Application of hybrid neural networks for monitoring and forecasting computer networks states

Abstract

Nowadays, monitoring and forecasting computer networks states are the most important components of network administration processes. For their realization, it is necessary to use the means having high adaptability and resistance to external noise. Hybrid neural networks possess such properties. The paper considers possibilities of application of hybrid neural networks as a basis for models of monitoring and forecasting of computer networks states. Hybrid neural networks have high adaptability and resistance to external noise. Results of the experiments showed that the offered models possess rather high precision of classification of the current and predicted states of computer networks.