Enhancing Technique for Intrusion Detection Using Neural Network and SVM Classifier

Information assurance and security has been a major issue of serious global concern in the wake of rapid expansion of computer systems. Intrusion Detection Systems (IDS) form a key part of system defence, where it identifies abnormal activities happening in a computer system. Different soft-computing based methods have been proposed in recent years for the development of intrusion detection systems. The proposed technique is a four step methodology of which, first step is to perform the Fuzzy C-means clustering. Then, neural network is trained, such that each of the data point is trained with the corresponding neural network associated with the cluster. Subsequently, a vector for SVM classification is formed and in the fourth step, final classification using SVM is performed to detect intrusion has happened or not. Data set used is the KDD cup 99 dataset and we have used sensitivity, specificity and accuracy as the evaluation metrics parameters. In the testing and training phase, about 27000 data points were considered each having 34 attributes. The technique yielded very good results and was compared with the other existing techniques and the comparison proved the validity of our proposed technique. It achieved about 96% accuracy in case of DOS attack and reached peaks of 99% accuracy in case of PROBE, RLA and URA attacks.