Continuous User Authentication on Touchscreen Using Behavioral Biometrics Utilizing Machine Learning Approaches
Amany Mahmoud Sarhan, Ahmed Ramadan · Advances in information security, privacy, and ethics book series · 2020
Nowadays, touchscreen mobile devices make up a larger share in the market, necessitating effective and robust methods to continuously authenticate touch-based device users. A classification framework is proposed that learns the touch behavior of a user and is able afterwards to authenticate users by monitoring their behavior in performing input touch actions. Two models of features are built; the low-level features (stoke-level) model or the high-level abstracted features (session-level) model. In building these models, two different methods for features selection and data classification were weighted features and PCA. Two classification algorithms were used; ANN and SVM. The experimental results indicate the possibility of continuous authentication for touch-input users with higher promises for session-level features than stroke-level features. Authors found out that using weighted features method and artificial neural networks in building the session-level model yields the most efficient and accurate behavioral biometric continuous user authentication.