NON-PARAMETRIC ESTIMATION OF THE SEPARATING SURFACE EQUATION UNDER LARGE SAMPLES AND ITS PROPERTIES

Authors

A.V. Lapko
Lapko V.A. Institute of Computational Modeling SB RAS, Russia

Keywords:

dividing surface, training sample, function, pattern recognition problem, blur coefficient, traditional nonparametric estimation

Abstract

The computational efficiency of nonparametric pattern recognition algorithms, which are based on estimates of the probability density of the Rosenblatt-Parzen type, decreases under conditions of large training samples

References

Parzen E. On estimation of a probability density function and mode // Ann. Math. Statistic, 1962, Vol.33. – p. 1065-1076.

Lapko A.V., Lapko V.A. Nonparametric Рattern Recognition Systems Based on Learning a Sample Decomposition by Its Dimension // Рattern recognition and image analysis, 2009. – Vol. 19. - №2. – P. 296 - 302.

Лапко А.В., Лапко В.А. Коллектив непараметрических решающих функций в двуальтернативной задаче распознавания образов // Системы управления и информационные технологии, 2009. – 3.1(37). – С. 156 – 160.