250 0
H0:5 A
0.3 0:3 0 0.2155 0.3832 0.5054 0.5802 0.6053 0.5802 0.5054 0.3832 0.2155 0
H0:5 A
0.4 0:4 0 0.2705 0.4437 0.5588 0.6254 0.6472 0.6254 0.5588 0.4437 0.2705 0
H0:5 A
0.7 0:7 0 0.4319 0.6124 0.7204 0.7797 0.7988 0.7797 0.7204 0.6124 0.4319 0
H0:5 A
1.0 1:0 0 0.6 0.8 0.9165 0.9798 1 0.9798 0.9165 0.8 0.6 0
H0:5 A
2.2 New Parametric Generalized R-norm Fuzzy Information Measures
27
28 2 Parametric Generalized R-norm Fuzzy Information
Thus HRm A is a concave function with respect to m and R and its maximum
value exists at lA xi 0:5.
Hence the measures (2.8) and (2.9) satisfy the property P2.
P3 (Resolution): Since HRa A and HRm A are increasing functions of lA xi in
the range 0; 0:5 and decreasing function in the range 0:5; 1, therefore
HRa A HRa A:
X
n
1 maxflA xi ; 1 lA xi g:
i1
Table 2.2 Different values of HRa A for a 0:7 and for different values of R
R lA xi 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.1 0:7 0 6.8071 8.6641 9.7356 10.3152 10.5 10.3152 9.7356 8.6641 6.8071 0
H0:1 A
0.2 0:7 0 1.2087 1.5379 1.7276 1.8301 1.8627 1.8301 1.7276 1.5379 1.2087 0
H0:2 A
0.4 0:7 0 0.5275 0.7174 0.8289 0.8896 0.9090 0.8896 0.8289 0.7174 0.5275 0
H0:4 A
0.6 0:7 0 0.3698 0.5467 0.6552 0.7153 0.7348 0.7153 0.6552 0.5467 0.3698 0
H0:6 A
0.8 0:7 0 0.2911 0.4652 0.5783 0.6429 0.6640 0.6429 0.5783 0.4652 0.2911 0
H0:8 A
2.2 New Parametric Generalized R-norm Fuzzy Information Measures
29
30 2 Parametric Generalized R-norm Fuzzy Information
X1 fx= x 2 X; lA xi lB xi g 2:10
X2
n h i
R X R R a
HRa A \ B 1 lAa \ B xi 1 lA \ B xi a R
R a i1
2X R R a
3
1 lBa xi 1 lB xi a R 2:13
R 66 X1
7
7
6 X 7
R a4 R R a 5
1 lA xi 1 lA xi
a a R
X2
Table 2.4 Different values of HRa A for m 0:7 and for different values of R
R lA xi 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.1 0:7 0 0.7630 0.9920 1.1244 1.1961 1.219 1.1961 1.1244 0.9920 0.7630 0
H0:1 A
0.2 0:7 0 0.6000 0.8000 0.9165 0.9798 1 0.9798 0.9165 0.8000 0.6000 0
H0:2 A
0.4 0:7 0 0.4396 0.6207 0.7288 0.7881 0.8071 0.7881 0.7288 0.6207 0.4396 0
H0:4 A
0.6 0:7 0 0.3549 0.5313 0.6402 0.7009 0.7205 0.7009 0.6402 0.5313 0.3549 0
H0:6 A
0.8 0:7 0 0.3004 0.4749 0.5872 0.6510 0.6718 0.6510 0.5872 0.4749 0.3004 0
H0:8 A
33
34 2 Parametric Generalized R-norm Fuzzy Information
HR A1=2 [ HR A [ HR A2 [ HR A3 [ HR A4 : 2:16
Table 2.6 Numerical values of the R-norm fuzzy entropy measure HRa
HRa A1=2 A A2 A3 A4
R 0:1 51.2045 47.8754 33.5312 22.6947 15.6592
R 0:5 3.8610 3.5240 2.1409 1.2058 0.6798
R 0:9 3.0597 3.3325 1.3683 0.5984 0.3062
R2 2.5640 2.1088 0.8703 0.3548 0.1527
R5 2.2744 1.7958 0.6744 0.2698 0.1156
clarify that the proposed new generalized R-norm fuzzy entropy measures HRa A
and HRm A satisfy the same:
Thus, the behaviour of new generalized R-norm fuzzy entropy measures HRa A
and HRm A is also consistent for the structured linguistic variables. Table 2.8 and
Fig. 2.5 display a declining trend in the numerical values of three entropy measures
corresponding to the logical order of fuzzy sets.
Figure 2.5 clearly depicts the similarity between the proposed entropy measures
and the existing one. An inequality among two proposed measures of fuzzy
information and the existing one may also be seen from Fig. 2.5. That is:
HR HRa HRm .
As mentioned in Chap. 1, in the recent years, many authors have introduced various
measures of divergence for probability distributions and their corresponding di-
vergence measures between fuzzy sets.
Hooda and Sharma [9] introduced the R-norm measure of divergence between
the probability distribution P p1 ; p2 ; . . .; pn and Q q1 ; q2 ; . . .; qn given by
2 !12 3
R 4 X n
DR P; Q pRi q1R 15; R [ 0; R 6 1: 2:19
R1 i1
X
n
pi
DP : Q pi log
i1
qi
Analogous to the measure (2.19), Hooda and Bajaj [6] dened the R-norm fuzzy
divergence measure for fuzzy sets A and B as
n h i
R X R 1R R1
IR A; B flRA xi l1R x i 1 l x
A i 1 l x
B i g 1
R 1 i1 B
R [ 0 6 1:
2:20
2.4 Generalized Measures of R-norm Fuzzy Divergence 37
JR A; B IR A; B IR B; A 2:21
and
2 3
1 1
R m 1 X
n flRm
A xi lRm
B xi
IRm A; B 4 5;
R m i1 1 lA xi Rm 1 1 lB xi Rm 1 gRm 1 1
1
R m 1 [ 0; R 6 m; R; m [ 06 1:
2:23
Theorem 2.4 The generalized R-norm fuzzy divergence measures IRa A; B and
IRm A; B dened in Eqs. 2.22 and 2.23 are valid measures of fuzzy divergence.
Proof It is clear from the denitions of measures IRa A; B and IRm A; B in (2.22)
and (2.23), respectively, that
(i) IRa A; B 0 and IRm A; B 0,
(ii) the equality holds if lA xi lB xi , 8i 1; 2; . . .; n.
(iii) We now prove the convexity of measure IRa A; B.
P
n P
n
Let us assume that lA xi a1 , lB xi b1 .
i1 i1
38 2 Parametric Generalized R-norm Fuzzy Information
where t1 Ra
R
and d1 Ra [ 0 6 1.
Then two cases arise.
Case 1: R\a
In this case we have t1 \0, d1 \1.
" #
X n
lA xi d1 lB xi 1d1
1 0
i1
a1 b1
2:24
X
n
or d1
lA xi lB 1d1
xi ad11 b1d
1
1
i1
Similarly,
X
n
1 lA xi d1 1 lB xi 1d1 n a1 d1 n b1 1d1 2:25
i1
ad11 b1d 1
n a1 d1 n b1 1d1 2:26
1
Case 2: R [ a
In this case we have t1 [ 0, d1 [ 1.
" #
X n
lA xi d1 lB xi 1d1
1 0
i1
a1 b1
2:27
X
n
or d1
lA xi lB 1d1
xi ad11 b1d
1
1
i1
2.5 Two New Generalized R-norm Fuzzy Divergence Measures 39
Similarly,
X
n
1 lA xi d1 1 lB xi 1d1 n a1 d1 n b1 1d1 2:28
i1
ad11 b1d 1
n a d1
n b 1d1 2:29
1 1 1
d1
Now let f a1 t ad11 b1d
1
1
n a 1 n b 1 1d1
n
So that we have
" #
0 a1 d1 1 n a1 d1 1
f a1 t1 d1 d1 2:30
b1 n b1
"
d1 2 d1 2 #
1 a 1 n a
f 00 a1 t1 d1 d1 1
1 1
[0 2:31
b1 b1 n b1 n b1
1
where t2 Rm
Rm and d2 R m 1 [ 0 6 1.
40 2 Parametric Generalized R-norm Fuzzy Information
i1
Similarly,
X
n
1 lA xi d2 1 lB xi 1d2 n a2 d2 n b2 1d2 2:33
i1
ad22 b1d 2
n a d2
n b 1d2 2:34
2 2 2
Case 2: R m 1 [ 1
In this case we have t2 [ 0, d2 [ 1.
" #
X n
lA xi d2 lB xi 1d2
1 0
i1
a2 b2
2:35
X
n
or d2
lA xi lB
1d2
xi ad22 b1d
2
2
i1
Similarly,
X
n
1 lA xi d2 1 lB xi 1d2 n a2 d2 n b2 1d2 2:36
i1
2.5 Two New Generalized R-norm Fuzzy Divergence Measures 41
a2 b1d 2
n a2 d2 n b2 1d2 2:37
2
d2
IRm A; B t2 ad22 b1d
2
2
n a 2 n b 2 1d2
n
So this shows that ga2 is a convex function of a2 whose minimum value exists
when ba2 nb
na2
, i.e. at a2 b2 and is equal to zero. So that ga2 [ 0 and vanishes
2 2
only when a2 b2 .
Hence in view of the denition of fuzzy divergence measure of Bhandari and Pal
[2] provided in Sect. 1.3 of Chap. 1, IRm A; B is a valid measure of R-norm fuzzy
divergence of type R and degree m and consequently it establishes the validity of
JRm A; B IRm A; B IRm B; A as measure of R-norm symmetric divergence of
type R and degree m between fuzzy sets A and B.
Limiting and Particular Cases:
(i) when a ! 1, m ! 1, measures (2.22) and (2.23) reduce to IR A; B .
(ii) when a ! 1, m ! 1 with R 1, measures (2.22) and (2.23) reduce to
Bhandari and Pal [2] fuzzy divergence measure IA; B.
The proposed generalized R-norm fuzzy divergence measures (2.22) and (2.23)
satisfy the following properties.
Theorem 2.5 For fuzzy sets A and B
(a) IRa A [ B; C IRa A; C IRa B; C and IRm A [ B; C IRm A; C IRm B; C.
(b) IRa A \ B; C IRa A; C IRa B; C and IRm A \ B; C IRm A; C IRm B; C.
42 2 Parametric Generalized R-norm Fuzzy Information
n h i
R m1X
xi lCRm 1 xi 1 lB xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
1
flRm
R m i1 B
n h i
R m1X 1
xi lCRm 1 xi 1 lA [ B xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
flARm
[B
R m i1
R m 1 X h Rm 1 i
xi lCRm 1 xi 1 lB xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
flB
R m X1
R m 1 X h Rm 1 i
xi lCRm 1 xi 1 lA xi Rm 1 1 lC xi Rm 1 gRm 1 1 0:
1
flA
R m X2
IRa A [ B; C
n h R i
R X 1R R R a
lAa [ B xi lC a xi 1 lA [ B xi a 1 lC xi 1a R 1
R a i1
Xh R 1R R R a
i
lAa xi lC a xi 1 lA xi a 1 lC xi 1a R 1
X1
Xh R 1R R R a
i
lBa xi lC a xi 1 lB xi a 1 lC xi 1 a R 1
X2
2:40
IRa A \ B; C
n h R i
R X 1R R R a
lAa \ B xi lC a xi 1 lA \ B xi a 1 lC xi 1a R 1
R a i1
Xh R 1R R R a
i
lBa xi lC a xi 1 lB xi a 1 lC xi 1a R 1
X1
Xh R 1R R R a
i
lAa xi lC a xi 1 lA xi a 1 lC xi 1 a R 1
X2
2:41
IRm A [ B; C
2 3
1
R m1X n flARm
[B xi lCRm 1 xi
4 5
R m i1 1 lA [ B xi Rm 1
1 lC xi Rm 1 Rm1 1
g 1
2 3
Rm 1 Rm 1
R m 1 X 4 flA xi lC xi
5
R m X1 1 lA xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
2 3
Rm 1
R m 1 X 4 flB xi lCRm 1 xi
5
R m X2 1 l x Rm 1 1 l x Rm 1 gRm 1 1
1
B i C i
2:42
44 2 Parametric Generalized R-norm Fuzzy Information
IRm A \ B; C
2 3
1 1
R m1X n flRm
A\B xi lRm
C xi
4 5
R m i1 1 lA \ B xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flB xi lRm
C xi
5
R m X1 1 lB xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flA xi lRm
C xi
5 2:43
R m X2 1 lA xi Rm 1 1 lC xi Rm 1 gRm 1 1
1
IRa A [ B; A \ B
n h R i
R X 1R R R a
lAa [ B xi lA \aB xi 1 lA [ B xi a 1 lA \ B xi 1a R 1
R a i1
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1a R 1
R a X1
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1 a R 1
R a X2
2:44
Now
IRa A \ B; A [ B
n h R i
R X 1R R R a
lAa \ B xi lA [aB xi 1 lA \ B xi a 1 lA [ B xi 1a R 1
R a i1
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1a R 1
R a X1
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1 a R 1
R a X2
2:45
2.6 Properties of New Generalized R-norm Fuzzy Divergence Measures 45
IRa A [ B; A \ B IRa A \ B; A [ B
n h R i
R X 1R R R a
lAa xi lB a xi 1 lA xi a 1 lB xi 1 a R 1
R a i1
n h R i
R X 1R R R a
lBa xi lA a xi 1 lB xi a 1 lA xi 1 a R 1
R a i1
IRa A; B IRa B; A
IRm A [ B; A \ B
2 3
1 1
R m1X n flARm
[B xi lRm
A\B xi
4 5
R m i1 1 lA [ B xi Rm 1 1 lA \ B xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flA xi lRm
B xi
5
R m X1 1 lA xi Rm 1 1 lB xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flB xi lRm
A xi
5
R m X2 1 l x Rm 1 1 l x Rm 1 gRm 1 1
1
B i A i
2:46
Now
IRm A \ B; A [ B
2 3
1 1
R m1X n flARm
\B xi lRm
A[B xi
4 5
R m i1 1 lA \ B xi Rm 1 1 lA [ B xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flB xi lRm
A xi
5
R m X1 1 lB xi Rm 1 1 lA xi Rm 1 gRm 1 1
1
2 3
Rm 1 1
R m 1 X 4 flA xi lRm
B xi
5
R m X2 1 l x Rm 1 1 l x Rm 1 gRm 1 1
1
A i B i
2:47
46 2 Parametric Generalized R-norm Fuzzy Information
IRm A [ B; A \ B IRm A \ B; A [ B
n h i
R m1X
flARm 1 xi lBRm 1 xi 1 lA xi Rm 1 1 lB xi Rm 1 gRm 1 1
1
R m i1
n h i
R m1X
flBRm 1 xi lARm 1 xi 1 lB xi Rm 1 1 lA xi Rm 1 gRm 1 1
1
R m i1
IRm A; B IRm B; A
IRa A; A [ B
n h R i
R X 1R R R a
lAa xi lA [aB xi 1 lA xi a 1 lA [ B xi 1 a R 1
R a i1
R X h Ra 1R R R a
i
lA xi lA a xi 1 lA xi a 1 lA xi 1 a R 1
R a X1
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1a R 1
R a X2
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1 a R 1
R a X2
2:48
Now
IRa A; A \ B
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1 a R 1
R a X1
R X h Ra 1R R R a
i
lA xi lA a xi 1 lA xi a 1 lA xi 1a R 1
R a X2
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1 a R 1
R a X1
2:49
2.6 Properties of New Generalized R-norm Fuzzy Divergence Measures 47
IRa A; A [ B IRa A; A \ B
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1a R 1
R a X1
R X h Ra 1R R R a
i
lA xi lB a xi 1 lA xi a 1 lB xi 1a R 1
R a X2
n h R i
R X 1R R R a
lAa xi lB a xi 1 lA xi a 1 lB xi 1a R 1
R a i1
IRa A; B
Hence, 2.8(a) holds for IRa A; B and similar result holds for IRm A; B.
2.8(b)
IRa B; A [ B
n h R i
R X 1R R R a
lBa xi lA [aB xi 1 lB xi a 1 lA [ B xi 1 a R 1
R a i1
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1 a R 1
R a X1
R X h Ra 1R R R a
i
lB xi lB a xi 1 lB xi a 1 lB xi 1a R 1
R a X2
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1 a R 1
R a X1
2:50
Now
IRa B; A \ B
n h R i
R X 1R R R a
lBa xi lA \aB xi 1 lB xi a 1 lA \ B xi 1 a R 1
R a i1
R X h Ra 1R R R a
i
lB xi lB a xi 1 lB xi a 1 lB xi 1 a R 1
R a X1
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1a R 1
R a X2
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1 a R 1
R a X2
2:51
48 2 Parametric Generalized R-norm Fuzzy Information
IRa B; A [ B IRa B; A \ B
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1a R 1
R a X1
R X h Ra 1R R R a
i
lB xi lA a xi 1 lB xi a 1 lA xi 1a R 1
R a X2
n h R i
R X 1R R R a
lBa xi lA a xi 1 lB xi a 1 lA xi 1a R 1
R a i1
IRa B; A
Hence, 2.8(b) holds for IRa A; B and similar result holds for IRm A; B.
Theorem 2.9 For fuzzy sets A and B
(a) IRa A; B IRa Ac ; Bc ; IRm A; B IRm Ac ; Bc .
(b) IRa Ac ; B IRa A; Bc ; IRm Ac ; B IRm A; Bc .
(c) IRa A; B IRa Ac ; B IRa Ac ; Bc IRa Ac ; B;
Proof 2.9(a)
IRa Ac ; Bc
n h R i
R X 1R R R a
lAa c xi lBc a xi 1 lAc xi a 1 lBc xi 1a R 1
R a i1
n h i
R X R R R
1R a
1 lA xi a 1 lB xi 1 a lAa xi lB a xi R 1
R a i1
IRa A; B
Now
IRm Ac ; Bc
2 3
1 1
R m1X n flARm
c xi lBRm
c xi
4 5
R m i1 1 lAc xi Rm 1 1 lBc xi Rm 1 gRm 1 1
1
2 3
R m1X n f1 lA xi Rm 1 1 lB xi Rm 1
4 5 I m A; B
R m i1 lARm 1 Rm 1
xi lB xi g
1
Rm 1 1
R
2.9(b)
IRa Ac ; B
n h R i
R X 1R R R a
lAa c xi lB a xi 1 lAc xi a 1 lB xi 1 a R 1
R a i1 2:52
X n h i
R R 1 R R R a
1 lA xi a lB a xi lAa xi 1 lB xi 1a R 1
R a i1
IRa A; Bc
n h R i
R X 1R R R a
lAa xi lBc a xi 1 lA xi a 1 lBc xi 1a R 1
R a i1 2:53
R X n h R R R 1R a
i
lAa xi 1 lB xi 1 a 1 lA xi a lB a xi R 1
R a i1
Theorem 2.10 For fuzzy sets A and B, the relation between HRa A and IRa A; B is
given by
" aR # aR
a 1 1 R
nR 1 R a
IR A; 1 HR A:
2 2 Ra 2
n h i
R X R R a
HRa A 1 lAa xi 1 lA xi a R
R a i1
n h R i 2:55
Rn R X R a
) HRa A lAa xi 1 lA xi a R
R a R a i1
2 Rm 1 3
Rm 1 1
n 6
fl xi 7
1 R m1X 6 A
6
2 7
7
m
IR A; 6 7
2 R m i1 4 1 Rm 1
5
1 lA xi Rm 1
1
gRm 1 1
2
2 3
X flA Rm 1
xi
R m1 1 n
4 5
Rm 2 i1 1 lA xi Rm 1 gRm 1
1
nR m 1
Rm
2:56
n h i
R m1X 1 Rm 1 Rm1 1
HRm A 1 lRm xi 1 l x i
R m i1 A A
) HRm A
n h i
nR m 1 R m 1 X 1
xi 1 lA xi Rm 1 Rm 1
1
lRm
Rm R m i1 A
2:57
In this chapter, we have proposed two new parametric generalizations of one of the
existing R-norm fuzzy information measures along with proof of their validity. It is
noted that the proposed generalized fuzzy measures of information are valid
measures which reduce to the known measure on substituting the particular values
of parameters. Some of the interesting properties of these measures have also been
studied. The similarity of proposed generalized measures of fuzzy information with
HR A is proved using a numerical example. In addition, two new parametric
generalized R-norm fuzzy divergence measures are introduced along with the proof
of their validity. The interesting properties of these proposed generalized R-norm
fuzzy divergence measures are also established. Finally, the relation between the
proposed generalized R-norm fuzzy entropy and divergence measures are intro-
duced. In view of application, the proposed measures are more flexible.
References
This chapter introduces and details a generalized methodology for measuring the
degree of difference between two fuzzy sets. We present a new parametric gener-
alized exponential measure of fuzzy divergence and study the essential properties of
this measure in order to check its authenticity.
Section 3.1 gives a discussion on some well-known concepts and notations
related to fuzzy set theory and fuzzy divergence measures. Thereafter, we introduce
a parametric generalized fuzzy exponential measure of divergence corresponding to
generalized fuzzy entropy given by Verma and Sharma [25] in Sect. 3.2. In
Sect. 3.3, we rst provide some interesting properties of the proposed measure of
fuzzy divergence and then a relation is established between generalized exponential
fuzzy entropy and the proposed fuzzy divergence measure. In Sect. 3.4, a com-
parison of the proposed divergence with some of existing generalized measures of
fuzzy divergence is presented with the help of table and graph. The application of
the proposed parametric generalized exponential measure of fuzzy divergence to
strategic decision-making is illustrated with the help of a numerical example in
Sect. 3.5. Section 3.6 presents the application of the proposed measure of fuzzy
divergence in the existing methods of strategic decision-making. In the same sec-
tion, a comparative analysis between the proposed method of strategic
decision-making and the existing methods is provided. The nal section concludes
the chapter.
This section is devoted to review some well-known notations and to recall the
axiomatic denition of a divergence measure for fuzzy sets.
As we have mentioned in Sect. 1.1.4 of Chap. 1 that Zadeh [26] gave some
notions related to fuzzy sets. Some of them, which we shall need in our discussion,
are as follows:
Springer International Publishing Switzerland 2016 53
A. Ohlan and R. Ohlan, Generalizations of Fuzzy Information Measures,
DOI 10.1007/978-3-319-45928-8_3
54 3 Parametric Generalized Exponential Fuzzy Divergence
n
X
lA xi 1 lA xi
Ia A : B lA xi log 1 lA xi log
i1
lB xi 1 lB xi
n
1X 1 alA xi 1 a1 lA xi
1 alA xi log f1 a1 lA xi g log
a i1 1 alB xi 1 a1 lB xi
3:4
1 X n h i
a
Da A; B log laA xi l1a x i 1 l x i 1 l x i 1a
;
a 1 i1 B A B
3:5
a 6 1; a [ 0:
@IEa a a
1 a1 lA xi a e1lB xi 1lA xi a 1elB xi lA xi
a a
@lA xi
@ 2 IE a h a a
i
aa 1 1 lA xi a1 e1lB xi 1lA xi la1
A xi e
laB xi laA xi
@lA xi
2
h a a a a
i
a2 1 lA xi 2a1 e1lB xi 1lA xi lA2a1 xi elB xi lA xi [ 0
for a [ 0:
@ 2 IEa
Similarly, @l2B xi
[ 0 for a [ 0.
56 3 Parametric Generalized Exponential Fuzzy Divergence
Thus IEa A; B is a convex function of fuzzy sets A and B and hence in view of
the denition of fuzzy divergence measure of Bhandari and Pal [2] provided in
Sect. 1.4 of Chap. 1, IEa A; B is a valid measure of fuzzy divergence and conse-
quently it establishes the validity of JEa A; B IEa A; B IEa B; A as a new
parametric generalized exponential symmetric measure of divergence between
fuzzy sets A and B.
In particular,
For a 1, IEa A; B reduces to IA; B given in (3.2).
The generalized exponential fuzzy divergence measure IEa A; B dened above has
the following properties. While proving these theorems we consider the separation
of X into two parts X1 and X2 , such that the set
X1 fx=x 2 X; lA xi lB xi g
and
X2 fx=x 2 X; lA xi \lB xi g:
In set X2 ,
Theorem 3.2
(a) IEa A [ B; A IEa A \ B; A IEa B; A.
(b) IEa A [ B; C IEa A \ B; C IEa A; C IEa B; C.
(c) IEa A [ B; A \ B IEa A \ B; A [ B.
3.3 Properties of Generalized Exponential Fuzzy Divergence Measure 57
Proof 3.2(a)
IEa A [ B; A
Xn h i
a a a a
1 1 lA [ B xi e1lA xi 1lA [ B xi lA [ B xi elA xi lA [ B xi
i1
Xh a
i
1lA xi a a a
1 1 lA xi e1lA xi lA xi elA xi lA xi
X1
Xh a a a a
i
1 1 lB xi e1lA xi 1lB xi lB xi elA xi lB xi
X2
Xh a
i
1lB xi a a a
1 1 lB xi e1lA xi lB xi elA xi lB xi
X2
3:7
Now
IEa A \ B; A
Xn h i
a a a a
1 1 lA \ B xi e1lA xi 1lA \ B xi lA \ B xi elA xi lA \ B xi
i1
Xh a
i
1lB xi a a a
1 1 lB xi e1lA xi lB xi elA xi lB xi
X1
Xh a a a a
i
1 1 lA xi e1lA xi 1lA xi lA xi elA xi lA xi
X2
Xh a
i
1lB xi a a a
1 1 lB xi e1lA xi lB xi elA xi lB xi
X1
3:8
IEa A [ B; A IEa A \ B; A
Xh a a a a
i
1 1 lB xi e1lA xi 1lB xi lB xi elA xi lB xi
X1
Xh a
i
1lB xi a a a
1 1 lB xi e1lA xi lB xi elA xi lB xi
X2
n h
X i
a
1lB xi a a a
1 1 lB xi e1lA xi lB xi elA xi lB xi
i1
IEa B; A
3.2(b) IEa A [ B; C
Xn h i
a a a a
1 1 lA [ B xi e1lC xi 1lA [ B xi lA [ B xi elC xi lA [ B xi
i1
Xh a
i
1lA xi a a a
1 1 lA xi e1lC xi lA xi elC xi lA xi
X1
Xh a
i
1lB xi a a a
1 1 lB xi e1lC xi lB xi elC xi lB xi
X2
3:9
IEa A \ B; C
Xn h i
a a a a
1 1 lA \ B xi e1lC xi 1lA \ B xi lA \ B xi elC xi lA \ B xi
i1
Xh a
i
1lB xi a a a
1 1 lB xi e1lC xi lB xi elC xi lB xi
X1
Xh a
i
1lA xi a a a
1 1 lA xi e1lC xi lA xi elC xi lA xi
X2
3:10
IEa A [ B; C IEa A \ B; C
Xn h i
a a a a
1 1 lA xi e1lC xi 1lA xi lA xi elC xi lA xi
i1
n h
X i
a a a a
1 1 lB xi e1lC xi 1lB xi lB xi elC xi lB xi
i1
IEa A; C IEa B; C:
and
IEa A \ B; A [ B
Xn h i
a a la x la x
1 1 lA \ B xi e1lA [ B xi 1lA \ B xi lA \ B xi e A [ B i A \ B i
i1
Xh a a la xi la xi
i
1 1 lA xi e1lB xi 1lA xi lA xi e B A
X1
Xh a a la xi la xi
i
1 1 lB xi e1lA xi 1lB xi lB xi e A B
X2
Xh a a a a
i
1 lA xi elB xi lA xi 1 lA xi e1lB xi 1lA xi
X1
Xh a
i
a a
1lB xi a
1 lB xi elA xi lB xi 1 lB xi e1lA xi
X2
IEa A [ B; A \ B
3.3(a) IEa A; A
Xn h i
a a la x la x
1 1 lA xi e1lA xi 1lA xi lA xi e A i A i
i1
n h
X i
a a a
laA xi
1 1 lA xi elA xi 1lA xi lA xi e1lA xi
i1
and
IEa A; A
X n h i
a a la x la x
1 1 lA xi e1lA xi 1lA xi lA xi e A i A i
i1
n h
X i
a a a a
1 lA xi e1lA xi lA xi 1 lA xi elA xi 1lA xi
i1
IEa A; A
60 3 Parametric Generalized Exponential Fuzzy Divergence
3.3(b) IEa A; B
X n h i
a a la x la x
1 1 lA xi e1lB xi 1lA xi lA xi e B i A i
i1
n h
X i
a
a a
1lA xi a
1 lA xi elB xi lA xi 1 lA xi e1lB xi
i1
IEa A; B
3.3(c) IEa A; B
X n h i
a a la x la x
1 1 lA xi e1lB xi 1lA xi lA xi e B i A i
i1
n h
X i
a a a
laA xi
1 1 lA xi elB xi 1lA xi lA xi e1lB xi
i1
Now IEa A; B
X n h i
a a la x la x
1 1 lA xi e1lB xi 1lA xi lA xi e B i A i
i1
n h
X i
a a a a
1 lA xi e1lB xi lA xi 1 lA xi elB xi 1lA xi
i1
IEa A; B
Proof
X n h i
1 a a a a
IEa A; 1 1 lA xi e0:5 1lA xi lA xi e0:5 lA xi
2 i1
" #
X n
e1lA xi
a a
elA xi
1 1 lA xi l x
A i
i1
e0:5a e0:5a
" #
X n
e11lA xi
a a
e1lA xi
1 1 lA xi lA xi 10:5a
i1
e10:5a e
1 X n h a a
i
n 10:5a 1 lA xi e11lA xi lA xi e1lA xi 1 1
e i1
n h i
1 X a a n
n 10:5a lA xi e1lA xi 1 lA xi e11lA xi 1 10:5a
e i1
e
1 a
n
n 10:5a n e10:5 1 Ea A 10:5a
e e
1 1 a
n 1 10:5a 10:5a n e10:5 1 Ea A
e e
a a
n e10:5 1 n e10:5 1
Ea A
e10:5a e10:5
a
10:5a
n e 1
1 Ea A
e10:5a
a
e10:5 1
) 10:5 a IEa A;
n e 1 2
1 Ea A
10:5a
Thus, Ea A 1 nee10:5a 1 IEa A; 12 .
Hence, 3.4 holds.
As we have already discussed in Sect. 3.1 that in recent years the applications of the
fuzzy divergence measure have been given in different areas: Poletti et al. [22] in
bioinformatics; Bhandari et al. [3], Fan et al. [10] and Bhatia and Singh [4] in image
thresholding; Ghosh et al. [13] in automated leukocyte recognition. We provide an
application of the proposed divergence measure in strategic decision-making.
Decision-making problem is the process of nding the best option from all of the
feasible alternatives. It is assumed that a rm X desires to apply m strategies
S1 ; S2 ; S3 ; . . .; Sm to meet its goal. Let each strategy has different degrees of effec-
tiveness if the input associated with it is varied, let fI1 ; I2 ; I3 ; . . .; In g be the input set
or set of alternatives. Let the fuzzy set Y denotes the effectiveness of a particular
strategy with uniform input. Therefore Y fY; lY Si =i 1; 2; . . .; mg. Further,
let Ij be a fuzzy set denotes the degree of effectiveness of a strategy when it is
implemented with input Ij .
3.5 Application of Parametric Generalized Exponential 63
.
Ij fIj ; lIj Si i 1; 2; . . .; mg
where j 1; 2; . . .; n.
Taking A Y and B Ij in the fuzzy divergence measure IEa Y; Ij dened in
Sect. 3.3 we calculate IEa Y; Ij . Then most effective Ij is determined by
MinfIEa Y; Ij g 1 j n . It is assumed that It 1 t n determines the minimum value
0\a 0:5
of fIEa Y; Ij g0\a 0:5 . With this It nds Max:flIj Si g1 i m , let it corresponds to
Sp , 1 p m. Hence, if the strategy Sp is implemented with input budget of It the
rm will meet its goal in the most input-effective manner.
An Illustrative Example
Let m n 5 in the above model. Table 3.2 shows the efciency of different
strategies at uniform inputs.
Table 3.3 illustrates the efciency of different strategies at particular inputs.
The numerical values of divergence measure fIEa Y; Ij g 1 j n are presented in
0\a 0:5
Table 3.4. The calculated numerical values of the proposed fuzzy divergence
measure indicate that alternative input I2 is more appropriate for values of
a0\a 0:5. An examination of the results presented in Tables 3.3 and 3.4 makes
it clear that the strategy S3 is most effective. Thus, a rm will achieve its goal most
effectively if the strategy S3 is applied with an input alternative I2 .
We now present an application of TOPSIS method developed by Hwang and Yoon [15]
and MOORA method developed by Brauers and Zavadskas [6] in a fuzzy context for
strategic decision-making using the proposed fuzzy divergence measure (3.6).
A methodology for solving decision-making problem using TOPSIS in fuzzy environ-
ment was developed by Chen [7], Jahanshahloo et al. [16] and Emami et al. [8]. Further,
Stanujkic et al. [24] and Gadakh et al. [12] extended the MOORA method for solving
decision-making problem with interval data and optimization problems in welding.
2. Construct the normalized fuzzy decision matrix. The normalized value nij is
calculated as
,v
uX
u m 2
nji xji t xji ; j 1; . . .; m; i 1; . . .; n: 3:11
j1
3. Construct the weighted normalized fuzzy decision matrix, the weighted nor-
malized value
vji wi nji ; j 1; . . .; m; i 1; . . .; n 3:12
3.6 Application of Fuzzy TOPSIS and MOORA Methods 65
A v
1 ; v2 ; . . .; vn min vij i2S respectively 3:14
j
Table 3.7 Distance of each alternative from positive and negative ideal solutions
IEa IEa
a 0:1 a 0:5 a 0:1 a 0:5
I1 0.1960 0.6621 0.1638 0.5308
I2 0.2367 0.8460 0.1268 0.4294
I3 0.0693 0.2505 0.2927 0.9497
I4 0.1522 0.5319 0.2020 0.6662
I5 0.1167 0.3974 0.2105 0.6866
Si v1 ; v2 ; . . .; vm max vij j2I and 3:16
i
S
i v
1 ; v2 ; . . .; vm min vij j2I respectively; 3:17
i
In this chapter, we have proposed and validated the generalized exponential mea-
sure of fuzzy divergence. A relation is established between generalized exponential
fuzzy entropy and the proposed fuzzy divergence measure. Particular case and some
of the properties are proved on this divergence measure. The efciency of the
proposed divergence measure is also presented. In addition, application of the
proposed divergence measure is discussed in strategic decision-making and a
numerical example is given for illustration. The application of the proposed mea-
sure of fuzzy divergence is presented in two existing methods of strategic
decision-making. A comparative analysis between the proposed method of strategic
decision-making and the existing methods of strategic decision-making has also
been provided. It is noted that the proposed measure of fuzzy divergence comprises
greater consistency and flexibility in applications because of the presence of the
parameter.
References
1. Bajaj RK, Hooda DS (2010) On some new generalized measures of fuzzy information. World
Acad Sci Eng Technol 62:747753
2. Bhandari D, Pal NR (1993) Some new information measures for fuzzy sets. Inf Sci 67
(3):209228
3. Bhandari D, Pal NR, Majumder DD (1992) Fuzzy divergence, probability measure of fuzzy
event and image thresholding. Inf Sci 13:857867
References 69
4. Bhatia PK, Singh S (2013) A new measure of fuzzy directed divergence and its application in
image segmentation. Int J Intell Syst Appl 4:8189
5. Bouchon-Meunier B, Rifqi M, Bothorel S (1996) Towards general measures of comparison of
objects. Fuzzy Sets Syst 84:143153
6. Brauers WKM, Zavadskas EK (2006) The MOORA method and its application to
privatization in transition economy. Control Cybern 35(2):443468
7. Chen CT (2000) Extension of TOPSIS for group decision making under fuzzy environment.
Fuzzy Sets Syst 114(1):19
8. Emami M, Nazari K, Fardmanesh H (2012) Application of fuzzy TOPSIS technique for
strategic management decision. J Appl Basic Sci Res 2(1):685689
9. Fan J, Xie W (1999) Distance measures and induced fuzzy entropy. Fuzzy Sets Syst 104
(2):305314
10. Fan S, Yang S, He P, Nie H (2011) Infrared electric image thresholding using two
dimensional fuzzy entropy. Energy Procedia 12:411419
11. Ferreri C (1980) Hyperentropy and related heterogeneity divergence and information
measures. Statistica 40(2):155168
12. Gadakh VS, Shinde VB, Khemnar NS (2013) Optimization of welding process parameters
using MOORA method. Int J Adv Manuf Technol 69:20312039
13. Ghosh M, Das D, Chakraborty C, Ray AK (2010) Automated leukocyte recognition using
fuzzy divergence. Micron 41:840846
14. Havrada JH, Charvat F (1967) Quantication methods of classication processes: concept of
structural a-entropy. Kybernetika 3(1):3035
15. Hwang CL, Yoon K (1981) Multiple attribute decision makingmethods and applications.
Springer, New York
16. Jahanshahloo GR, Hosseinzadeh LF, Izadikhah M (2006) Extension of TOPSIS method for
decision making problems with fuzzy data. Appl Math Comput 181:15441551
17. Kapur JN (1997) Measures of fuzzy information. Mathematical Sciences Trust Society, New
Delhi
18. Kullback S, Leibler RA (1951) On information and sufciency. Ann Math Stat 22(1):7986
19. Montes S, Couso I, Gil P, Bertoluzza C (2002) Divergence measure between fuzzy sets. Int J
Approximate Reasoning 30:91105
20. Pal NR, Pal SK (1989) Object background segmentation using new denition of entropy. IEE
Proc Comput Digital Tech 136(4):248295
21. Parkash O, Sharma PK, Kumar S (2006) Two new measures of fuzzy divergence and their
properties. SQU J Sci 11:6977
22. Poletti E, Zappelli F, Ruggeri A, Grisan E (2012) A review of thresholding strategies applied
to human chromosome segmentation. Comput Methods Programs Biomed 108:679688
23. Renyi A (1961) On measures of entropy and information. In: Proceeding of fourth Berkeley
symposium on mathematics, statistics and probability, vol 1, pp 547561
24. Stanujkic D, Magdalinovic N, Tojanovic S, Jovanovic R (2012) Extension of ratio system part
of MOORA method for solving decision making problems with interval data. Informatica 23
(1):141154
25. Verma R, Sharma BD (2011) On generalized exponential fuzzy entropy. World Acad Sci Eng
Technol 69:14021405
26. Zadeh LA (1965) Fuzzy sets. Inf Control 8:338353
Chapter 4
Sequence and Application of Inequalities
Among Fuzzy Mean Difference Divergence
Measures in Pattern Recognition
Fuzzy divergence measures and inequalities have recently been applied widely in
the fuzzy comprehensive evaluation and information theory. In this chapter, we
introduce a sequence of fuzzy mean difference divergence measures and establish
the inequalities among them to explore the fuzzy inequalities. The advantage of
establishing the inequalities is to make the computational work much simpler. The
technique of inequalities provides a better comparison among fuzzy mean diver-
gence measures.
In Sect. 4.1 we present preliminaries on fuzzy mean divergence measures and
important results related to fuzzy measure that we use in our study. A sequence of
fuzzy mean difference divergence measures and the essential properties for their
validity are provided in Sect. 4.2. Section 4.3 provides some of inequalities among
selected proposed fuzzy divergence measures. In the same section the proof of these
inequalities is also presented. In Sect. 4.4 the applications of proposed fuzzy mean
difference divergence measures in the context of pattern recognition have been
presented using a numerical example. In the next section, it is shown that the
proposed fuzzy mean difference divergence measures are well suited to use with
linguistic variables. Finally, concluding remarks are drawn in Sect. 4.6.
The fuzzy mean divergence measures of Singh and Tomar [1] corresponding to
seven geometrical mean measures given in Taneja [2] are presented in Table 4.1.
We have the following Lemma in fuzzy context corresponding to the Lemma of
Taneja [3]:
Lemma
1 4.1 Let f : I R ! R be a convex and differentiable function satisfying
f 2 0. Consider a function
5. Fuzzy n 2
P
lA xi l2B xi 1lA xi 2 1lB xi 2
CA; B lA xi lB xi 2lA xi lB xi
contra-harmonic i1
mean measure
q q
6. Fuzzy P
n 2 lA xi lB xi
2 2
1lA xi 1lB xi 2
SA; B 2 2
root-mean-square i1
mean measure
0 1
7. Fuzzy centroidal 2l2A xi lA xi lB xi l2B xi
B
n B 3lA xi lB xi C
mean measure
RA; B
P
B C
C
B 2 1 lA xi 2 1 lA xi 1 lB xi 1 lB xi 2 C
i1 @ A
32 lA xi lB xi
b
uf a; b af ; a; b [ 0;
a
Then the function uf a; b is convex R2 . Additionally, if f 0 1=2 0, then the
following inequality holds:
ba
0 uf a; b uf 0 a; b:
a
b
uf a; b af ; a; b [ 0:
a
>
i1 >
l2A xi l2B xi
1lA xi 2 1lB xi >
2 2
>
: 2 ;
4:1
4:4
74 4 Sequence and Application of Inequalities Among Fuzzy
4:5
vi DCH A; B CA; B HA; B
8h 2 i 9
n <> lA xi l2B xi 1lA xi 2 1lB xi 2 >
=
X lA xi lB xi 2lA xi lB xi
h i
> 2lA xi lB xi 21lA xi 1lB xi >
i1 : ;
l xi l xi A 2l xi l xi
B A B
4:6
vii DSA A; B SA; B AA; B
8 q q
9
Xn >
<
l2A xi l2B xi
1lA xi 2 1lB xi 2 >
=
2 2
h i
>
i1 :
>
;
lA xi 2 lB xi 2lA xi2lB xi
4:7
viii DSN A; B SA; B NA; B
8 q q
9
>
> l2A xi l2B xi >
1lA xi 2 1lB xi >
2 2
>
> >
>
Xn >
<
2 >
=
2 p 3
lA xi lA xi lB xi lB xi
>
> 4 >
>
i1 >
>
>
3
p 5 >
>
: 2lA xi lB xi 1lA xi 1lB xi > ;
3
4:8
4:10
4.2 Sequence of Fuzzy Mean Difference Divergence Measures 75
4:11
4:14
4:16
Theorem 4.1 All the proposed measures from Eqs. (4.1) to (4.18) are valid
measures of fuzzy mean difference divergence.
Proof
(a) Nonnegativity
From one of inequalities given in Taneja [2], for two fuzzy sets A and B we
have HA; B GA; B NA; B AA; B RA; B SA; B CA; B:
Hence, the condition of nonnegativity of measures from Eqs. (4.1) to (4.18) is
proved.
(b) Also clearly, DA1 B1 A; A 0 for all measures from Eqs. (4.1) to (4.18) where
A1 and B1 belong to the fuzzy mean divergence measures given in Table 4.1.
(c) Convexity
We now shall prove the condition of convexity of measures for Eqs. (4.1)
(4.18) with the help of Lemma 4.1.
For simplicity, let us write DA1 B1 bfA1 B1 where fA1 B1 z fA1 z fB1 z with
A1 B1 .
Let us take lA z ) lB 1 z. So, corresponding to measures of Eqs. (4.1)
(4.18) we have the following generating functions:
2 s3
z2 1 z2 5
1: fCS z 24z2 1 z2 4:19
2
4.2 Sequence of Fuzzy Mean Difference Divergence Measures 77
" p#
2 1 z1 z
2: fCN z 2 z 1 z
2
4:20
3
h pi
3: fCG z 2 z2 1 z2 z1 z 4:21
" #
2z2 1 z2 z1 z
2
4: fCR z 2 z 1 z
2
4:22
3
z 1 z
5: fCA z 2 z2 1 z2 4:23
2
h i
6: fCH z 2 z2 1 z2 2z1 z 4:24
2s 3
2
z 1 z
2 z 1 z5
7: fSA z 24 4:25
2 2
2s p 3
z 1 z
2 2
z z1 z 1 z5
8: fSN z 24 4:26
2 3
2s 3
z 1 z
2 2 p
9: fSG z 24 z1 z5 4:27
2
2s 3
2
z 1 z
2
10: fSH z 24 2z1 z5 4:28
2
" #
2z2 1 z2 z1 z z 1 z
11: fRA z 2 4:29
3 2
" p #
2z2 1 z2 z1 z 1 z1 z
12: fRN z 2 4:30
3 3
" #
2z2 1 z2 z1 z p
13: fRG z 2 z1 z 4:31
3
78 4 Sequence and Application of Inequalities Among Fuzzy
" #
2z2 1 z2 z1 z
14: fRH z 2 2z1 z 4:32
3
" p#
z 1 z 1 z1 z
15: fAN z 2 4:33
2 3
z 1 z p
16: fAG z 2 z1 z 4:34
2
z 1 z
17: fAH z 2 2z1 z 4:35
2
" p #
1 z1 z p
18: fNG z 2 z1 z 4:36
3
Now
in all
the cases
from Eqs. (4.19) to (4.36), we can easily check that
fA1 B1 12 fA1 12 fB1 12 12 12 0. It is understood that z 2 0; 1.
The rst- and second-order derivatives of the functions Eqs. (4.19)(4.36) are as
follows:
0 22z 1
fCS z 42z 1 q ;
1: 2z2 1 z2 4:37
00 4
fCS z 8 3=2 [ 0
2z2 1 z2
0 2z 1
fCN z 42z 1 p ;
2: 3 z z2 4:38
00 4
fCN z 8 [0
6z z2 3=2
0 2z 1
fCG z 42z 1 p ;
3: z z2 4:39
00 4
fCG z 8 [0
2z z2 3=2
4: 0 82z 1 00 16 4:40
fCR z ; fCR z [0
3 3
4.2 Sequence of Fuzzy Mean Difference Divergence Measures 79
0
00
5: fCA z 42z 1; fCA z 8 [ 0 4:41
0
00
6: fCH z 82z 1; fCH z 16 [ 0 4:42
p
0 22z 1 00 2
7: fSA z q ; fSA z 3=2 [ 0 4:43
2z2 1 z2 z2 1 z2
0 22z 1 2z 1
fSN z q p ;
2z2 1 z2 3 z z
2
8: p 4:44
00 2 4
fSN z 3=2 [0
z2 1 z2 6z z2 3=2
0 22z 1 2z 1
fSG z q p ;
9: 2z2 1 z 2 z z2 4:45
00 4 1
fSG z 3=2 [0
2z2 1 z2 2z z2 3=2
0 42z 1
fSH z q 42z 1;
10: 2z2 1 z2 4:46
p
00 2
fSH z 8 3=2 [ 0
z2 1 z2
0 42z 1 00 8
11: fRA z ; fRA z [0 4:47
3 3
0 42z 1 2z 1
fRN z p ;
3 3 z z2
12: " # 4:48
00 8 1 z2 1 z2
fRN z [0
3 3 z z2 3=2
0 42z 1 2z 1
fRG z p ;
3 z z2
13: 4:49
00 4 1
fRG z [0
3 2z z2 3=2
80 4 Sequence and Application of Inequalities Among Fuzzy
0 162z 1 00 32
14: fRH z ; fRH z [0 4:50
3 3
0 2z 1 00 1
15: fAN z p ; fAN z [0 4:51
3 z z2 6z z2 3=2
0 2z 1 00 1
16: fAG z p ; fAG z [0 4:52
z z2 z z2 3=2
0
00
17: fAH z 42z 1; fAH z 8 [ 0 4:53
0 22z 1 00 2z2 1 z2
18: fNG z p ; fNG z [0 4:54
3 z z2 3z z2 3=2
We see that in the all cases second-order derivative is positive and satises
0
fA1 B1 12 0 for all z 2 0; 1. Thus, according to Lemma 4.1 and Eqs. (4.39)
(4.54), we get the convexity of the measures for Eqs. (4.1)(4.18).
Hence in view of the denition offuzzy divergence measure of Bhandari and Pal [4],
all the dened measures for Eqs. (4.1)(4.18) are valid measures of fuzzy mean dif-
ference divergence. Moreover, we can easily check for measures in Eqs. (4.1)(4.18)
such that DA1 B1 A; B DA1 B1 B; A where A1 and B1 belong to the fuzzy mean
divergence measures given in Table 4.1. Hence all the dened measures for
Eqs. (4.1)(4.18) are valid measures of fuzzy symmetric mean difference divergence.
1 3 3 1 3
iii DSA DSH DCR DCN DCG DRG 3DAN ;
3 4 7 3 5
3 1 3
iv DSA DSN DSG DRG 3DAN :
4 2 5
Proof The proof of the above theorem is based on Lemma 4.2 and is given in parts
in the following propositions.
Proposition 4.1 We have DSA 34 DSN
Proof Let us consider the function
00
p
fSA z 6 2z z2 3=2
gSA SN z 00 z p 3=2
fSN
6 2z z2 3=2 z2 1 z2
This gives
p
0 3 22z 1z z2 1=2 2z2 2z 11=2 4z2 4z 1 [ 0 for z\1=2
gSA SN z h p i2 \ 0 for z [ 1=2
6 2z z2 3=2 2z2 2z 13=2
We have
1 3
b sup gSA SN z gSA SN 4:55
z 20;1 2 4
Applying Lemma 4.2 for the difference of fuzzy means DSA A; B and DSN A; B
and using Eq. (4.55) we get
3
DSA DSN :
4
Proposition 4.2 We have DSA 13 DSH
Proof Let us consider the function
00
p
fSA z 2
gSA SH z 00 p
fSH z 82z2 2z 13=2 2
This gives
p 1=2
0 12 22z2 2z 1 4z 2 [ 0 for z\1=2
gSA SH z h pi2 \ 0 for z [ 1=2
82z2 2z 13=2 2
82 4 Sequence and Application of Inequalities Among Fuzzy
We have
1 1
b sup gSA SH z gSA SH 4:56
z 20;1 2 3
Applying Lemma 4.2 for the difference of fuzzy means DSA A; B and DSH A; B
and using Eq. (4.56) we get
1
DSA DSH :
3
Proposition 4.3 We have DSH 94 DCR
Proof Let us consider the function
00 3=2 p
fSH z 242z2 2z 1 3 2
gSH CR z 00
fCR z 162z2 2z 13=2
This gives
p
0 46 22z 1 [ 0 for z\1=2
gSH CR z 5=2 \ 0 for z [ 1=2
492z2 2z 1
And we have
1 9
b sup gSH CR z gSH CR 4:57
z 20;1 2 4
Applying Lemma 4.2 for the difference of fuzzy means DSH A; B and DCR A; B
and using Eq. (4.57) we get
9
DSH DCR :
4
Proposition 4.4 We have DCR 47 DCN
Proof Let us consider the function
00 3=2
fCR z 32z z2
gCR CN z 00 z
fCN 48z z2 3=2 1
4.3 Inequalities Among Fuzzy Mean Difference Divergence Measures 83
This gives
0 48z z2 1=2 1 2z [ 0 for z\1=2
gCR z h i2
CN \ 0 for z [ 1=2
48z z2 3=2 1
We have
1 4
b sup gCR CN z gCR CN 4:58
z 20;1 2 7
Applying Lemma 4.2 for the difference of fuzzy means DCR A; B and
DCN A; B and using Eq. (4.58) we get
4
DCR DCN :
7
00
fCR z 164z2 4z 23=2 z z2 3=2
gCR CN z 00 z
fSG 24z z2 3=2 34z2 4z 23=2
This gives
h i
0
82z 1z z2 1=2 4z2 4z 21=2 32z z2 5=2 4z2 4z 25=2
gCR SG z h i2
8z z2 3=2 4z2 4z 23=2
[ 0 for z\1=2
\ 0 for z [ 1=2
We have
1 2
b sup gCR SG z gCR SG 4:59
z 20;1 2 3
Applying Lemma 4.2 for the difference of fuzzy means DCR A; B and DSG A; B
and using Eq. (4.59) we get
2
DCR DSG :
3
84 4 Sequence and Application of Inequalities Among Fuzzy
00
fSN z 24z z2 3=2 4z2 4z 23=2
gSN CN z 00 z
fCN 48z z2 3=2 14z2 4z 23=2
This gives
h i
0
72z z2 1=2 4z2 4z 21=2 1 2z 1 96z z2 5=2 4z2 4z 25=2
gSN CN z 2
48z z2 3=2 1 4z2 4z 23
[ 0 for z\1=2
\ 0 for z [ 1=2
We have
1 4
b sup gSN CN z gSN CN 4:60
z 20;1 2 7
Applying Lemma 4.2 for the difference of fuzzy means DSN A; B and DCN A; B
and using Eq. (4.60) we get
4
DSN DCN :
7
Proposition 4.7 We have DSN 23 DSG
Proof Let us consider the function
00
fSN z 24z z2 3=2 4z2 4z 23=2
gSN SG z 00 z
fSG 24z z2 3=2 34z2 4z 23=2
This gives
0 144z z2 1=2 4z2 4z 21=2 1 2z [ 0 for z\1=2
gSN SG z h i2 \0 for z [ 1=2
24z z2 3=2 34z2 4z 23=2
We have
1 2
b sup gSN SG z gSN SG : 4:61
z 20;1 2 3
4.3 Inequalities Among Fuzzy Mean Difference Divergence Measures 85
Applying Lemma 4.2 for the difference of fuzzy means DSN A; B and DSG A; B
and using Eq. (4.61) we get
2
DSN DSG :
3
Proposition 4.8 We have DCN 73 DCS
Proof Let us consider the function
" #" #1
00
fCN z 1 4
gCN CS z 00 8 8
fCS z 6z z2 3=2 4z2 4z 23=2
This gives
0 2z 14z2 4z 21=2 f4z2 4z 284z2 4z 23=2 4 16z z2 48z z2 3=2 1g
gCN CS z
4z z2 5=2 84z2 4z 23=2 42
[ 0 for z\1=2
\ 0 for z [ 1=2
We have
1 7
b sup gCN CS z gCN CS 4:62
z 20;1 2 3
Applying Lemma 4.2 for the difference of fuzzy means DCN A; B and DCS A; B
and using Eq. (4.62) we get
7
DCN DCS :
3
Proposition 4.9 We have DCS 3DAN
Proof Let us consider the function
" #
00
fCS z 84z2 4z 23=2 4 h i
2 3=2
gCS AN z 00 6 z z
fAN z 4z2 4z 23=2
This gives
n h io
362z 1z z2 4z z2 1=2 4z2 4z 2 24z2 4z 23=2 1
0 [ 0 for z\1=2
gCS AN z 5=2 \ 0 for z [ 1=2
4z2 4z 2
86 4 Sequence and Application of Inequalities Among Fuzzy
We have
1
b sup gCS AN z gCS AN 3 4:63
z 20;1 2
Applying Lemma 4.2 for the difference of fuzzy means DCS A; B and DAN A; B
and using Eq. (4.63) we get
DCS 3DAN :
This gives
0 161 2zz z2 1=2 [ 0 for z\1=2
gCN z h i2
CG \ 0 for z [ 1=2
16z z2 3=2 1
We have
1 7
b sup gCN CG z gCN CG 4:64
z 20;1 2 9
Applying Lemma 4.2 for the difference of fuzzy means DCN A; B and
DCG A; B and using Eq. (4.64) we get
7
DCN DCG :
9
Proposition 4.11 We have DSG 65 DRG
Proof Let us consider the function
00
fSG z 24z z2 3=2 34z2 4z 23=2
gSG RG z 00 z
fRG 4z2 4z 23=2 8z z2 3=2 3
4.3 Inequalities Among Fuzzy Mean Difference Divergence Measures 87
This gives
8 9
< 4z2 4z 25=2 32z z2 5=2 8z z2 3=2 3 =
92z 1
: 38z z2 3=2 4z2 4z 23=2 4z2 4z 2 ;
0
gSG RG z h i2
24z2 4z 25=2 z z2 8z z2 3=2 3
0 [ 0 for z\1=2
gSG RG z
\ 0 for z [ 1=2
We have
1 6
b sup gSG RG z gSG RG 4:65
z 20;1 2 5
Applying Lemma 4.2 for the difference of fuzzy means DSG A; B and DRG A; B
and using Eq. (4.65) we get
6
DSG DRG :
5
Proposition 4.12 We have DCG 95 DRG
Proof Let us consider the function
00
fCG z
gCG RG z 2 3=2
00 z 6 458z z 31
fRG
This gives
0 540z z2 1=2 1 2z [ 0 for z\1=2
gCG RG z
\0 for z [ 1=2
8z z2 3=2 3
We have
1 9
b sup gCG RG z gCG RG 4:66
z 20;1 2 5
Applying Lemma 4.2 for the difference of fuzzy means DCG A; B and
DRG A; B and using Eq. (4.66) we get
9
DCG DRG :
5
88 4 Sequence and Application of Inequalities Among Fuzzy
This gives
[ 0 for z\1=2
g0RG 2 3=2
AN z 8z z 3
\0 for z [ 1=2
We have
1
b sup gRG AN z gRG AN 5 4:67
z 20;1 2
Applying Lemma 4.2 for the difference of fuzzy means DRG A; B and
DAN A; B and using Eq. (4.67) we get
DRG 5DAN :
We now present the application of the proposed fuzzy mean difference divergence
measures in the context of pattern recognition. Next, an example related to pattern
recognition is given to demonstrate the results obtained for the fuzzy mean dif-
ference divergence measures for Eqs. (4.1)(4.18).
In order to demonstrate the application of the introduced fuzzy mean difference
divergence measures to pattern recognition, suppose that we are given three known
patterns P1 ; P2 and P3 which have classications C1 ; C2 and C3 respectively.
These patterns are represented by the following fuzzy sets in the universe of dis-
course X fx1 ; x2 ; x3 ; x4 g:
For convenience we use the notation i in Table 4.2 to present the divergence
value computed from equation i.
Our aim here is to classify Q to one of the classes C1 ; C2 and C3 . According to
the principle of minimum divergence information between fuzzy sets, the process of
assigning Q to Ck is described by k arg minfDAB Pk ; Qg.
k
Table 4.2 (above) presents DAB Pk ; Q, k f1; 2; 3g. It is observed that Q has
been classied to C3 correctly.
Let us establish that the proposed fuzzy mean difference divergence measures
Eqs. (4.1)(4.18) are reliable in applications with compound linguistic variables.
Example Let F fx; lF x=x 2 Xg be a fuzzy set in X. Tomar and Ohlan [5, 6]
dened for any positive real number n, from the operation of power of a fuzzy set
90 4 Sequence and Application of Inequalities Among Fuzzy
It can be seen from the numerical results presented in Table 4.3 that the pro-
posed fuzzy mean difference divergence measures for Eqs. (4.1)(4.18) satisfy the
requirement of Eqs. (4.68)(4.71). Thus, the proposed fuzzy mean difference
divergence measures are consistent in the application with compound linguistic
measures.
Table 4.3 Divergence values calculated by Eqs. (4.1)(4.18)
L. M.L.L. V.L. V.V.L. L. M.L.L. V.L. V.V.L. L. M.L.L. V.L. V.V.L. L. M.L.L. V.L. V.V.L.
L. M.L.L. V.L. V.V.L.
0.0000(6) 0.0884(6) 0.1304(6) 0.4250(6) 0.0884(6) 0.0000(6) 0.4523(6) 0.8566(6) 0.1304(6) 0.4523(6) 0.0000(6) 0.1114(6) 0.4250(6) 0.8566(6) 0.1114(6) 0.0000(6)
4.5 Numerical Example
0.0000(7) 0.2249(7) 0.2945(7) 0.9380(7) 0.2249(7) 0.0000(7) 1.0043(7) 1.9197(7) 0.2945(7) 1.0043(7) 0.0000(7) 0.2485(7) 0.9380(7) 1.9197(7) 0.2485(7) 0.0000(7)
(8) (8) (8) (8) (8) (8) (8) (8) (8) (8) (8) (8) (8) (8) (8)
0.0000 0.3013 0.3836 1.2589 0.3013 0.0000 1.3277 2.5594 0.3836 1.3277 0.0000 0.3262 1.2589 2.5594 0.3262 0.0000(8)
(9) (9) (9) (9) (9) (9) (9) (9) (9) (9) (9) (9) (9) (9) (9)
0.0000 0.1124 0.1666 0.5183 0.1124 0.0000 0.5617 0.6388 0.1666 0.5617 0.0000 0.1399 0.5183 0.6388 0.1399 0.0000(9)
(10) (10) (10) (10) (10) (10) (10) (10) (10) (10) (10) (10) (10) (10) (10)
0.0000 0.1866 0.2499 0.7775 0.1866 0.0000 0.8426 1.5441 0.2499 0.8426 0.0000 0.2097 0.7775 1.5441 0.2097 0.0000(10)
0.0000(11) 0.4092(11) 0.4998(11) 1.5550(11) 0.4092(11) 0.0000(11) 1.6852(11) 3.0882(11) 0.4998(11) 1.6852(11) 0.0000(11) 0.4194(11) 1.5550(11) 3.0882(11) 0.4194(11) 0.0000(11)
0.0000(12) 0.0982(12) 0.1195(12) 0.3525(12) 0.0982(12) 0.0000(12) 0.3903(12) 0.6875(12) 0.1195(12) 0.3903(12) 0.0000(12) 0.0983(12) 0.3525(12) 0.6875(12) 0.0983(12) 0.0000(12)
0.0000(13) 0.1365(13) 0.1641(13) 0.5130(13) 0.1365(13) 0.0000(13) 0.5520(13) 1.0631(13) 0.1641(13) 0.5520(13) 0.0000(13) 0.1371(13) 0.5130(13) 1.0631(13) 0.1371(13) 0.0000(13)
(14) (14) (14) (14) (14) (14) (14) (14) (14) (14) (14) (14) (14) (14) (14)
0.0000 0.2129 0.2532 0.8339 0.2129 0.0000 0.8754 1.7028 0.2532 0.8754 0.0000 0.2148 0.8339 1.7028 0.2148 0.0000(14)
(15) (15) (15) (15) (15) (15) (15) (15) (15) (15) (15) (15) (15) (15) (15)
0.0000 0.3208 0.3694 1.1300 0.3208 0.0000 1.2329 2.2316 0.3694 1.2329 0.0000 0.3080 1.1300 2.2316 0.3080 0.0000(15)
(16) (16) (16) (16) (16) (16) (16) (16) (16) (16) (16) (16) (16) (16) (16)
0.0000 0.0742 0.0833 0.2592 0.0742 0.0000 0.2809 0.9053 0.0833 0.2809 0.0000 0.0698 0.2592 0.9053 0.0698 0.0000(16)
0.0000(17) 0.1125(17) 0.1279(17) 0.4197(17) 0.1125(17) 0.0000(17) 0.4426(17) 1.2809(17) 0.1279(17) 0.4426(17) 0.0000(17) 0.1086(17) 0.4197(17) 1.2809(17) 0.1086(17) 0.0000(17)
0.0000(18) 0.1889(18) 0.2170(18) 0.7406(18) 0.1889(18) 0.0000(18) 0.7660(18) 1.9206(18) 0.2170(18) 0.7660(18) 0.0000(18) 0.1863(18) 0.7406(18) 1.9206(18) 0.1863(18) 0.0000(18)
0.0000(19) 0.2968(19) 0.3332(19) 1.0367(19) 0.2968(19) 0.0000(19) 1.1235(19) 2.4494(19) 0.3332(19) 1.1235(19) 0.0000(19) 0.2795(19) 1.0367(19) 2.4494(19) 0.2795(19) 0.0000(19)
(20) (20) (20) (20) (20) (20) (20) (20) (20) (20) (20) (20) (20) (20) (20)
0.0000 0.0383 0.0446 0.1605 0.0383 0.0000 0.1617 0.3756 0.0446 0.1617 0.0000 0.0388 0.1605 0.3756 0.0388 0.0000(20)
(21) (21) (21) (21) (21) (21) (21) (21) (21) (21) (21) (21) (21) (21) (21)
0.0000 0.1147 0.1337 0.4814 0.1147 0.0000 0.4851 1.0153 0.1337 0.4851 0.0000 0.1165 0.4814 1.0153 0.1165 0.0000(21)
(22) (22) (22) (22) (22) (22) (22) (22) (22) (22) (22) (22) (22) (22) (22)
0.0000 0.2226 0.2499 0.7775 0.2226 0.0000 0.8426 1.5441 0.2499 0.8426 0.0000 0.2097 0.7775 1.5441 0.2097 0.0000(22)
0.0000(23) 0.0764(23) 0.0891(23) 0.3209(23) 0.0764(23) 0.0000(23) 0.3234(23) 0.6397(23) 0.0891(23) 0.3234(23) 0.0000(23) 0.0777(23) 0.3209(23) 0.6397(23) 0.0777(23) 0.0000(23)
For convenience we use the notation i in Table 4.3 to present the divergence value computed from equation i.
91
92 4 Sequence and Application of Inequalities Among Fuzzy
References
1. Singh RP, Tomar VP (2014) On fuzzy mean divergence measures and their inequalities. In: 5th
national conference at MIT, Academy of Engineering, Alandi(D), Pune, Maharashtra
2. Taneja IJ (2012) Inequalities having seven means and proportionality relations. http://arxiv.org/
pdf/1203.2288v1.pdf. Accessed 18 April 2015
3. Taneja IJ (2005) Renement of inequalities among means. http://arxiv.org/pdf/math/
0505192v2.pdf. Accessed 20 April 2015
4. Bhandari D, Pal NR (1993) Some new information measures for fuzzy sets. Inf Sci 67
(3):209228
5. Tomar VP, Ohlan A (2014a) Two new parametric generalized Rnorm fuzzy information
measures. Int J Comput Appl 93(13):2227
6. Tomar VP, Ohlan A (2014b) Sequence of inequalities among fuzzy mean difference divergence
measures and their applications. SpringerPlus 3 (623):120
Chapter 5
Applications of Generalized Fuzzy
Divergence Measure in Multi-criteria
Decision-Making and Pattern Recognition
Fuzzy divergence measures have broad applications in many areas such as pattern
recognition, fuzzy clustering, decision-making, signal and image processing,
speech recognition, bioinformatics, fuzzy aircraft control, feature selection, etc.
This chapter provides a new parametric generalized measure of fuzzy divergence
along with its applications in the context of pattern recognition and multi-criteria
decision-making. In addition, the efciency of the proposed measure is presented by
proving its elegant properties.
In Sect. 5.1, we propose a new parametric measure of fuzzy divergence corre-
sponding to Taneja [1] divergence measure. Following this, particular cases are
discussed in the same section. Some more elegant properties of the proposed
measure are studied in a number of theorems in Sect. 5.2. The applications of the
proposed generalized divergence measure to pattern recognition and multi-criteria
decision-making with numerical examples are then discussed in Sect. 5.3. Finally,
the concluding remarks are drawn in Sect. 5.4. It is noted that the presence of the
parameter in the proposed divergence measure provides greater flexibility for pat-
tern recognition and multi-criteria decision-making.
Xn
pi qi 2 pi qi t
Lt P; Q p ; P; Q 2 Cn Cn ; t 2 Z
i1 2t pi qi t 1
( , ) 5:1
Xn
where Cn P p1 ; p2 ; . . .; pn pi [ 0; pi 1 ; n 2:
i1
d1 : DA; B DB; A
d2 : DA; A 0
d3 : maxfDA [ C; B [ C; DA \ C; B \ Cg DA; B
X1 fx=x 2 X; lA x lB x lC xg;
X2 fx=x 2 X; lA x lC x\lB xg;
X3 fx=x 2 X; lB x\lA x lC xg;
X4 fx=x 2 X; lB x lC x\lA xg;
X5 fx=x 2 X; lC x\lA x lB xg;
X6 fx=x 2 X; lC x\lB x\lA xg:
5.1 A New Parametric Generalized Measure of Fuzzy Divergence 95
Lt A [ C; B [ C Lt C; C 0;
Lt A \ C; B \ C Lt A; B;
So, maxfLt A [ C; B [ C; Lt A \ C; B \ Cg Lt A; B.
Similarly, in the sets X2 , X3 , X4 , X5 , X6 we have maxfLt A [ C; B [ C;
Lt A \ C; B \ Cg Lt A; B:
Thus, maxfLt A [ C; B [ C; Lt A \ C; B \ Cg Lt A; B for all A; B; C 2 FX.
(iv) We now show that Lt A; B [ 0.
Consider two standard fuzzy sets A 0:3; 0:4; 0:2; 0:1; 0:5, B 0:2; 0:2;
0:3; 0:4; 0:4. "
Pn
lA xi lB xi 2 lA xi lB xi t
We take Lt A; B ei , where ei 2t
p t 1
i1 lA xi lB xi
#
t
2lA xi lB xi
p t 1
1lA xi 1lB xi
for i 1; 2; 3; . . .n:
Table 5.1 gives positive values of Lt A; B for any value of t 0; 1; 2; . . .: with
two standard fuzzy sets considered above. It is clear from the results presented in
Table 5.1 that for arbitrary values of t 0, Lt A; B [ 0.
Hence in view of the denition of fuzzy divergence measure of Couso et al. [3],
Lt A; B is a valid measure of fuzzy divergence.
In particular,
(i) For t 1, Lt A; B reduces to 2DA; B, where DA; B is the fuzzy tri-
angular divergence measure of Singh and Tomar [4].
(ii) For t 1, Lt A; B reduces to 12 wA; B, where wA; B is the fuzzy
Chi-square divergence measure of Singh and Tomar [4].
(iii) For t 2, Lt A; B reduces to 12 wM A; B, where wM A; B is the symmetric
chi-square, arithmetic and geometric mean divergence measure of Kumar
and Johnson [5] in fuzzy environment.
In this section we provide some more properties of the proposed generalized fuzzy
divergence measure (5.2) in the following theorems. While proving these theorems
we consider the separation of X into two parts X1 and X2 such that
X1 fx=x 2 X; lA xi lB xi g 5:3
Theorem 5.2
(a) Lt A [ B; A \ B Lt A; B.
(b) Lt A [ B; A Lt A \ B; A Lt A; B.
(c) Lt A [ B; C Lt A; C Lt B; C.
Proof 2 3
lA [ B xi lA \ B xi t
X 6 pt 1
2 7
n
l A [ B xi lA \ B xi 6 lA [ B xi lA \ B xi 7
5:2a Lt A [ B; A \ B 6 7
2t 4 2lA [ B xi lA \ B xi t 5
i1 pt 1
1lA [ B xi 1lA \ B xi
" #
X l xi l xi 2 l xi l xi t 2 lA xi lB xi t
A
t
B
pt 1 pt 1
A B
X1
2 lA xi lB xi 1 lA xi 1 lB xi
" #
X l xi l xi 2 l xi l xi t 2 lB xi lA xi t
B A B A
pt 1 pt 1
X2
2t lB xi lA xi 1 lB xi 1 lA xi
" #
Xn
lA xi lB xi 2
lA xi lB xi t
2 lA xi lB xi t
p t1
p t1
i1
2t lA xi lB xi 1 lA xi 1 lB xi
Lt A; B:
Proof
2 3
lA [ B xi lC xi t
X 6 pt 1
2 7
n
l A [ B xi lC xi 6 lA [ B xi lC xi 7
5:3a Lt A [ B; C Lt A \ B; C 6 7
2t 4 2lA [ B xi lC xi t 5
i1 pt 1
1lA [ B xi 1lC xi
2 t 3
lA \ B xi lC xi
p t 1
X l lC xi 2 6 7
A \ B xi
n
lA \ B xi lC xi
6 7
2t 4 2lA \ B xi lC xi t 5
i1 pt 1
1lA \ B xi 1lC xi
" #
X l xi l xi 2 l xi l xi t 2 lA xi lC xi t
A C A C
pt 1 pt 1
X1
2t lA xi lC xi 1 lA xi 1 lC xi
" #
X l xi l xi 2 l xi l xi t 2 lB xi lC xi t
B C
p
B C
t1
p t1
X2
2t lB xi lC xi 1 lB xi 1 lC xi
" #
X l xi l xi 2 l xi l xi t 2 lB xi lC xi t
B C
p
B C
t1
p t1
X1
2t lB xi lC xi 1 lB xi 1 lC xi
" #
X l xi l xi 2 l xi l xi t 2 lA xi lC xi t
A
t
C
pt 1 pt 1
A C
X2
2 lA xi lC xi 1 lA xi 1 lC xi
" #
Xn
lA xi lC xi 2
lA xi lC xi t
2 lA xi lC xi t
t pt 1 pt 1
i1
2 lA xi lC xi 1 lA xi 1 lC xi
" #
X n
lB xi lC xi 2 lB xi lC xi t 2 lB xi lC xi t
t pt 1 pt 1
i1
2 l xi l xi 1 l xi 1 l xi
B C B C
Lt A; C Lt B; C:
98 5 Applications of Generalized Fuzzy Divergence Measure
X1
2t l xi l xi 1 l xi 1 l xi
A B A B
Now
2 3
lB xi lA [ B xi t
X 6 pt 1
2 7
lB xi lA [ B xi 6
n
lB xi lA [ B xi 7
Lt B; A [ B 6 7
2t 4 2lB xi lA [ B xi t 5
i1 pt 1
1lB xi 1lA [ B xi
" #
X l xi l xi 2 l xi l xi t 2 lB xi lA xi t
B A
pt 1 pt 1
B A
X1
2t l xi l xi 1 l xi 1 l xi
B A B A
Lt A; A \ B:
Now
2 3
lB xi lA \ B xi t
p
t 1
Xn
lB xi lA \ B xi 2 6
6 lB xi lA \ B xi
7
7
Lt B; A \ B 6 7
2 t 4 2lB xi lA \ B xi t 5
i1 pt 1
1lB xi 1lA \ B xi
" #
X l xi l xi 2 l xi l xi t 2 lB xi lA xi t
B A
pt 1 pt 1
B A
X2
2t lB xi lA xi 1 lB xi 1 lA xi
Lt A; A [ B:
5.2 Properties of Proposed Generalized Fuzzy Divergence Measure 99
Proof
2 3
2 t
Xn
lA xi lA xi 6 lA xi lA xi t 2 lA xi lA xi 7
5:4a Lt A; A 4pt 1 q
2 t
l x l x t 1 5
i1 A i A i 1 lA xi 1 lA xi
" #
X lA xi 1 lA xi
n 2
lA xi 1 lA xi t
2 lA xi 1 lA xi t
p t 1 pt 1
i1
2t l xi 1 l xi 1 l xi l xi
A A A A
X
n
2lA xi 12
p t 1
i1 2t1 lA xi 1 lA xi
Now
2 3
2 t
Xn
lA xi lA xi 6 lA xi lA xi t 2 lA xi lA xi 7
Lt A; A 4pt 1 q
t 1 5
2t lA xi lA xi
i1 1 lA xi 1 lA xi
" #
Xn
1 lA xi lA xi 2
1 lA xi lA xi t
2 1 lA xi lA xi t
pt 1 pt 1
i1
2t 1 l xi l xi l xi 1 l xi
A A A A
X
n
2lA xi 12
p t 1
i1 2t1 lA xi 1 lA xi
Lt A; A:
2 3
2 t
Xn
lA xi lB xi 6 lA xi lB xi t 2 lA xi lB xi 7
5:4c Lt A; B 4pt 1 q
2t lA xi lB xi t 1 5
i1 1 lA xi 1 lB xi
" #
Xn
lA xi lB xi 12 1 lA xi lB xi t 1 lA xi lB xi t
pt 1 pt 1
i1
2t l xi 1 l xi 1 l xi l xi
A B A B
Now
2 3
2 t
Xn
lA xi lB xi 6 lA xi lB xi t
2 lA xi lB xi 7
Lt A; B 4pt 1 q
t 1 5
2t lA xi lB xi
i1 1 lA xi 1 lB xi
" #
X 1 l xi l xi
n 2
1 lA xi lB xi t
1 lA xi lB xi t
A B
pt 1 pt 1
i1
2t 1 l xi l xi l xi 1 l xi
A B A B
Lt A; B:
In recent years, the applications of the fuzzy divergence measure have been given in
different areas: Poletti et al. [6] in bioinformatics; Bhandari et al. [7], Fan et al. [8]
and Bhatia and Singh [9] in image thresholding; Ghosh et al. [10] in automated
leukocyte recognition. We present the application of the proposed generalized fuzzy
divergence measure in the context of pattern recognition and multi-criteria
decision-making.
Dealing with uncertainties is a common problem in pattern recognition and the use
of fuzzy set theory has given rise to a lot of new methods of pattern recognition. We
now present application of the proposed generalized fuzzy divergence measure in
the context of pattern recognition.
In order to demonstrate the applications of the introduced generalized fuzzy
divergence measure in pattern recognition, suppose that we are given m known
patterns P1 ; P2 ; P3 ; . . .; Pm which have classications C1 ; C2 ; C3 ; . . .; Cm respec-
tively. The patterns are represented by the following fuzzy sets in the universe of
discourse X fx1 ; x2 ; x3 ; . . .; xn g:
5.3 Applications of the Proposed Generalized Fuzzy Divergence Measure 101
Pi xj ; lPi xj xj 2 X
where i 1; 2; . . .; m and j 1; 2; . . .; n.
Given an unknown pattern Q, represented by the fuzzy set
Qi xj ; lQi xj xj 2 X :
According to this algorithm, the given pattern can be recognized so that the best
class can be selected. It is a practical application of minimum divergence measure
principle of Shore and Gray [11] to pattern recognition.
A Numerical Example
We consider the problem having four known patterns P1 ; P2 ; P3 and P4 which have
classications C1 ; C2 ; C3 and C4 respectively. These are represented by the fol-
lowing fuzzy sets in the universe of discourse X fx1 ; x2 ; x3 g:
Our aim now is to classify Q to one of the classes C1 ; C2 ; C3 and C4 . From the
formula (5.2), we can compute the values of generalized fuzzy divergence measure
Lt Pk ; Q; k f1; 2; 3; 4g, for any t 0 and are presented in Table 5.2 as follows.
It is observed that Q can be classied to C3 correctly.
where lij indicates the degree that the option Mi satises the criterion Cj .
We introduce the approach to solve the above multi-criteria fuzzy
decision-making problem using Eq. (5.2). The computational procedure of the same
is as follows.
Step 1 Find out
the positive-ideal
solution
M and negative-ideal
solution M
:
M l1 ; l2 ; . . .; ln and M l 1 ; l
2 ; . . .; l
n ,
where for each j 1; 2; 3; . . .; n,
D E
9
lj max lij > >
=
D E i
: 5:5
>
>
lj min lij ;
i
Lt M ; Mi
where Lt Mi ; i 1; 2; 3; . . .; m: 5:6
Lt M ; Mi Lt M ; Mi
Step 4 Rank the preference order of all alternatives according to the relative
fuzzy divergence measure.
Step 5 Select the best alternative Mk with the smallest Lt Mk .
5.3 Applications of the Proposed Generalized Fuzzy Divergence Measure 103
A Numerical Example
We now demonstrate the applicability of the proposed generalized fuzzy divergence
measure to solve a real problem related to multi-criteria decision-making. For this
we consider a customer decision-making problem in purchasing a car.
Example Consider a customer who wants to buy a car. Let ve types of cars, i.e.
the alternatives M fM1 ; M2 ; M3 ; M4 ; M5 g be available in the market. To buy a car
the customer takes the following four criteria to decide: (i) Quality of Product (C1 ),
(ii) Price (C2 ), (iii) Technical Capability (C3 ) and (iv) Fuel Economy (C4 ).
The ve possible options are to be evaluated by the decision-maker under the
above four criteria in the following form:
Step 2 Tables 5.3 and 5.4 show the calculated numerical values of Lt M ; Mi
and Lt M ; Mi using Eq. (5.2) for t 0.
Step 3 Calculated values of the relative fuzzy divergence measure Lt Mi for
i 1; 2; 3; 4; 5 with t 0 are presented in Table 5.5.
Step 4 According to the calculated numerical values of relative divergence for
different values of t 0, ranking order of alternatives is as follows:
For t 0; M3 [ M5 [ M4 [ M2 [ M1 :
For t 1; M3 [ M4 [ M5 [ M2 [ M1 :
For t 2; M3 [ M4 [ M5 [ M1 [ M2 :
For t 10; M3 [ M4 [ M5 [ M1 [ M2 :
For t 50; M3 [ M4 [ M5 [ M2 [ M1 :
This chapter proposed and validated a new parametric generalized measure of fuzzy
divergence. The particular cases have been discussed in detail along with some of
the properties of this fuzzy divergence measure. The applications of the proposed
generalized fuzzy divergence measure are shown to pattern recognition and
multi-criteria fuzzy decision-making. Finally, a numerical example is provided to
illustrate the multi-criteria decision-making process proposed by us. We note that
the presence of the parameter in the proposed divergence measure provides greater
flexibility for pattern recognition and multi-criteria decision-making.
References 105
References
In real-life situations, the human thinking style involves subjectivity which intro-
duces vagueness in decision-making. This vagueness or uncertainty is easily han-
dled by fuzzy set theory. This chapter deals with the issue of Hellingers
generalized measure of discrimination in fuzzy environment. Multi-criteria
decision-making (MCDM) has been one of the rapidly developing areas based
on the altering in the commercial area. For this, decision-maker requires a decision
support to decide among different options or alternatives by surpassing the less
suitable options quickly. At the managerial and industrial point Duckstein and
Opricovic [3] considered multiple-criteria decision-making as a compound and
dynamic process. Hwang and Yoon [7] and Dyer et al. [4] precisely presented the
method of weights and ranking of the criteria in the classical MCDM methods.
Zeleny [11] and Hwang and Yoon [7] presented a survey of the MCDM methods
based on the concept that the preferred alternative should have a smaller distance
from the positive-ideal solution and the largest distance from the negative-ideal
solution.
Bellman and Zadeh [1] introduced the theory of solving MCDM problems in a
fuzzy environment which is acknowledged as fuzzy multi-criteria decision-making
(FMCDM) problems. Erginel et al. [5] analyzed that the fuzzy theory is often used
in the study of multi-criteria decision-making approaches since MCDM problems
are mainly depend on individual views and qualitative information.
Hellingers measure of discrimination was rst introduced by Hellinger [6].
Further, Taneja [10] proposed a generalized Hellingers discrimination given by
p p2t 1
Xn
pi qi
ht P; Q pt ; P; Q 2 Cn Cn ; t 2 N:
i1 pi qi
( , ) 6:1
Xn
where Cn P p1 ; p2 ; . . .; pn pi [ 0; pi 1 ; n 2:
i1
ht A; B
"p p2t 1 p p2t 1 #
X n
lA xi lB xi 1 lA xi 1 lB xi
pt pt ; t 2 N:
i1 l x l
A i B i x 1 lA xi 1 lB xi
6:2
@ht A; B
@lA xi
p p2t 1
2t 2 lA xi lB xi 1
p
lA xi lB xi t=2 2 lA xi
t p p2t 2
lA xi lB xi
lB xi
lA xi lB xi 21
t
2
p p2t 1
1 lA xi 1 lB xi 1
2t 1 pt p 1
1 lA xi 1 lB xi 2 1 lA xi
p p2t 2
t 1 lA xi 1 lB xi
1 lB xi
2 1 lA xi 1 lB xi 2 1
t
@ 2 ht A; B
@l2A xi
2 p p2t 3
lA xi lB xi
6 pt 7
t 12t 1 6
6 lA xi lA xi lB xi
7
7
6 p
p
7
2 6 1 lA xi 1 lB xi
2t 7
4 5
pt
1 lA xi 1 lA xi 1 lB xi
2 p p2t 1 3
lA xi lB xi
6 3=2 pt 7
t 12t 1 6
6 lA xi lA xi lB xi
7
7
6 p p2t 1 7
2 6 1 lA xi 1 lB xi 7
4 5
3=2
pt
1 lA xi 1 lA xi 1 lB xi
2 p p2t 2 3
lA xi lB xi
6 pt 7
tt 2 6
6 lA xi lA xi lB xi
2 7
7
6 p
p
7 [ 0; t 2 N:
4 6 1 lA xi 1 lB xi
2t 2 7
4 5
p t
1 lA xi 2 1 lA xi 1 lB xi
@ 2 ht A; B
Similarly, @l2B xi
[ 0; t 2 N:
Thus ht A; B is a convex function of fuzzy sets A and B and hence in view of the
denition of fuzzy divergence measure of Bhandari and Pal [2] ht A; B is a valid
measure of fuzzy divergence. Moreover, we can easily check for measure ht A; B
110 6 Generalized Hellingers Fuzzy Divergence Measure
X1 fx=x 2 X; lA xi lB xi g 6:3
Theorem 6.2
(a) ht A [ B; A ht A \ B; A ht A; B.
(b) ht A [ B; A \ B ht A \ B; A [ B ht A; B.
(c) ht A; A ht A; A.
Proof 6.2(a)
ht A [ B; A ht A \ B; A
"p p2t 1 p p2t 1 #
X n
lA [ B xi lA xi 1 lA [ B xi 1 lA xi
pt pt
i1 lA [ B xi lA xi 1 lA [ B xi 1 lA xi
"p p2t 1 p p2t 1 #
X n
lA \ B xi lA xi 1 lA \ B xi 1 lA xi
pt pt
i1 lA \ B xi lA xi 1 lA \ B xi 1 lA xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt
X2 lA xi lB xi 1 lA xi 1 lB xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt
X1 lA xi lB xi 1 lA xi 1 lB xi
"p p2t 1 p p2t 1 #
X n
lA xi lB xi 1 lA xi 1 lB xi
pt pt
i1 l x l
A i B i x 1 lA xi 1 lB xi
ht A; B:
6.2 Properties of Generalized Fuzzy Hellingers Divergence Measure 111
ht A [ B; A \ B
2p 2t 1 3
p p p2t 1
X n lA [ B xi lA \ B xi l x l x
6 1 A[B i 1 A\B i 7
4 pt t 5
p
i1 l x l
A[B i A\B i x 1 l A[B i x 1 l A\B i x
"p p2t 1 p p2t 1 #
X
n
1 lA [ B xi 1 lA \ B xi lA [ B xi lA \ B xi
pt pt
i1 1 lA [ B xi 1 lA \ B xi lA [ B xi lA \ B xi
"p p2t 1 p p2t 1 #
X 1 lA xi 1 lB xi lA xi lB xi
pt pt
X1 1 lA xi 1 lB xi lA xi lB xi
"p p2t 1 p p2t 1 #
X 1 lB xi 1 lA xi lB xi lA xi
pt pt
X2 1 lB xi 1 lA xi lB xi lA xi
ht A; B:
Now
ht A \ B; A [ B
"p p2t 1 p p2t 1 #
X n
lA \ B xi lA [ B xi 1 lA \ B xi 1 lA [ B xi
pt p
t
i1 lA \ B xi lA [ B xi 1 lA \ B xi 1 lA [ B xi
"p p2t 1 p p2t 1 #
X lA xi lB xi 1 lA xi 1 lB xi
pt pt
X1 lA xi lB xi 1 lA xi 1 lB xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt
X2 lB xi lA xi 1 lB xi 1 lA xi
"p p2t 1 p p2t 1 #
X 1 lA xi 1 lB xi lA xi lB xi
pt pt
X1 1 l x
A i 1 l x
B i lA xi lB xi
"p p2t 1 p p2t 1 #
X 1 lB xi 1 lA xi lB xi lA xi
pt pt
X2 1 lB xi 1 lA xi lB xi lA xi
ht A; B:
Hence, ht A [ B; A \ B ht A \ B; A [ B ht A; B.
112 6 Generalized Hellingers Fuzzy Divergence Measure
Hence, ht A; A ht A; A.
Hence, 6.2(c) holds.
Theorem 6.3
(a) ht A [ B; C ht A \ B; C ht A; C ht B; C.
(b) ht A; A [ B ht B; A \ B.
(c) ht A; A \ B ht B; A [ B.
Proof 6.3(a)
ht A [ B; C ht A \ B; C
"p p2t 1 p p2t 1 #
X n
lA [ B xi lC xi 1 lA [ B xi 1 lC xi
pt pt
i1 lA [ B xi lC xi 1 lA [ B xi 1 lC xi
"p p2t 1 p p2t 1 #
X n
lA \ B xi lC xi 1 lA \ B xi 1 lC xi
pt pt
i1 lA \ B xi lC xi 1 lA \ B xi 1 lC xi
"p p2t 1 p p2t 1 #
X lA xi lC xi 1 lA xi 1 lC xi
pt pt
X1 l x l
A i C i x 1 lA xi 1 lC xi
"p p2t 1 p p2t 1 #
X lB xi lC xi 1 lB xi 1 lC xi
pt pt
X2 lB xi lC xi 1 lB xi 1 lC xi
6.2 Properties of Generalized Fuzzy Hellingers Divergence Measure 113
ht A; A [ B
"p p2t 1 p p2t 1 #
X n
lA xi lA [ B xi 1 lA xi 1 lA [ B xi
pt pt
i1 lA xi lA [ B xi 1 lA xi 1 lA [ B xi
"p p2t 1 p p2t 1 #
X lA xi lB xi 1 lA xi 1 lB xi
pt pt
X2 l x l
A i B i x 1 lA xi 1 lB xi
Now
ht B; A \ B
"p p2t 1 p p2t 1 #
X n
lB xi lA \ B xi 1 lB xi 1 lA \ B xi
pt pt
i1 lB xi lA \ B xi 1 lB xi 1 lA \ B xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt
X2 l x l
B i A i x 1 lB xi 1 lA xi
ht A; A [ B:
ht A; A \ B
"p p2t 1 p p2t 1 #
X n
lA xi lA \ B xi 1 lA xi 1 lA \ B xi
pt pt
i1 lA xi lA \ B xi 1 lA xi 1 lA \ B xi
"p p2t 1 p p2t 1 #
X lA xi lB xi 1 lA xi 1 lB xi
pt pt
X1 lA xi lB xi 1 lA xi 1 lB xi
114 6 Generalized Hellingers Fuzzy Divergence Measure
Now
ht B; A [ B
"p p2t 1 p p2t 1 #
X n
lB xi lA [ B xi 1 lB xi 1 lA [ B xi
pt pt
i1 lB xi lA [ B xi 1 lB xi 1 lA [ B xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt
X1 lB xi lA xi 1 lB xi 1 lA xi
ht A; A \ B:
Proof 6.4(a)
ht A [ B; A \ B
"p p2t 1 p p2t 1 #
X n
lA [ B xi lA \ B xi 1 lA [ B xi 1 lA \ B xi
pt pt
i1 lA [ B xi lA \ B xi 1 lA [ B xi 1 lA \ B xi
"p p2t 1 p p2t 1 #
X lA xi lB xi 1 lA xi 1 lB xi
pt pt
X1 l x l
A i B i x 1 lA xi 1 lB xi
"p p2t 1 p p2t 1 #
X lB xi lA xi 1 lB xi 1 lA xi
pt pt s
X2 lB xi lA xi 1 lB xi 1 lA xi
"p p2t 1 p p2t 1 #
X n
lA xi lB xi 1 lA xi 1 lB xi
pt pt
i1 lA xi lB xi 1 lA xi 1 lB xi
ht A; B:
6.4(b)
"p p2t 1 p p2t 1 #
X
n
lA xi lB xi 1 lA xi 1 lB xi
ht A; B pt p t
i1 lA xi lB xi 1 lA xi 1 lB xi
"p p2t 1 p p2t 1 #
Xn
1 lA xi 1 lB xi lA xi lB xi
pt pt
i1 1 lA xi 1 lB xi lA xi lB xi
ht A; B:
Fuzzy sets (FSs) are a suitable tool to handle with imprecisely dened facts and
data, as well as with the vague knowledge. Next we provide the applications of the
generalized Hellingers fuzzy divergence measure in multi-criteria decision-making
and medical diagnosis problems. Thereafter we present the application of the
generalized Hellingers fuzzy divergence in the context of multi-criteria
decision-making.
116 6 Generalized Hellingers Fuzzy Divergence Measure
Ai Cj ; lij ; Cj 2 C ; i 1; 2; 3; . . .; m and j 1; 2; 3; . . .; n
where lij indicates the degree that the option Ai satises the criterion Cj .
We introduce the approach to solve the above multi-criteria fuzzy
decision-making problem using Eq. (6.2). The computational procedure of the same
is in order.
Step 1: Construct a fuzzy decision matrix
C1 C2 C3 ... Cn
A1 d11 d12 d13 ... d1n
A2 d21 d22 d23 ... d2n
A3 d31 d32 d33 ... d3n
Am dm1 dm2 dm3 ... dmn
A l1 ; l2 ; . . .; ln and A l
1 ; l2 ; . . .; ln
ht A ; Ai
ht Ai ; i 1; 2; 3; . . .; m 6:7
ht A ; Ai ht A ; Ai
Step 6: Rank the preference order of all alternatives according to the relative fuzzy
divergence measure.
Step 7: Select the best alternative Ak with smallest ht Ak .
Numerical Example 6.1
We here demonstrate the applicability of the proposed generalized fuzzy divergence
measure to solve a problem related to real multi-criteria decision-making. For this,
we consider the decision-making problem of a customer while purchasing an air
conditioner.
Example A customer wants to buy an air conditioner. There are four options (air
condition systems) Ai i 1; 2; 3; 4 available in the market. Suppose that three
criteria: (i) C1 , economical; (ii) C2 , functional; and (iii) C3 , being operative, are
taken into consideration during the selection procedure.
The four possible options (alternatives) are to be estimated by the customer
under the above three criteria in the form of following sets:
Step 2: Here C1 is the cost criteria while other two are benet criteria. We trans-
form the cost criteria C1 into benet criteria C1c ; we get the normalized fuzzy
decision matrix in Table 6.2.
Step 3: The positive-ideal solution A and negative-ideal solution A , respectively
are
Step 4: Tables 6.3 and 6.4 show the calculated numerical values of ht A ; Ai and
ht A ; Ai using Eq. (6.2) for t 2 N.
Step 5: Calculated values of the relative fuzzy divergence measure ht Ai for
i 1; 2; 3; 4 with t 2 N are presented in Table 6.5.
Step 6: According to the calculated numerical values of relative divergences for
different values of t 2 N, ranking order of alternatives is given in Table 6.6.
Step 7: Thus we here nd that variation in the values of t does not bring change in
ranking. So A1 is the most preferable alternative.
Many researchers have found the application of fuzzy sets theory in the diagnosis of
diseases. Fuzzy pattern recognition algorithms are suitable tool to solve the medical
diagnosis problems for recognizing the disease which is a challenging research area
from a practical point of view. In order to illustrate the efciency of the proposed
generalized fuzzy divergence measure in solving medical diagnosis problems, using
the algorithm given in Ohlan [8], we provide a numerical example.
Numerical Example 6.2
Suppose that the universe of discourse X is a set of symptoms
Q1 fhx1 ; 0:7i; hx2 ; 0:2i; hx3 ; 0:0i; hx4 ; 0:7i; hx5 ; 0:1ig;
Q2 fhx1 ; 0:4i; hx2 ; 0:3i; hx3 ; 0:1i; hx4 ; 0:4i; hx5 ; 0:1ig;
Q3 fhx1 ; 0:1i; hx2 ; 0:2i; hx3 ; 0:8i; hx4 ; 0:2i; hx5 ; 0:2ig;
Q4 fhx1 ; 0:1i; hx2 ; 0:0i; hx3 ; 0:2i; hx4 ; 0:2i; hx5 ; 0:8ig;
Q5 fhx1 ; 0:3i; hx2 ; 0:6i; hx3 ; 0:2i; hx4 ; 0:2i; hx5 ; 0:1ig:
120 6 Generalized Hellingers Fuzzy Divergence Measure
References
4. Dyer JS, Fishburn PC, Steuer RE, Wallenius J, Zionts S (1992) Multiple criteria decision
making, multiattribute utility theory: the next ten years. Manage Sci 38(5):645654
5. Erginel N, akmak T, entrk S (2010) Determining the preference of GSM operators in
turkey with fuzzy topsis after mobile number portability system application. Anadolu Univ J
Sci Tech Appl Sci Eng 11(2):8193
6. Hellinger E (1909) Neue Begrundung der theorie der quadratischen formen von unendlichen
vielen veranderlichen. Journal fr die reine und angewandte Mathematik 136:210271
7. Hwang CL, Yoon K (1981) Multiple attribute decision makingmethods and applications.
Springer, New York
8. Ohlan A (2016) Intuitionistic fuzzy exponential divergence: application in multi-attribute
decision making. J Intell Fuzzy Syst 30:15191530
9. Singh RP, Tomar VP (2008) On fuzzy divergence measures and their inequalities.
Proceedings of 10th National Conference of ISITA 4143
10. Taneja IJ (2013) Seven means, generalized triangular discrimination and generating
divergence measures. Information 4:198239
11. Zeleny M (1973) Compromise programming. In: Cochrane JL, Zeleny M (eds) Multiple
criteria decision making. University of South Carolina Press, Columbia, SC, pp 262301
Chapter 7
Intuitionistic Fuzzy Exponential
Divergence and Multi-attribute
Decision-Making
In this chapter, we apply the exponential approach on intuitionistic fuzzy sets (IFSs)
and propose a new information-theoretic divergence measure, called intuitionistic
fuzzy exponential divergence, to compute the difference between two IFSs.
Indeed, the notion of Atanassovs IFSs was rst originated by Atanassov [1]
which found to be well suited to deal with both fuzziness and lack of knowledge or
non-specicity. It is noted that the concept of an IFS is the best alternative approach
to dene a fuzzy set (FS) in cases where existing information is not enough for the
denition of imprecise concepts by means of a conventional FS. Thus, the concept
of Atanassov IFSs is the generalization of the concept of FSs. Gau and Buehrer [9]
introduced the notion of vague sets. But, Bustince and Burillo [6] presented that the
notion of vague sets was equivalent to that of Atanassov IFSs. As a very signicant
content in fuzzy mathematics, the study on the divergence measure between IFSs
has received more attention in recent years. Divergence measures of IFSs have been
widely applied to many elds such as pattern recognition [10, 12, 18, 24], linguistic
variables [11], medical diagnosis [7, 27], logical reasoning [14] and decision-
making [17, 21, 22, 28]. Since the divergence measures of IFSs have been applied
to many real-world situations, it is expected to have an efcient divergence measure
which deals with the aspect of uncertainty, i.e. fuzziness and non-specicity or lack
of knowledge.
Section 7.1 is devoted to review some relevant concepts related to fuzzy set
theory and intuitionistic fuzzy set theory. In Sect. 7.2 we introduce the intuitionistic
exponential divergence measure between IFSs and prove its essential properties.
Some of elegant properties are proved on the proposed divergence measure in
Sect. 7.3. In what follows, Sect. 7.4 presents the efciency of the proposed intu-
itionistic fuzzy exponential divergence in pattern recognition by comparing it with
some existing measures with the help of a numerical example. In this way, a
method is presented to solve MADM problem using the proposed divergence
measure in the intuitionistic fuzzy environment in Sect. 7.5. Section 7.6 presents
7.1 Preliminaries
We begin by reviewing some relevant concepts related to fuzzy set theory and
intuitionistic fuzzy set theory.
Denition 7.1 (Fuzzy Set [26]) A fuzzy set A0 dened on a nite universe of
discourse X fx1 ; x2 ; . . .; xn g is given as
Similarly, we have
2 3
mB xi 1 lB xi lB xi lA xi mB xi mA xi
n 61 e 7
2
X 6 2 7
IB; A 6 7 7:6
4 l x i 1 m B x i lA xi lB xi mA xi mB xi 5
i1 B
e 2
2
126 7 Intuitionistic Fuzzy Exponential Divergence
6 2 e 2
7
X6
n 2 2 7
6 7
4 lA xi 1 mA xi mB xi 1 lB xi lB xi lA xi mB xi mA xi 5
i1
e 2
2 2
DEIFS A; B
2 3
lA xi lB xi mA xi mB xi lA xi lB xi mA xi mB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l x
A i l x
B i m x
A i m x
B i mA xi mB xi lA xi lB xi 5
i1
1 e 2
2
7:7
A 1; 0; B 0; 1 or A 0; 1; B 1; 0:
n 6
2 1 e 2
7
X 6 2 7
6 7
4 lA xi lC xi mA xi mC xi mA xi mC xi lA xi lC xi 5
i1
1 e 2
2
2 3
lB xi lC xi mB xi mC xi lB xi lC xi mB xi mC xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l x i l x i m B xi m C x i mB xi mC xi lB xi lC xi 5
i1
1 B C
e 2
2
2 3
lA [ B xi lC xi mA [ B xi mC xi lA [ B xi lC xi mA [ B xi mC xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l x l x m A[B i x m x mA [ B xi mC xi lA [ B xi lC xi 5
1 A[B
i1 i C i C i
e 2
2
2 3
lA xi lC xi mA xi mC xi lA xi lC xi mA xi mC xi
6 2 1 e 2
7
X6 2 7
6 7
X1
4 l x i l x
C i m x
A i m x
C i mA xi mC xi lA xi lC xi 5
1 A e 2
2
2 3
lB xi lC xi mB xi mC xi lB xi lC xi mB xi mC xi
6 2 1 e 2
7
X6 2 7
6 7
X2
4 lB xi lC xi mB xi mC xi mB xi mC xi lB xi lC xi 5
1 e 2
2
0:
128 7 Intuitionistic Fuzzy Exponential Divergence
Proof DEIFS A [ B; C
2 3
lA [ B xi lC xi mA [ B xi mC xi lA [ B xi lC xi mA [ B xi mC xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l A[B x i l x i m A [ B x i m C x i mA [ B xi mC xi lA [ B xi lC xi 5
i1
1 C
e 2
2
2 3
lB xi lC xi mB xi mC xi lB xi lC xi mB xi mC xi
2 1 e 2
X6 6 2 7
7
6 7
X1
4 l x i l x i mB xi m C x i mB xi mC xi lB xi lC xi 5
1 B C
e 2
2
2 3
lA xi lC xi mA xi mC xi lA xi lC xi mA xi mC xi
2 1 e 2
X6 6 2 7
7
6 7
X2
4 l x i l x i mA x i m C x i mA xi mC xi lA xi lC xi 5
1 A C
e 2
2
7:8
Now DEIFS A \ B; C
2 3
lA \ B xi lC xi mA \ B xi mC xi lA \ B xi lC xi mA \ B xi mC xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l A\B x i l x i m A \ B x i m C x i mA \ B xi mC xi lA \ B xi lC xi 5
i1
1 C
e 2
2
2 3
lA xi lC xi mA xi mC xi lA xi lC xi mA xi mC xi
2 1 e 2
X6 6 2 7
7
6 7
X1
4 l x i l x i mA xi m C x i mA xi mC xi lA xi lC xi 5
1 A C
e 2
2
2 3
lB xi lC xi mB xi mC xi lB xi lC xi mB xi mC xi
2 1 e 2
X6 6 2 7
7
6 7
X2
4 l x i l x i mB x i m C x i mB xi mC xi lB xi lC xi 5
1 B C
e 2
2
7:9
7.3 Properties of Intuitionistic Fuzzy Exponential Divergence Measure 129
2 3
lB xi lA xi mB xi mA xi lB xi lA xi mB xi mA xi
2 1 e 2
X6 6 2 7
7
6 7
X1
4 lB xi lA xi mB xi mA xi mB xi mA xi lB xi lA xi 5
1 e 2
2
2 3
lA xi lB xi mA xi mB xi lA xi lB xi mA xi mB xi
2 1 e 2
X6 6 2 7
7
6 7
X2
4 l x i l x i m A x i m B x i mA xi mB xi lA xi lB xi 5
1 A B
e 2
2
DEIFS A; B:
Proof 7.5(a)
130 7 Intuitionistic Fuzzy Exponential Divergence
DEIFS A; A [ B
2 3
2 1 lA xi lA [ B xi m A xi mA [ B xi
6 2 7
6 lA xi lA [ B xi mA xi mA [ B xi 7
X6n
6e
7
7
2
6 7
i1 6
6 1 A i A[B i l x l x m x
A i m A[B ix 7
2 7
4 5
mA xi mA [ B xi lA xi lA [ B xi
e 2
2 3
lA xi lB xi mA xi mB xi lA xi lB xi mA xi mB xi
2 1 e 2
X6 6 2 7
7
6 7
X1
4 lA xi lB xi mA xi mB xi m x
A i m x
B i l x
A i l x
B i 5
1 e 2
2
7:10
Now
DEIFS A; A \ B
2 3
lA xi lA \ B xi mA xi mA \ B xi lA xi lA \ B xi mA xi mA \ B xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l xi lA \ B xi mA xi mA \ B xi mA x i mA\B xi l A xi l A\B xi 5
i1
1 A e 2
2
2 3
lA xi lB xi mA xi mB xi lA xi lB xi mA xi mB xi
X6 2 1 e 2
7
6 2 7
6 7
4 l x i l x
B i m x
A i m x
B i mA xi mB xi lA xi lB xi 5
X2 1 A e 2
2
7:11
2 3
lAc xi lBc xi mAc xi mBc xi lAc xi lBc xi mAc xi mB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 lAc xi lBc xi mAc xi mBc xi mAc xi mBc xi lAc xi lBc xi 5
i1
1 e 2
2
2 3
mA xi mB xi lA xi lB xi mA xi mB xi lA xi lB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 mA xi mB xi lA xi lB xi lA xi lB xi mA xi mB xi 5
i1
1 e 2
2
2 3
lA xi lB xi mA xi mB xi mA xi mB xi lA xi lB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 lA xi lB xi mA xi mB xi lA xi lB xi mA xi mB xi 5
i1
1 e 2
2
DEIFS A; B:
n 6
2 1 e 2
7
X 6 2 7
6 7
4 lAc xi lB xi mAc xi mB xi mAc xi mB xi lAc xi lB xi 5
i1
1 e 2
2
2 3
mA xi lB xi lA xi mB xi mA xi lB xi lA xi mB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 mA xi lB xi lA xi mB xi lA xi mB xi mA xi lB xi 5
i1
1 e 2
2
Now DEIFS A; Bc
2 3
lA xi lBc xi mA xi mBc xi lA xi lBc xi mA xi mBc xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l x
A i l c x i m x
A i m Bc xi mA xi mBc xi lA xi lBc xi 5
i1
1 B
e 2
2
2 3
lA xi mB xi mA xi lB xi lA xi mB xi mA xi lB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 l x
A i m x
B i m x
A i l x
B i mA xi lB xi lA xi mB xi 5
i1
1 e 2
2
2 3
mA xi lB xi lA xi mB xi mA xi lB xi lA xi mB xi
n 6
2 1 e 2
7
X 6 2 7
6 7
4 m x
A i l x
B i l x
A i m x
B i lA xi mB xi mA xi lB xi 5
i1
1 e 2
2
DEIFS Ac ; B:
132 7 Intuitionistic Fuzzy Exponential Divergence
Hung and Yang [11] dened the distance between intuitionistic fuzzy sets A and
B using hausdorff distance as follows:
1
dH A; B HIA xi ; IB xi 7:13
n
where
" #
X
n
lA xi mA xi
IIFS A; B lA xi In 1 mA xi In 1
i1 2 lA x i l B x i 2 m A x i mB xi
where I l A; B In llA x i
is the amount of discrimination of lA xi from lB xi .
B xi
Zhang and Jiang [27] presented a measure of divergence between IFSs/vague
sets A and B as
where
DA; B
2 3
lA xi 1 mA xi lA xi 1 mA xi =2
log2 1
Xn 6 7
4 flA xi 1 mA xi lB xi 1 mB xi g
6 2 7
6 7:
4 1 lA xi mA xi 1 lA xi mA xi =2 5
i1
log2 1
2 4f 1 l x
A i m x
A i 1 l x
B i m x
B i g
According to this algorithm, the given pattern can be recognized so that the best
class can be selected.
Example 7.1 Given three known patterns P1 ; P2 and P3 which have classications
C1 ; C2 and C3 respectively. These are represented by the following IFSs in the
universe of discourse X fx1 ; x2 ; x3 g:
where for i 1; 2; 3
Pi x1 ; lPi x1 ; mPi x1 ; x2 ; lPi x2 ; mPi x2 ; x3 ; lPi x3 ; mPi x3 :
134 7 Intuitionistic Fuzzy Exponential Divergence
our aim here is to classify Q to one of the classes C1 ; C2 and C3 . From the formulae
(7.7), (7.12)(7.15), we compute the values of different intuitionistic fuzzy diver-
gence measures and are presented in Table 7.1.
From the computed numerical values of different existing measures and the
proposed measure DEIFS presented in Table 7.1, it is observed that the pattern Q
should be classied to C2 . Thus, the proposed intuitionistic fuzzy exponential
divergence measure is consistent for the application point of view in the context of
pattern recognition.
where i 1; 2; . . .; n; j 1; 2; . . .; m.
and dijc is the complement of dij with dijc mij ; lij .
Step II: Specify the Options by the Characteristic Sets
With the normalized matrix R rij n m we specify the option Aj , by the charac-
teristic sets given by
Aj Mi ; lij ; mij Mi 2 M ; i 1; 2; . . .; n; j 1; 2; . . .; m 7:17
8 j 1; 2; . . .m
136 7 Intuitionistic Fuzzy Exponential Divergence
Example 7.2 Let us assume that a car company wants to select a suitable material
supplier among ve alternative suppliers A1 ; A2 ; A3 ; A4 and A5 while considering six
attributes: (i) Quality of Product (M1 ), (ii) Price (M2 ), (iii) Technical Capability
(M3 ), (iv) Delivery (M4 ), (v) Service (M5 ), and (vi) Flexibility (M6 ).
Table 7.2 shows the intuitionistic fuzzy decision matrix D dij 6 5 having the
characteristics of the alternatives Aj j 1; 2; 3; 4; 5.
Step I: Here M2 is the cost attribute while other ve are benet attributes.
We transform the cost attribute M2 into benet attribute M2c we get
M2c f0:1; 0:8; 0:3; 0:5; 0:3; 0:7; 0:4; 0:3; 0:1; 0:7g
A1 fM1 ; 0:4; 0:5; M2 ; 0:1; 0:8; M3 ; 0:7; 0:3; M4 ; 0:6; 0:2; M5 ; 0:5; 0:4; M6 ; 0:3; 0:4g;
A2 fM1 ; 0:7; 0:2; M2 ; 0:3; 0:5; M3 ; 0:3; 0:4; M4 ; 0:8; 0:1; M5 ; 0:2; 0:6; M6 ; 0:4; 0:5g;
A3 fM1 ; 0:6; 0:1; M2 ; 0:3; 0:7; M3 ; 0:6; 0:2; M4 ; 0:4; 0:1; M5 ; 0:3; 0:4; M6 ; 0:8; 0:2g;
A4 fM1 ; 0:5; 0:4; M2 ; 0:4; 0:3; M3 ; 0:8; 0:1; M4 ; 0:7; 0:2; M5 ; 0:6; 0:1; M6 ; 0:7; 0:1g;
A5 fM1 ; 0:4; 0:3; M2 ; 0:1; 0:7; M3 ; 0:5; 0:2; M4 ; 0:9; 0:1; M5 ; 0:8; 0:0; M6 ; 0:6; 0:4g:
Step III: The ideal solution A obtained using Eqs. (7.18) and (7.19) is given by
A f0:7; 0:1; 0:4; 0:3; 0:8; 0:1; 0:9; 0:1; 0:8; 0:0; 0:8; 0:1g
Step IV: Table 7.4 gives the values of DEIFS Ak ; A using measure (7.20).
Step V: The best alternative is A4 obtained corresponding to the smallest value of
divergence measure DEIFS Aj ; A for j 4.
Hence, the numerical example shows that the proposed intuitionistic fuzzy
exponential divergence measure is a very suitable measure to solve the
multi-attribute decision-making problems.
0 ,v ,v1
uX uX
u m 2 u m 2
rij lij ; mij @lij t lij ; mij t mij A;
j1 j1 7:21
i 1; 2; . . .; n; j 1; 2; . . .; m
7:23
IFNIS A fv 0
1 ; v2 ; . . .; vn g fmin vij =i 2 J; max vij =i 2 J g 7:24
j j
where J is associated with the benet attribute and J 0 is associated with the cost
attribute.
5. Calculate the separation measures DEIFS and DEIFS using the divergence measure
in (7.7).
6. Calculate the relative closeness of the ideal solution.
The relative closeness of alternative Aj with respect to PIS is dened by
DEIFS
Cj ; j 1; 2; . . .; m: 7:25
DEIFS DEIFS
M fv
1 ; v2 ; . . .; vn g fmin vij ; max vij =j 2 Jg; 7:27
i i
References
8. Fan J, Xie W (1999) Distance measures and induced fuzzy entropy. Fuzzy Sets Syst 104
(2):305314
9. Gau WL, Buehrer DJ (1993) Vague sets. IEEE Trans Syst Man Cybern 23:610614
10. Hatzimichailidis AG, Papakostas GA, Kaburlasos VG (2012) A novel distance measure
of intuitionistic fuzzy sets and its application to pattern recognition problems. Int J Intell Syst
27(4):396409
11. Hung WL, Yang MS (2004) Similarity measures of intuitionistic fuzzy sets based on
Hausdorff distance. Pattern Recogn Lett 25:16031611
12. Hung WL, Yang MS (2008) On the J-divergence of intuitionistic fuzzy sets and its application
to pattern recognition. Inf Sci 178(6):16411650
13. Hwang CL, Yoon K (1981) Multiple attribute decision makingmethods and applications.
Springer, New York
14. Jiang YC, Tang Y, Wang J, Tang S (2009) Reasoning within intuitionistic fuzzy rough
description logics. Inf Sci 179:23622378
15. Li F, Lu ZH, Cai LJ (2003) The entropy of vague sets based on fuzzy sets. J Huazhong Univ
Sci Technol (Nature Science) 31:13
16. Li DF (2004) Some measures of dissimilarity in intuitionistic fuzzy structures. J Comput Syst
Sci 68(1):115122
17. Li DF (2005) Multi-attribute decision-making models and methods using intuitionistic fuzzy
sets. J Comput Syst Sci 70(1):7385
18. Papakostas GA, Hatzimichailidis AG, Kaburlasos VG (2013) Distance and similarity
measures between intuitionistic fuzzy sets: a comparative analysis from a pattern recognition
point of view. Pattern Recogn Lett 34(14):16091622
19. Shore JE, Gray RM (1982) Minimization cross-entropy pattern classication and cluster
analysis. IEEE Trans Pattern Anal Mach Intell 4(1):1117
20. Stanujkic D, Magdalinovic N, Tojanovic S, Jovanovic R (2012) Extension of ratio system part
of MOORA method for solving decision making problems with interval data. Informatica
23(1):141154
21. Verma R, Sharma BD (2012) On generalized intuitionistic fuzzy divergence (relative
information) and their properties. J Uncertain Syst 6(4):308320
22. Verma R, Sharma BD (2013) Intuitionistic fuzzy jensen-rnyi divergence: applications to
multiple-attribute decision making. Informatica 37(4):399409
23. Verma R, Sharma BD (2013) Exponential entropy on intuitionistic fuzzy sets. Kybernetika
49:114127
24. Vlachos IK, Sergiadis GD (2007) Intuitionistic fuzzy information-application to pattern
recognition. Pattern Recogn Lett 28(2):197206
25. Xu Z, Hu H (2010) Projection models for intuitionistic fuzzy multiple attribute decision
making. Int J Inf Technol Decis Making 9(2):267280
26. Zadeh LA (1965) Fuzzy sets. Inf Control 8:338353
27. Zhang QS, Jiang SY (2008) A note on information entropy measures for vague sets and its
applications. Inf Sci 178:41844191
28. Zhang SF, Liu SY (2011) A GRA-based intuitionistic multi-criteria decision making method
for personnel selection. Expert Syst Appl 38(9):1140111405
Epilogue
We hope the reader has enjoyed the material provided in this book. The book
focused on generalization of fuzzy information measures, fuzzy divergence mea-
sures and intuitionistic fuzzy divergence measures. The characterization and gen-
eralization of various measures of fuzzy information were used to introduce and
validate new generalized R-norm fuzzy information and divergence measures. The
relations were established between generalized fuzzy entropy measure and their
fuzzy divergence measures. A new parametric generalized exponential fuzzy
divergence measure was obtained in the process of generalizations.
In addition, a comparative application was provided in the context of strategic
decision-making. A sequence of fuzzy mean difference divergence measures was
introduced with a number of inequalities among them. In this way, their efciency
in pattern recognition and compound linguistic variables was achieved. Moreover, a
newly generalized fuzzy divergence measure and its application to multi-criteria
decision-making and pattern recognition were provided. Finally, the book addres-
sed the issue of intuitionistic fuzzy set theorya generalization of fuzzy set
theorywith a new denition for exponential divergence measure in intuitionistic
fuzzy nature. A number of interesting and elegant properties were found that
enhanced the importance of proposed exponential divergence measure.
A comparison of the proposed method of multi-attribute decision-making in an
intuitionistic fuzzy environment with the existing TOPSIS and MOORA methods
was presented with discussion. On an application level, the proposed generaliza-
tions of fuzzy information measures open the way to numerous applied areas. It was
shown that the proposed measure and method were more appropriate to solve the
real-world problems related to multi-attribute decision-making.
The present work is based on both the quantitative and the qualitative gener-
alized fuzzy information measures, divergence measures and intuitionistic fuzzy
divergence measures. It is found that these measures are more flexible from
application point of view in different elds. Some of the proposed measures could
be further compared with existing measures of divergence. The corresponding
A G
Applications of generalized fuzzy divergence Generalized information measures, 17, 18, 24,
measures, 9, 15, 17 25, 51, 143
C H
Characterization of information measures, 17, Hausdorff distance, 12, 132
18, 2426, 51, 143 Hellingers fuzzy divergence measure, 10, 16,
Comparative analysis, 13, 14, 53, 68, 124, 137, 18, 108, 115
141
Complement sets, 32 I
Compound variables, 89, 92, 143 Image processing, 6, 18, 23, 93
Control theory, 1 Image segmentation, 1
Crisp sets, 2, 3, 6, 24, 26 Imprecise information, 102
Inequalities among fuzzy mean difference
D divergence measures, 9, 71, 80
Decision making, 1, 11, 1315, 18, 23, 53, 62, Information measures, 17, 24, 26
64, 66, 68, 93, 100, 102104, 107, 108, Intuitionistic fuzzy divergence measures, 12,
115, 117, 120, 123, 124, 134, 137, 140, 13, 17, 132, 134, 143
141, 143 Intuitionistic fuzzy environment, 15, 16, 123,
Decision theory, 1, 13 133, 137, 141, 143, 144
Dissimilarity measure, 12, 13, 132 Intuitionistic fuzzy sets theory, 10, 123, 124,
Divergence measures, 810, 12, 13, 1517, 23, 143
24, 36, 37, 41, 49, 51, 53, 54, 61, 62,
7173, 76, 80, 8890, 93, 123, 132, 134, L
143 Linguistic variables, 11, 13, 17, 18, 34, 35, 71,
89, 90, 123, 143
F
Fuzziness, 2, 6, 7, 13, 23, 123 M
Fuzzy entropy, 6, 7, 8, 23, 24, 34, 35, 51, 53, Mean difference divergence, 10, 16, 17, 71, 73,
54, 68, 143 76, 80, 8890, 143
Fuzzy environment, 7, 15, 16, 64, 95, 107, 108, Medical diagnosis, 1, 11, 13, 17, 18, 108, 115,
120, 123, 133, 137, 141, 143 119, 120, 123
Fuzzy sets, 24, 5, 7, 8, 10, 11, 16, 23, 3436, Membership function, 24, 30, 124
39, 41, 44, 46, 4850, 53, 55, 56, 61, 73, 76, Multi-attribute decision making, 16, 18, 124,
8890, 94, 95, 100, 101, 108, 109, 115, 119 134, 135, 137, 141, 143
Fuzzy sets theory, 1
N S
Non-membership value, 10 Sequence of fuzzy divergence measures, 10,
Non-specicity, 10, 123 18, 71
Strategic decision making, 15, 18, 53, 62, 64,
P 66, 68, 143
Parametric generalizations, 18, 23, 51
Pattern recognition, 1, 6, 1113, 16, 18, 23, 71, U
88, 92, 93, 100, 101, 104, 119, 123, Uncertainty and vagueness, 15
132134, 141, 143, 144
Probabilistic entropy, 7 V
Probability distribution, 1, 6, 7, 23, 25, 36 Vague sets, 11, 12, 123, 132