This Article 
 Bibliographic References 
 Add to: 
The Quality of Training Sample Estimates of the Bhattacharyya Coefficient
January 1990 (vol. 12 no. 1)
pp. 92-97

The quality, in terms of the bias and variance, of estimates of the Bhattacharyya coefficient based on n training samples from two classes described by multivariate Gaussian distributions is considered. The case where the classes are described by a common covariance matrix, as well as the case where each class is described by a different covariance matrix, is analyzed. Expressions for the bias and the variance of estimates of the Bhattacharyya coefficient are derived, and numerical examples are used to show the relationship between these parameters, the number of training samples, and the dimensionality of the observation space.

[1] S. John, "Errors in discrimination,"Ann. Math. Statist., vol. 32, pp. 1125-1144, 1961.
[2] L. N. Kanal, Ed.,Pattern Recognition. Washington, DC: Thompson, 1968.
[3] K. Fukunaga,Introduction to Statistical Pattern Recognition. New York: Academic, 1972.
[4] C. H. Chen,Statistical Pattern Recognition. Rochelle Park, NJ: Hayden, 1973.
[5] G. T. Toussaint, "An upper bound on the probability of misclassification in terms of the affinity,"Proc. IEEE, vol. 65, pp. 275-276. Feb. 1977.
[6] P. A. Devijver and J. Kittler,Pattern Recognition: A Statistical Approach, Englewood Cliffs, NJ: Prentice-Hall, 1982.
[7] T. Kailath, "The divegence and Bhattacharyya distance measures in signal detection,"IEEE Trans. Comm. Tech., vol. COM-15, pp. 52-60, Feb. 1967.
[8] K. S. Fu, P. J. Min, and T. J. Li, "Feature selection in pattern recognition."IEEE Trans. Syst. Sci. Cybern., vol. SSC-6, pp. 33-39, Jan. 1970.
[9] A. K. Jain and W. G. Waller, "On the optimal number of features in the classification of multivariate Gaussian data,"Pattern Recognition, vol. 10, pp. 365-374, 1978.
[10] I. D. Longstaff, "On extensions to Fisher's linear discriminant function,"IEEE Trans. Patt. Anal. Machine Intell., vol. PAMI-9, no. 2, pp. 321-325, Mar. 1987.
[11] D. G. Lainiotis, " A class of upper bounds on the probability of error for multihypothesis pattern recognition,"IEEE Trans. Inform. Theory, vol. IT-15, pp. 730-731, Nov. 1969.
[12] F. D. Garber and A. Djouadi, "Bounds on the Bayes classification error based on pairwise risk functions,"IEEE Trans. Pattern Anal. Machine Intell., vol. 10, pp. 281-288, Mar. 1988.
[13] F. D. Garber, N. F. Chamberlain, andÖ. Snorrason, "Time-domain and frequency-domain feature selection for reliable radar target identification," inProc. IEEE 1988 Nat. Radar Conf., Ann Arbor, MI, Apr. 1988, pp. 79-84.
[14] A. K. Jain, "On an estimate of the Bhattacharyya distance,"IEEE Trans. Syst. Man, Cybern., vol. SMC-6, pp. 763-766, Nov. 1976.
[15] C. H. Chen, "Finite sample size considerations in statistical pattern recognition," inIEEE Comput. Soc. Conf. Pattern Recognition Image Processing, May 1978, pp. 188-192.

Index Terms:
pattern recognition; statistical analysis; training sample estimates; Bhattacharyya coefficient; bias; variance; multivariate Gaussian distributions; covariance matrix; dimensionality; observation space; pattern recognition; statistical analysis
A. Djouadi, Ö. Snorrason, F.D. Garber, "The Quality of Training Sample Estimates of the Bhattacharyya Coefficient," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 12, no. 1, pp. 92-97, Jan. 1990, doi:10.1109/34.41388
Usage of this product signifies your acceptance of the Terms of Use.