Masquerade Detection System Based on Correlation Eigen Matrix and Support Vector Machine

This article presents a masquerade detection system based on correlation eigen matrix and support vector machine (SVM). The system first creates a profile defining a normal user's behavior by correlation eigen matrix, and then compares the similarity of a current behavior with the created profile to decide whether the input instance is valid user or masquerader. In order to avoid overfitting and reduce the computational burden, user behavior principal features are extracted by the PCA method. SVM is used to distinguish valid user or masquerader for user behavior after training procedure has been completed by learning. In the experiments for performance evaluation the system achieved a correct detection rate equal to 82.6% and a false detection rate equal to 3.0%, which is consistent with the best results reports in the literature for the same data set and testing paradigm