Tadeusz Gerstenkorn , Jo a n n a Gerstenkorn *
REMARKS ON THE GENERALIZED PROBABILITY
OF THE BIFUZZY EVENT
A b str a c t. The presentation is a continuation o f a paper at M S A ’04 (T. Gerstenkorn,
J.
Gerstenkorn (2 0 0 7 )). In 1978 Ph. Sm ets proposed the so -ca lled g-p rob ab ility o f a fu zzy even t as a generalization o f the L. Z adeh’s probability o f 1968. In 1980 S. H eilp em also d iscu sscd ^-probability and analysed its properties. In 1992 Ph. Sm ets d iscussed o n ce again the sam e his ow n problem and dem onstrated its axiom atic proper ties. In this elaboration w e desire to d iscuss the g-probability o f the b ifu zzy (intuitionis- tic) event and its properties as consistent with K o lm o g o ro ff axiom atics.K ey w o rd s: b ifu zzy (intuitionistic) event, generalized-probability, fu zzy set.
I. INTRODU CTIO N
In 1965 L. Zadeh introduced the notion o f a fuzzy set as a generalization of
the Cantor’s set that was dominated to this moment in science. Conception o f the
fuzzy set allowed the mathematical modelling o f not sharp formulated notions,
usable very often in the so-called soft-sciences, as e.g. economy, humanistics,
law, medicine. The characteristics o f the fuzzy set comes after the introducing
the so-called membership function of an element x o f a considered space X to
a fuzzy set A, defined as follows:
A = {(x, /лл (х))1 x ' . x & X ) ,
(1)
where X -> [0,1], i.e. f.tA(x ) is a characteristic function o f the set A, with this
difference to the characteristic function (flA (x) o f the Cantor s set A that it can
take all the values o f the interval [0, 1] and not only the values 0 or 1.
P rofessor em eritus o f the Ł ód ź U n iv., Fac. o f M athem atics, Prof. o f U niv. ot Trade in Łódź.
Already in 1968 L. Zadeh introduced also and discussed the idea o f prob
ability o f the fuzzy event in relation to the fuzzy set. He has brought out the dif
ference between the fuzziness and accidentalness.
Assuming a probability space (.X, .с/, P), the probability o f the fuzzy event
A
e S ’/has been defined by
P ( A ) = \ n A( x ) P ( d x ),
(2)
x
where the meaning o f juA( x ) and o f the space X is as above and . с / is a ď -
Algebra.
One can easy see that the membership function
jlia(x)has replaced in (2)
the characteristic function <pA(x ) o f the normal (crisp) set A.
The considerations relating to the probability of the fuzzy event were con
tinued in following years by many authors. A review o f these problems one can
find in papers o f T. Gerstenkorn and J. Mańko (1994, 1996).
The idea o f the fuzzy set has been developed, extended and generalized for
the so-called bifuzzy set, called by K. Atanassov (1983, 1985, 1986) the in-
tuitionistic set.
The generalization consisted in introducing to considerations, besides the
membership function, also the so-called non-membership function o f the ele
ment x to the set A, i.e. one has proposed the following definition o f that set:
A = {(M
a( * ) . ( * ) ) / x : x e X } ,
(3)
where /лл ( х ) , v A( x ) : X —> [0,1] with the condition
0 < /uA( x ) + vA(x) < 1 f o r x e X .
(4)
The above statement assumes the existence o f the function
* л ( х ) = 1 - М л ( х ) - ул ( х )
called the intuitionistic index and the number я A{x) e [0,1] is treated as
a measure o f the hesitancy (indecision) connected with a valuation o f the degree
o f membership or non-membership o f the element x to the set A. Examples of
such interpretation and procedure one can find, e.g. in T. Gerstenkorn and
J. Mańko (2006a, 2006b).
Alike as for the fuzzy set, the probability o f the intuitionistic fuzzy set has
been introduced. The review o f some standpoints in this question was presented,
e.g. in T. Gerstenkorn and J. Gerstenkorn (2007) and in papers o f T. Gerstenkorn
and J. Mańko (1995-2006).
II. THE GENERALIZED PROBABILITY OF TIIE FUZZY EVENT
In 1978 Philippe Smets proposed the so-called g-probability o f the fuzzy
event as a generalization o f the probability of that event given by L. Zadeh
(1968, (5), p. 423). The definition is the following:
P { A ) = \g { /.iA( x ) ) d F ( x ) ,
(6)
x
where g is a monotonic and non-decreasing function with conditions:
g (0 ) = 0, g ( l) = 1 and F ( x ) is a distribution function in the probability space
(*, s f, P).
But two years later Stanisław Heilpem considered also the g-probability and
analysed inquiringly its properties.
In 1982 Ph. Smets came back to his considerations with the g-probability but
in this case he presented the axiomatic grounds o f its correctness and concor
dance with the axiomatics o f Kolmogoroff.
III. THE GENERALIZED PROBABILITY OF THE BIFUZZY
EVENT
Definition. Let X be any set and S f a cr -Algebra o f its bifuzzy sets in X.
Then by g-probability o f the bifuzzy event A e S t we call a non-negative func
tion Р determined on A with the values on [0,1], as follows:
P(A) = Jg C M * )-
M )P ( d x ),
(7)
where g is a monotonic, non-decreasing function with conditions. g(0,0) - 0,
£(1,1) = 1 and P i s a probabilistic measure on X.
Similarly as it was done by Ph. Smets, we show that the function (7) fulfils
the KolmogorofPs axioms.
The fulfilment o f this condition is evident in view o f the postulated condi
tions for f.i(x) and v(x).
Axiom 2: P ( X ) = \.
It suffices to notice that
jjx(
x) = 1 and vx ( x ) = 0 and to take (7).
Axiom 3:
Let
A
qX
and
B
qX
and
A n B = {<0},
then
P ( A <J B ) = P ( A ) + P ( B )
(the events A and В are bifuzzy ones and being excluded.).
We assume that:
(*) = H
a( * ) л M
bW ' thc °Peration minimum,
(8)
И
au /i(x ) = /*a(x ) v / /jv(x ) ' the °P eration maximum.
(9)
Proof. The bifuzzy events A and В arc excluded by assumption therefore the
sets A and В are disjoint. It means that Vx e X the membership function
/■1Аглв(х ) ~ 0, i.e. /.
ia(
x)
ajUB(x ) = 0, whereas the non-membership function
v ^ g i * ) has the value 1 for the product of the bifuzzy events A and B. If we
assume that the element x <£ A then we form the set X A = { x : /.iA(x) = 0 and
vA (x) = 1}. But if we take x £ B then we set up the set X B = { x : f.iB (x) = 0
and vfi(x) = l}. Let us notice that in account of V x G X A Г \ Х B it will be
/JA^ B(x) = 0
and
at
the
same
time
уж л Д * ) = 1*
Then
als0
g ( / W * ) > vA ,* (* )) = 0 - Next
there is / / ^ ( x ) = f i B( x) and
\ / x e X B there is M
a^
b(x ) = M
a(x ) and similarly for vAuB( x ) (it is evident
regarding (9)).
Therefore
P ( A ) = J g(M A( x ) , v A( x ) ) P { d x ) =
J
g (/J A( x ) , v A( x ) )P (d x ) =
X X H
=
*ЛиВ( х ) ) Р № ) ,
(/.iA( x ) > 0 only if /uB( x ) = 0 , i.e. if x e X B).
Similarly
Therefore
P ( A
k jB) = \ j + J \ ё ( / лАив(х )‘>
= P ( A ) + P (B ).
IV. T H E G E N ER A LIZED CO ND ITIO N A L PR O B A B ILITY O F TH E
BIFUZZY EVENT
If there are given two bifuzzy events A and В defined on X then we present
the g-probability o f the event A under the given event В traditionally , i.e.
Р ( А \ В ) - П *
п ^ g )- .
0 0 )
assuming the foregoing conditions on g and P ( B ) > 0 .
Proof. It is sufficient to notice that the probability P ( B) fulfils the axioms
° f Kolmogoroff, as it was shown above and in the domain o f bifuzzy sets, we
have
A n B = {(//л (л) л M
bW . v/< W v
vbW ) ' x : x e
■
R E F E R E N C E S
A tanassov K. (1 9 8 3 ), In tu itio n is tic fu z z y sets, ITKR's S cien tific S e ssio n , S ofia, June 1983. D ep o sed in Central S ei. Techn. Library o f B ulg. A cad, o f S ei. 1697 /8 4 (in B ulg).
A tanassov K ., S toeva S. (1 9 8 5 ), Intuitionistic fu zzy sets, Proc. o f the P olish Sym posium on Interval & F u zzy M athem atics, A ugust 2 6 - 2 9 , 1983. W ydaw n. P olitech n ik i P o znańskiej, eds. J. A lbrycht and H. W iśniew ski, Poznań, pp. 2 3 - 2 6 .
A tanassov K. (1 9 8 6 ), I n tu itio n is tic fu z z y sets, F u zzy Sets and S ystem s, 2 0 , 8 7 -9 6 . Gerstenkorn T ., Gerstenkorn J., (2 0 0 7 ), P ro b a b ility o f a f u z z y even t. R e v ie w o f p r o b le m s ,
2 3 rd A nnual C onference on M ultivariate Statistical A n a ly sis, M S A 2 0 0 4 , N ovem b er 8 -1 0 " ’, 2 0 0 4 , U n iversity o f Ł ódź, Poland, A cta U niversitatis L od zien sis, Folia
O econ om ica, 2 0 6 , 3 1 1 -3 1 9 . ,
Gerstenkorn T M ańko J (1 9 9 4 ), F ilo zo fia ro zm y to śc i a m a te m a ty k a lo so w o sc i, S tu d ia P h ilo s o p h ia e C h ris tia n a e (The p h ilo s o p h y o f fu z z in e s s a n d th e m a th e m a tic s o f ra n - d o m n e s s — in P olish ), Studia Philosophiae Christianae 30 (2), 83 7.
Gerstenkorn T ., M ańko J. (1 9 9 5 ), O n p r o b a b ility a n d in d e p e n d e n c e in in tu itio n istic fu z z y se t th e o r y , N o te s on Intuitionistic F u zzy Sets 1, 36—39.
Gerstenkorn T ., M ańko J. (1 9 9 6 ), F u zzin e ss a n d ra n d o m n e ss: v a rio u s c o n c e p tio n s o f p r o b a b ility . Proc. del III C ongreso Internacionál de la S o cicd a d Internacionál de
G estión у E con om ia F uzzy (SIG E F), 1 0 -1 3 N o v . 1996, B u en os A ires, Argentina. Facultad de C ien cias E con om icas, U niversidad de B u en o s A ires. Editores: Emma Fernandez Loureiro, L uiza L. Lazzari, E m ilio A .M . M achado, R o d o lfo H. Perez, A ntonio T erceno, V ol. I ll, paper 2 .4 5 , 21 pages.
Gerstenkorn T ., M ańko J. (1 9 9 8 a ), B ifu z z y p r o b a b ility o f in tu itio n is tic f u z z y sets, N otes on Intuitionistic F u zzy Sets 4 ( 1 ) , 8 -1 4 .
Gerstenkorn T ., M ańko J. (1 9 9 8 b ), A p r o b le m o f b ifu zzy p r o b a b ility o f b ifu z z y e ven ts, B U S E F A L 7 6 ,4 1 - 4 7 .
G erstenkorn T ., M ańko J. (1 9 9 9 ), R a n d o m n e s s in th e b ifu zzy s e t th e o r y , C A SY S-Intern. J. o f C om puting A nticipatory S ystem s, Ed. b y D aniel M. D u b o is, U niv. o f L ićge, B elgiu m - Third Intern. C onf. on C om puting A nticipatory S y stem s, H E C -L iôge, B elgiu m , A ugust 9 - 1 4 , 1999, Partial Proc., V ol. 7, pp. 8 9 -9 7 .
Gerstenkorn T ., M ańko J. (2 0 0 0 ), R e m a r k s on th e c la s s ic a l p r o b a b ility o f b ifu z z y events, C A SY S-Intern. J. o f C om puting A nticipatory S ystem s. Ed. b y D an iel M. D ubois, U niv. o f L ieg e, B elg iu m - Fourth Intern. C onf. on C om puting A nticipatory S y s tem s, H E C -L iege, B elg iu m , A ugust 7 - 1 2 , 2 0 0 0 , Partial Proc., V o l. 8, pp. 1 9 0 -1 9 6 . G erstenkorn T ., M ańko J. (2 0 0 1 ), O n a h e sita n c y m a rg in a n d a p r o b a b ility o f in tu itio n is
tic f u z z y e v e n ts, N o tes on Intuitionistic F uzzy Sets 7 ( 1 ) 4 - 9 .
Gerstenkorn T ., M ańko J. (2 0 0 5 ), P r o b a b ilitie s o f in tu itio n istic f u z z y se ts , First W arsaw International Sem inar on Intelligent S ystem s, M ay 2 1 , 2 0 0 4 , W arsaw -P olan d , S y s tem s R esearch Institute, Institute o f Com puter S cien ce - P olish A cad em y o f S c i en ces, Poland, Issu es in Intelligent System s-P aradigm s. Eds.: O. H ryniew icz, J. K acprzyk, J. K oronacki, S. T. Wierzchoń, Akademicka O ficyna W ydaw nicza EXIT, W arszawa 2005. In series: Problems o f present science - T heory and A pplications - Informatics (Problem y współczesnej nauki - Teoria i zastosow ania - Informatyka), pp. 6 3 -6 8 .
G erstenkorn T ., M ańko J. (2 0 0 6 a ), P ro b a b ility o f in tu itio n istic f u z z y e v e n ts w ith h e lp o f m o d a l o p era to rs, C ybernetics and S ystem s 2 0 0 6 - Proc. o f the E ighteenth European M eeting on C ybernetics and System R esearch (E M C SR 2 0 0 6 ), U n iversity o f V i enna, A pril 1 8 -2 1 , 2 0 0 6 . Ed. b y Robert Trappl, M edical U niv. o f V ienna and A u s trian S o c ie ty for C ybernetic Studies, V ol. I: M athem atical M ethods in C ybernetics and S ystem s T heory, pp. 5 2 -5 6 .
G erstenkorn T ., M ańko J. (2 0 0 6 b ), U tility a n d h e lp fu ln e s s o f p r o b a b ility o f th e fu z z y e v e n ts in s o m e e c o n o m ic p r o b le m s , C A SY S-Intern. J. o f C om puting A nticipatory System s, V o l. 18, p. 1 8 7 -1 9 5 . Ed. b y D aniel M. D u b ois, p ublished b y C H A O S 2 0 0 6 as Proc. o f C A S Y S ’0 5 , T he Seventh Intern. C onf. on C om puting A nticipatory S y s tem s, L ieg e, B elg iu m , A u gu st 8 - 1 3 , 2005.
H eilp em S. (1 9 8 0 ), W ybrane zagadnienia z teorii zb iorów rozm ytych (S elected problem s in fu zzy set theory -in P o lish ), M atem atyka Stosow ana 16, 2 7 - 3 8 .
Sm ets Ph. (1 9 7 8 ), P r o b a b ility o f a fu z z y even t: a n a x io m a tic a p p ro a c h , C olloque International sur la T heorie et les A pplications d es S o u s-E n sem b les F lou s, M arseille 1978, 1, 1 -3 .
Sm ets Ph. (1 9 8 2 ), P r o b a b ility o f a fu z z y even t: an a x io m a tic a p p ro a c h , F u zzy S ets and S y stem s 7, 1 5 3 -1 6 4 .
Zadeh L.A. (1965), Fuzzy sets, Inform. Control 8, 338-353.
Zadeh L.A. (1968), Probability measure o f fuzzy events, J. Math. Anal. Appl. 23,421-427.
Tadeusz Gerstenkorn, Joanna Gerstenkorn
UW AGI O U O G Ó LN IO N Y M PR A W D O PO D O B IEŃ STW IE ZD A R ZEN IA D W O IST O R O Z M Y T E G O
Niniejsza prezentacja jest kontynuacją pracy pt. Probability o f fu zzy event. Review
o f problems (Prawdopodobieństwo zdarzenia rozmytego. Przegląd zagadnień), przed
stawionej na WAS'05 Acta Univ. Lodz., Folia Oeconomica 2007.
W 1978 r. Philippe Smets zaproponował tzw. ^-prawdopodobieństwo zdarzenia rozmytego jako pewne uogólnienie prawdopodobieństwa tegoż zdarzenia podanego Przez Lotfi Zadeha w 1968 r.
W 1980 r. Stanisław Heilpem także rozważał g-prawdopodobieństwo i analizował jego własności.
W 1982 r. Ph. Smets ponownie i szeroko rozpatrywał ^-prawdopodobieństwo i do wodził jego aksjomatycznych własności.
W przedstawianym opracowaniu pragniemy rozpatrzyć g-prawdopodobieństwo zda rzenia dwoistorozmytego (intuicjonistycznego) i jego własności jako zgodne z aksjoma- tyką Kołmogorowa.