The Application Of Ensemble Learning Based On Feature Selection In Intrusion Detection

With the rapid development of modern network and other information technology,computer network has become an essential part in the social life.However,because of the openness and sharing of the Internet itself brought convenience to human society also brought more security problems at the same time,the security problem of network suffered serious challenges.Intrusion detection technology is a new safety protection technology developed in recent years,which is a new kind of network security protection technology after the firewall,information encryption and other security equipment.With the rapid development of the field,the trend of its research need to be more intelligent and reliable,which requires the machine learning methods applied in intrusion detection.Research on intrusion detection algorithm based on machine learning,further improving the effectiveness,adaptability and efficiency of the detection algorithm,is of great significance.This paper mainly studied the machine learning method can be applied in intrusion detection related issues,such as feature selection,classification,and proposed an intrusion detection method based on feature selection ensemble learning.Ensemble learning method can effectively improve the performance of the classification model,considering how to integrate the base classifier is important.Based on the data of feature selection,produce different feature subset,on this basis,to generate a diversity training set,and thus for the base classifier integration,generate the ensemble classification model based on feature selection,as a new kind of detection method.In this paper,the main researches include the following three aspects:Firstly,compares three different measure of feature selection algorithm,experiment showed the information measure is more useful to select the features for network data,based on the result of experiment,advances a method of feature selection by information measure and shows the effectiveness of the proposed method to identify the data characteristics combining with experimental analysis;Secondly,shows the advantages of the ensemble learning method,which can be applied in intrusion detection by the analysis of the classification algorithm of machine learning,according to the principle of the ensemble learning method,proposes a kind of ensemble learning classification model by feature selection,experimental analysis shows that the method can further improve the classification accuracy of intrusion detection;Finally,designs a hybrid intrusion detection framework with the methods aforementioned,expounds the related modules of the framework,and shows the advantage of the framework,the experimental data show that the detection framework can achieves a good detection performance when have a lot unknown attacks in the network data.