SVM Based Intrusion Detection Using Nonlinear Scaling Scheme

Intrusion is one of major security issues of internet with the rapid growth in smart and Internet of Thing (IoT) devices, and it becomes important to detect attacks and set out alarm. In this paper, Support Vector Machine (SVM) with nonlinear input data scaling scheme is proposed to detect attacks, which is different than the existing linear scaling based machine learning methods. Experiments on the NSL-KDD dataset show that the performances of the proposed method are compared favorably with existing works. The detection rate from the new method is 82.2% for binary-classification, compared to 81.2% by existing Artificial Neural Networks (ANN) based works. For multi-classification, the proposed method shows similar performances of ANN. Further more, the detection rate of Denial of Service (DoS) is 86.5%, compared to 77.7% by existing ANN based works.