METHOD OF FAST BANDWIDTH SELECTION IN A NONPARAMETRIC CLASSIFICATION SYSTEM CORRESPONDING TO THE A POSTERIORI PROBABILITY MAXIMUM CRITERION
A. V. Lapko1,2, V. A. Lapko1,2
1Institute of Computational Modeling, Siberian Branch, Russian Academy of Sciences, Krasnoyarsk, Russia
2Reshetnev Siberian University of Science and Technology, Krasnoyarsk, Russia
Keywords: непараметрический алгоритм распознавания образов, максимум апостериорной вероятности, ядерная оценка плотности вероятности, быстрый выбор коэффициентов размытости, оценка плотности вероятности типа Розенблатта-Парзена, nonparametric algorithm of pattern recognition, a posteriori probability maximum, kernel estimate of the probability density, fast bandwidth selection, estimate of the probability density of the Parzen - Rosenblatt type
Abstract
A method of fast bandwidth selection in a nonparametric algorithm of pattern recognition corresponding to the a posteriori probability maximum criterion is proposed. The algorithm is based on the results of studying the asymptotic properties of the nonparametric estimate of the separating surface equation and probability densities in solving a two-alternative problem of pattern recognition. The proposed method is compared with the traditional approach based on minimizing the classification error probability estimate.
|
