Integrated nonparametric estimations of probability density of stochastic processes by atomic functions

Integral non-parametric estimations of the probability density and its derivatives [5-8] found on the basis of atomic functions (AF) [1-4] and of results of works [5-7] are presented. This approach enables us to obtain smoother estimations of the probability density, which improves the efficiency of solving the problems of image classification and recognition. The mathematical apparatus of the integral nonparametric statistics enables us to estimate more precisely the sequence characteristics without a priori parametric information.