The resolution of signals classes mixture, described by Poly-Gaussian probability density distribution

Abstract

The paper is devoted to resolution of the signals class, at that the problem is solved for a case when classes amount and their weights are known, and mean values and variance of particular distributions are unknown. For obtain of poly-Gaussian distribution of probability density the method of maximal likelihood is used. At that likelihood function and the system of likelihood equations are obtained. Solving the equation system with iteration Newton method, we obtain numerical results, showing algorithm operability. Obtained algorithm can be used for creation of specialized computer, applied for signals class resolution. © Allerton Press, Inc., 2012.