Fusion of Detectors on the Basis of Recirculation Neural Networks for Intrusion Detection

Abstract

The identification of attack class plays great role in intrusion detection. In this paper the method of recognition of a class of attack by means of the cumulative classifier with nonlinear recirculation neural networks as private detectors is described, strategy of detector selection – by a relative reconstruction error, relative cost of recognition error and mutual cost of recognition error are considered. Results of experiments are compared to results of similar researches.