For example,Бобцов

NEURAL NETWORKS APPLICATION TO NETWORK ATTACK DISCOVERY

Annotation

A method for network attacks discovery based on the choice of useful neural network parameters that characterize abnormal traffic, is proposed. Useful parameters are obtained with ranking all parameters of the neural network according to the degree of significance for detecting each attack. The ranking is carried out according to a system of rules that account for three criteria of a neural network efficiency: the general accuracy of the parameters classification, the training time of the network, and the time of its testing. To train the neural network, the well-known NSL – KDD attack database is used, which characterizes each attack by 41 information signs. The ranking makes it possible to reduce the number of features to 10. The neural network trained on useful parameters showed a high detection rate and classification accuracy for most of the attacks under consideration.

Keywords

Постоянный URL

Articles in current issue

CONTENT © ITMO UNIVERSITY 1993-2024