Professional Documents
Culture Documents
i =1
Keywords—Entropy, Uncertainty, Fuzzy Entropy, Concavity,
Symmetry. ; α > 0, α ≠1, α≠1
And called it (α–β) fuzzy entropy which includes some
well known fuzzy entropies.
I. INTRODUCTION Rudas [16] discussed elementary entropy function based
fuzzy entropy construction and fuzzy operations. The author
I N real life situation, uncertainty arises in decision-making
problem either due to lack of knowledge or due to inherent
vagueness. Such types of problems can be solved using
defined new generalized operations as minimum and
maximum entropy operations. Guo and Xin [2] studied some
probability theory and fuzzy set theory respectively. new generalized entropy formulas for fuzzy sets and proposed
Fuzziness, a feature of imperfect information, results from the new idea for further development of fuzzy information.
lack of crisp distinction between the elements belonging and Parkash and Sharma [12] provided the applications of fuzzy
not belonging to a set. A measure of fuzziness which is often measures to prove some noiseless coding theorems. Parkash
used and citied in the literature of fuzzy information is an and Gandhi [11] used fuzzy divergence measures to generate
entropy first mentioned by Zadeh [17]. However, the two some new measures of fuzzy entropy. Parkash, Sharma and
functions measure fundamentally different types of Mahajan [12, 13] delivered the applications of weighted
uncertainty. Basically, the Shannon’s [15] entropy measures measures of fuzzy entropy for the study of maximum entropy
the average uncertainty in bits associated with the prediction principles.
of outcomes in a random experiment. Some other interesting findings related with theoretical
De Luca and Termini [2] introduced some requirements measures of fuzzy entropy and their applications have been
which capture our intuitive comprehension of the degree of investigated by Zadeh [17], Kapur [6], Gurdial, Petry and
fuzziness and consequently developed a measure of fuzzy Beaubouef [3], Pal and Bezdek [8], Hu and Yu [5] etc. In
entropy which corresponds to Shannon’s [17] probabilistic section 2, we have introduced two new trigonometric
entropy, given by: measures of fuzzy entropy and provided their applications to
(1) the field of geometry.
H ( A) = −∑ µ A ( xi ) log µ A ( xi ) + (1 − µ A ( xi ) ) log (1 − µ A ( xi ) )
n
i =1
Om Prakash is in Department of Mathematics, Guru Nanak Dev I. We, first propose a new measure of fuzzy entropy,
University, Amritsar- 143005 (India). (omparkash777@yahoo.co.in.). given by
C. P. Gandhi, is in Department of Mathematics, Guru Nanak Dev
University, Amritsar- 143005 (India). (cchanderr@yahoo.co.in).
International Scholarly and Scientific Research & Innovation 4(1) 2010 61 ISNI:0000000091950263
World Academy of Science, Engineering and Technology
International Journal of Mathematical and Computational Sciences
n
π π π (5)No:1,subtended by n sides of a convex polygon at the centre of
H1 ( A) = −∑ sin + sin − n sin 2 Vol:4, 2010
i =1 2nµA ( xi ) 2n(1 − µA ( xi )) 2n circle so t
; n >3 B1 + B2 + B3 + ... + Bn = 2π (6)
The lengths of the sides of the polygon are given by
Differentiating equation (5) w.r.t. µ A ( xi ) , we get B1 B B
∂H1 ( A) π 1 π 1 π
2a sin , 2a sin 2 ,...., 2a sin n
= cos − cos 2 2 2
∂µ A ( xi ) 2n µ A2 ( xi ) 2nµ A ( xi ) (1 − µ A ( xi ))2 2n(1 − µ A ( xi ))
Let us define the fuzzy distribution given by
π
Also µ A ( xi ) = (7)
nBi
∂ H1 ( A)
2
= Thus, the perimeter of the polygon is given by
∂µ A2 ( xi )
π π π (8)
P = 2a sin + sin + ... + sin
n µ A ( x1 ) n µ A ( x2 ) n µ A ( x )
n
1 π 1 π
4 sin + sin Since, is a measure of fuzzy entropy, its maximum value
µ ( x ) 2 n µ ( x ) (1 − µ ( x )) 4
2 n (1 − µ ( x ))
π2
A i A i A i A i
arises when the distribution is most fuzzy, that is, when
4n
2
−
2n 1 π
+
1 π µ A ( x1 ) = 1- µ A ( x1 ) = 1/2 ∀ i
cos cos
π µ A3 ( xi ) 2nµ A ( xi ) (1 − µ A ( xi ))3 2n(1 − µ A ( xi ))
Open Science Index, Mathematical and Computational Sciences Vol:4, No:1, 2010 waset.org/Publication/4944
Thus, we have
Using the relation, µ A ( xi ) = 1 − µ A ( xi ) we get:
∂ 2 H1 ( A) 1
= H1 ( A) ≤ H1
∂µ A2 ( xi ) 2
or
2n π 1 π π π π π π π
− 4
n n
International Scholarly and Scientific Research & Innovation 4(1) 2010 62 ISNI:0000000091950263
World Academy of Science, Engineering and Technology
International Journal of Mathematical and Computational Sciences
∂H 2 ( A) π π π π Vol:4, No:1, 2010n
= sec 2 − sec 2 π π
∂µ A ( xi ) 2nµ A2 ( xi ) 2nµ A ( xi ) 2n(1 − µ A ( xi )) 2 2n(1 − µ A ( xi ))
∑ tan +
2nµ A ( xi )
tan
2n(1 − µ A ( xi ))
i =1
Also
Using the fact that µ A ( xi ) = 1- µ A ( xi ) , we have
2 π π π
sec π sec2 tan
∂ 2 H 2 ( A) 2π 2nµ A ( xi ) 2nµ A ( xi ) 2nµ A ( xi )
=− +
∂µ A2 ( xi )
µ A3 ( xi ) nµ A4 ( xi )
n
π π
∑ tan 2nµ
n
≥ n tan (13)
i =1 A ( xi ) n
< 0 ∀n > 3 Multiplying both sides equation (13) by 2a and using (11),
Hence, H 2 ( A) is a concave function of µ A ( xi ) and its we get
µ A ( xi ) = 1/2 ∀ i π
maximum value arises when P ≥ 2an tan (14)
Thus, H 2 ( A) introduced above satisfies the following n
properties: Similarly, multiplying both sides equation (13) by a 2 and
using (14), we get
(i) H 2 ( A) is a concave function of µ A ( xi ) .
(ii) H 2 ( A) is an increasing function of µ A ( xi ) when π
A ≥ a 2 n tan (15)
1 n
0 ≤ µ A ( xi ) ≤
Open Science Index, Mathematical and Computational Sciences Vol:4, No:1, 2010 waset.org/Publication/4944
2
(iii) H 2 ( A) is a decreasing function of µ A ( xi ) when
III. CONCLUSION
1
≤ µ A ( xi ) ≤ 1 From equation (14), we conclude that the minimum
2 π
perimeter of the polygon of n sides is 2an tan and this
(iv) H 2 ( A) does not change when µ A ( xi ) is changed n
to 1- µ A ( xi ) . arises when the polygon is regular. Also
π
(v) H 2 ( A) = 0 when µ A ( xi ) = 0 or 1 [ P ]min = 2a n tan → 2aπ as n → ∞ which is also
n
Under the above five conditions, the measure proposed in
the circumference of the circle of radius .
equation (10) is a valid trigonometric measure of fuzzy
Also, from equation (15), we conclude that the minimum
entropy.
area of the polygon of n sides is given by
Next, we apply the trigonometric measure introduced in
π
equation (10) to the field of geometry. [ A]min = a 2 n tan → a 2π as n → ∞ which is also the
Now, if we take a polygon circumscribing a circle of radius n
a, we get P, the perimeter of polygon as area of circle of radius a .
t a n 1 + t a n 2 + . . . + t a n n
B B B
P=2a REFERENCES
2 2 2
[1] Bhandari, D. and Pal, N.R. (1993): "Some new information measures
(11) for fuzzy sets", Inform. Sci., 67, 209-228.
and area A of the polygon is given by [2] De Luca, A. and Termini, S. (1972): "A definition of non-probabilistic
entropy in setting of fuzzy set theory ", Inform. and Control, 20,
B1 B2 Bn
tan 2 + tan 2 + ... + tan 2
2 301-312.
A= a (12) [3] Ebanks, B.R.(1983): "On measures of fuzziness and thei
representations”, Jour. of Mathematical Analysis and Applications, 94,
We know that H 2 ( A) is maximum at µ A ( xi ) = 1/2 , we 24-37.
[4] Guo, X. Z. and Xin, X. L. (2006): "Some new generalized entropy
have formulas of fuzzy sets”, J. Northwest Univ.,36 (4), 529-532.
[5] Havrada, J.H. and Charvat, F. (1967): "Quantification methods of
1
H 2 ( A) ≤ H 2 classification process: Concept of structural -entropy", Kybernetika,
3, 30-35.
2 [6] Hu,Q.and Yu, D. (2004): "Entropies of fuzzy indiscernibility relation
and its operations", Internat. J. Uncertain. Fuzziness Knowledge-Based
This gives the following mathematical expression: Systems, 12 (5), 575-589.
[7] Kapur, J.N. (1997): "Measures of Fuzzy Information", Mathematical
n
π π π Sciences Trust Society, New Delhi.
−∑ tan + tan − n 2 tan [8] Klir, G.J. and Folger, T.A. (1988): "Fuzzy Sets, Uncertainty and
i =1 2nµ A ( xi ) 2n(1 − µ A ( xi )) 2n Indetermination", Prentice Hall, New York.
[9] Liu, S.T. and Kao, C. (2002): “Fuzzy measures for correlation
n
π π π
≤ −∑ tan + tan − n 2 tan
coefficient of fuzzy numbers”, Fuzzy Sets and Systems, 128, 267-275.
[10] Parkash, O. (1998): “A new parametric measure of fuzzy entropy,”
i =1 n n 2n Inform. Process. and Management of Uncertainty, 2, pp. 1732-1737.
or
International Scholarly and Scientific Research & Innovation 4(1) 2010 63 ISNI:0000000091950263
World Academy of Science, Engineering and Technology
International Journal of Mathematical and Computational Sciences
[11] Parkash, O. and Gandhi, C. P. (2005): “Generating measures of
Vol:4, No:1, 2010
fuzzy entropy through fuzzy directed divergence”, Ultra Scientist
of Physical Sci., 17(3), 419-428.
[12] Parkash, O. and Sharma, P. K. (2004): “Noiseless coding
theorems corresponding to fuzzy entropies”, Southeast Asian
Bulletin of Maths., 27, 1073-1080.
[13] Parkash, O., Sharma, P. K.and Mahajan, R (2008): “New measures of
weighted fuzzy entropy and their applications for the study of
maximum weighted fuzzy entropy principle”, Information Sciences,
178, 2389-2395.
[14] Parkash, O., Sharma, P. K. and Mahajan, R (2008): “Optimization
principle for weighted fuzzy entropy using unequal constraints”,
Southeast Asian Bulletin Maths.(Accepted).
[15] Renyi, A. (1961): "On measures of entropy and information", Proc. 4th
Ber. Symp. Math. Stat. and Prob., 1, 547-561.
[16] Rudas, I.J.(2001):“Measures of fuzziness : theory and applications ,”
Advance in fuzzy systems and evolutionary computation, 187-192,
Artfi. Intell. Series (Athens) World Science Eng. Soc. Press, Athens.
[17] Shannon, C. E. (1948):"A mathematical theory of communication”,
Bell. Sys. Tech. Jr., 27, 379-423, 623-659.
[18] Zadeh,L.A.(1968): "Probability measures of fuzzy events", Jr. Math.
Ann. Appli., 23, 421-427.
Open Science Index, Mathematical and Computational Sciences Vol:4, No:1, 2010 waset.org/Publication/4944
International Scholarly and Scientific Research & Innovation 4(1) 2010 64 ISNI:0000000091950263