Direct Kernel Method for Machine Learning With Support Vector Machine

Support vector machine (SVM) based intrusion detection system (IDS) presently working as the machine learning approach for classification. It helps to detect new attacks from the datasets which are used in the machine learning. At IDS, the task of the machine learning method is to construct a projectile model which can be distinguished between normal and illegitimate activity. Any IDS can be developed to get high accuracy, high detection rate and low false positive rate, which show the efficiency of that intrusion detection system. In this paper, we use a direct kernel method with SVM classifier to get the high accuracy and detection rate, also low false positive rate. For the performance evaluation of the projected system we use KDDCup99 dataset, NSL-KDD dataset and Kyoto 2006+ datasets.