Функционально связанная МГУА-вэйвлет-нейро-фаззи система и робастный алгоритм ее обучения

Abstract

In the paper the functional linked GMDH-wavelet-neuro-fuzzy system that is hybrid architecture of computational intelligence is proposed. The robust learning algorithm based on logistic optimization criterion is synthesized. Such hybrid system allows to process signals under significant level of a noise with abnormal outliers.