Network intrusion detection by a hybrid method of rough set and RBF neural network

Abstract

In order to detect the intrusion for computer network accurately, the network intrusion detection method should be developed continuously. The hybrid method of rough set and RBF neural network is presented to network intrusion detection. The collected 390 cases in KDD-CUP99 applied to research the performance of rough set and RBF neural network compared with RBF neural network, BP neural network. The collected 390 cases include 200 normal data, 50 Probe fault data, 40 U2R fault data, 50 DoS fault data and 50 R2L fault data. It is indicated that the detection accuracies of rough set-RBF neural network are higher than those of normal RBF neural network and BP neural network.