Imbalanced network traffic classification based on ensemble feature selection

Abstract

In order to improve the classification efficiency of large scale imbalanced network traffic, a classification method based on ensemble feature selection is proposed. The method firstly based on the characteristics of SU algorithm on different data sets to generate the feature subset. According to the data set of support degree and the threshold to produce integrated feature subset, based on the accuracy and recall rate, ROC area three criteria in the decision tree model compared the different feature selection methods of class effect. Experimental results show that the ensemble feature selection method in imbalanced network traffic classification performance is better than the general SU algorithm.