Computer network security based on Support Vector Machine approach

At present, incidents of computer networks attack are significantly increasing therefore the effective Intrusion Detection Systems (IDS) are essential for information systems security. In order for the IDS to be effective, they have the capability of detecting new arrival attacks. Additionally, the correct detection rate should also be at high level whereas the low false positive detection rate is preferred. This paper proposes the new intrusion detection technique by using hybrid methods of unsupervised/supervised learning scheme. The proposed technique integrates the Principal Component Analysis (PCA) with the Support Vector Machine (SVM). The PCA is applied to reduce high dimensional data vectors and distance between vectors including its projection onto the subspace. SVM is then used to classify different groups of data, Normal and Anomalous. The results show that the proposed technique can improve the performance of anomaly intrusion detection, the intrusion detection rate and generate fewer false alarms.