Comparison of Learning Hierarchical Dynamic Neuro-neighborhood Models Based on Perceptron and Radial-basic Functions

This article will describe hierarchical dynamic neuro-neighborhood models. A review of previous studies in the field of hierarchical dynamic neuro-neighborhood modeling is carried out. The idea of hierarchical dynamic neuro-neighborhood models is the possibility of representing complex systems and their subprocesses using a neighborhood model with neural networks nested in its nodes. An algorithm for identifying a model of the «input-state-output» type is presented. The training of hierarchical dynamic neuro-neighborhood models based on a perceptron and a network of radial basis functions was compared.