Network Intrusion Detection Method Based on AO-BP Framework.

Abstract

Aiming at the low detection rate of network intrusion detection, a intrusion detection framework AO-BP is presented. It combines feature selection with artificial neural network. Firstly, SMOTE technology and random sampling technology are adopted to equalize data. Secondly, applying crucial features deal with data dimension reduction with the integration method in internet intrusion data. Finally, a classified experiment on the intrusion data is conducted using the optimized BP neural network. The experiment results express that the presented model shorten modeling time of the traditional BP neural network. It increases the detection accuracy of U2R and R2L. Compared with the SVM and NaiveBayes classification methods, experiments prove that the suggested method also has a highest accuracy, precision and recall.