Dimensional Reduction in Behavioral Biometrics Authentication System

摘要

Authentication plays important role in digital security. It confirms the identity of the user to access the application or the system. One of the ways that used for authentication system is the user's biometric. Every user will be identified by their biometric to determine the authentication of the user. User's biometrics are categorized as psychological biometric and behavioral biometric. One of user's behavior biometric that can be captured is the ways the user plays the mouse. Therefore, in this study we will use the user's behavioral using mouse to identify their identity for the authentication system. Machine learning can be used to identify the behavior of the user by the data. Although it's not easy work because besides of accuracy, one of the important things in the authentication system is how long the system identifies the user. The huge of the dimension of the data becomes a problem because it makes the authentication process gets slower. Hence, in this work, we propose PCA (Principal Component Analysis) for dimensional reduction. Then, SVM (Support Vector Machine) is used to model the data so that the system can identify the user by the model that be built. PCA has reduced the authentications time to 50%.