{"references": ["A. Thomas and B. J. Oommen, \"Optimal \"Anti-Bayesian\" Parametric Pattern Classification Using Order Statistics Criteria,\" in Progress in Pat-tern Recognition, Image Analysis, Computer Vision, and Applications, vol. 7441 of Lecture Notes in Computer Science, pp. 1-13, Springer Berlin / Heidelberg, 2012. This was a Plenary/Keynote Talk at the Conference.", "A. Thomas and B. J. Oommen, \"Optimal \"Anti-Bayesian\" Parametric Pattern Classification for the Exponential Family Using Order Statistics Criteria,\" in Image Analysis and Recognition, vol. 7324 of Lecture Notes in Computer Science, pp. 11-18, Springer Berlin / Heidelberg, 2012.", "M. Ahsanullah and V. B. Nevzorov, Order Statistics: Examples and Exercises. Nova Science Publishers, Inc, 2005.", "K. W. Morris and D. Szynal, \"A goodness-of-fit for the Uniform Distribution based on a Characterization,\" Journal of Mathematical Science, vol. 106, pp. 2719-2724, 2001.", "G. D. Lin, \"Characterizations of Continuous Distributions via Expected values of two functions of Order Statistics,\" Sankhya\u2022 The Indian Journal of Statistics, vol. 52, pp. 84-90, 1990.", "Y. Too and G. D. Lin, \"Characterizations of Uniform and Exponential Distributions,\" Academia Sinica, vol. 7, no. 5, pp. 357-359, 1989.", "A. Thomas, Pattern Classification using Novel Order Statistics and Border Identification Methods. PhD thesis, School of Computer Science, Carleton University, 2012. (To be Submitted).", "A. Thomas and B. J. Oommen, \"The Fundamental Theory of Optimal \"Anti-Bayesian\" Parametric Pattern Classification Using Order Statistics Criteria,\" Pattern Recognition, vol. 46, pp. 376-388, 2013.", "A. Thomas and B. J. Oommen, \"Optimal Order Statistics-based \"Anti\u00acBayesian\" Parametric Pattern Classification for the exponential family,\" 2012. (To be submitted).\n[10]\tL. Devroye, Non-Uniform Random Variate Generation. Springer-Verlag, New York, 1986.\n[11]\tP. R. Krishnaih and M. H. Rizvi, \"A note on Moments of Gamma Order Statistics,\" Technometrics, vol. 9, pp. 315-318, 1967.\n[12]\tP. R. Tadikamalla, \"An approximation to the moments and the per\u00ac centiles of Gamma Order Statistics,\" Sankhya\u2022 The Indian Journal of Statistics, vol. 39, pp. 372-381, 1977.\n[13]\tD. H. Young, \"Moment relations for Order Statistics of the standard\u00acized Gamma distribution and the inverse multinomial distribution,\" Biometrika, vol. 58, pp. 637-640, 1971."]}

Although the field of parametric Pattern Recognition (PR) has been thoroughly studied for over five decades, the use of the Order Statistics (OS) of the distributions to achieve this has not been reported. The pioneering work on using OS for classification was presented in [1] for the Uniform distribution, where it was shown that optimal PR can be achieved in a counter-intuitive manner, diametrically opposed to the Bayesian paradigm, i.e., by comparing the testing sample to a few samples distant from the mean. This must be contrasted with the Bayesian paradigm in which, if we are allowed to compare the testing sample with only a single point in the feature space from each class, the optimal strategy would be to achieve this based on the (Mahalanobis) distance from the corresponding central points, for example, the means. In [2], we showed that the results could be extended for a few symmetric distributions within the exponential family. In this paper, we attempt to extend these results significantly by considering asymmetric distributions within the exponential family, for some of which even the closed form expressions of the cumulative distribution functions are not available. These distributions include the Rayleigh, Gamma and certain Beta distributions. As in [1] and [2], the new scheme, referred to as Classification by Moments of Order Statistics (CMOS), attains an accuracy very close to the optimal Bayes’ bound, as has been shown both theoretically and by rigorous experimental testing.