To search, Click below search items.


All Published Papers Search Service


On the three levels security policy comparison between PCA and SVM


A. Radi, A. Kartit, B. Regragui, M. El Marraki, D. Aboutajdine, A. Ramrami


Vol. 11  No. 2  pp. 69-79


The omnipresence of the computer system tools intensified every year in all companies. They integrate equipments, data and services that constitute a wealth to protect. A lot of mechanisms have been developed to assure the computer systems security. Conventional intrusions detection systems “IDS” have shown their insufficiencies and limits. In the previous articles, we have proposed an exact algorithm for the deployment of security policies for single computer systems [1] and an enhanced three levels security policy for complex computer systems [2] to improve computer systems security approach. However, manual analysis of the huge volume of data generated and audit data are usually impractical. To overcome this problem and evaluate our system proposed in [2], we use Support Vector Machines (SVM) which becomes one of the most important techniques for anomaly intrusion detection due to their good generalization nature and the ability to overcome the curse of dimensionality [3, 4] with applications involve large number of events as well as large number of features. Experimental analysis and comparison show that the proposed system in [2] outperformed other recent systems [5, 6] in precision, computation time, false positive and false negative rate.


Intrusions detection, Security policy, Support vector machine, Principal component, Classification