Continuous multibiometric authentication for online exam with machine learning

Multibiometric authentication has been received great attention over the past decades with the growing demand of a robust authentication system. Continuous authentication system verifies a user continuously once a person is login in order to prevent intruders from the impersonation. In this study, we propose a continuous multibiometric authentication system for the identification of the person during online exam using two modalities, face recognition and keystrokes. Each modality is separately processed to generate matching scores, and the fusion method is performed at the score level to improve the accuracy. The EigenFace and support vector machine (SVM) approach are applied to the facial recognition and keystrokes dynamic accordingly. The matching score calculated from each modality is combined using the classification by the decision tree with the weighted sum after the score is split into three zones of interest.