Applying SMOTE for a Sequential Classifiers Combination Method to Improve the Performance of Intrusion Detection System
Sornxayya Phetlasy, Satoshi Ohzahata, Celimuge Wu, Toshihito Kato · 2019
Network Intrusion Detection System must detect malicious traffic effectively to protect the network system. The classification of Intrusion Detection System (IDS) distinguishes normal and malicious traffic, however, there is false in the classification affects the performance of IDS. In order to improve the false detection, we have proposed a method of sequential classifiers combination. In the proposed method, the sequential classifiers are able to detect more malicious traffic which is undetected by the previous classifier. With combining the sequential classifiers, the sensitivity and the accuracy of the final result is improved. Although the previous proposed method outperforms the previous works, the imbalanced class dataset is not well handled. In order to improve the previous our research, Synthetic Minority Oversampling Technique (SMOTE) is applied to increase the number of minority class in training phase for building classifier model. The experimental result shows that the sequential classifiers model with SMOTE enables to detect more malicious traffic and improve the sensitivity and the accuracy.