Cancellation of dimension of input realization vectors in neural net circuits of signal processing in short range detection systems

摘要

The problems of cancellation of the input realization vectors dimensions in the neural net circuits of signal processing in the short-range detection systems, are considered. Application of the regression and neural-net methods is substantiated; these methods allow to solve the problems under weak conditionality of the data and degeneracy of their probability distribution, as well as when there is no a priori information about mathematical expectation of the informative features.