Practical User Identification for Masquerade Detection

Masquerade detection discovers suspicious activities in a computer system by creating userspsila normal profiles, then raising an alert when the audited behavior does not fit. We propose to apply the SVM algorithm to the concurrently employed patterns that have been weighted according to their frequencies in order to identify masquerading attacks. Our approach not only reduces the complexity of the system but also is more robust in controlling noisy instances of the audited behavior.