Research on Intrusion Detection System Neural Networks and Principal Component Analysis
Weiting Zhao · Jisuanji fangzhen · 2011
Research on the network security question of intrusion detection.According to the features of high dimensional,nonlinear and redundant of the network intrusion data,and the problems that the it is difficult to reduce the dimension and the testing rate is low in traditional methods,An intrusion detecting method is proposed based on the analysis of the main genetic neural network methods.First,the dimension of network intrusion data is reduced using the principal component analysis to eliminate the redundant information and simplify the neural network's inputs.Then,using genetic algorithm of neural network weights,the learning speed of neural network is accelerated.Finally,the neural network model is adopted to optimize the data after the principal component analysis and draw the nonlinear rule of network intrusion detection data.Through the network intrusion KDD CUP 99 algorithm for data collection verification experiment,experimental results showed that,compared with other network intrusion detection methods,this method is fast in learning,and has high detection accuracy and low fail and error rate.It is a kind of efficient and real good network intrusion detection method.