A method of imbalanced traffic classification based on ensemble learning

In real environment, the protocol distribution of Network traffic is imbalance, and the generalization ability of supervised learning algorithm such as algorithm to C4.5 is poor. In order to improve the classification accuracy and stability of network traffic, a network traffic classification method based on Rotation Forest was proposed. In the method, PCA was used for feature reduction and C4.5 algorithm was used to train base classifier. The experimental results show that traffic classification method based on Rotation Forest has higher accuracy and stronger generalization ability compared with C4.5 and Bagging algorithm, and more suitable for imbalanced network traffic classification.