Intrusion Detection System Using Modified Support Vector Machine

In network security, intrusion detection system plays an important role as it is able to detect various types of attacks in the network. The main idea of the intrusion detection system is to recognize the malicious attacks which intimidate the security from the information system’s normal activities. The intrusion detection system can be formulated basically as a problem of binary classification, so that it can be solved using effective classification technique. Support Vector Machine (SVM) is the most prominent classification algorithms in the area of data mining, but it has limitation such as extensive training time. To rectify this limitation, a modified version of SVM is introduced in this work. In this work, classification is done using modified SVM and evaluation of the proposed method is done using KDD dataset by conducting experiments. The experimental result proved that the extensive time is reduced using modified SVM by performing proper dataset.