Few-Shot Continuous Authentication for Mobile-Based Biometrics

The rapid growth of smartphone financial services raises the need for secure mobile authentication. Continuous authentication is a user-friendly way to strengthen security smartphones by implicitly monitoring user’s identity through sessions. Mobile continuous can be viewed as an anomaly detection problem in which models discriminate between one genuine user and rest impostors (anomalies). In practice, complete impostor profiles are hardly available due time monetary cost, while leveraging data alone yields poorly generalized lack knowledge about impostors. To address this challenge, we recast few-shot problem, aiming enhance inference robustness unseen using partial profiles. Specifically, propose novel deep learning-based model, namely local feature pooling-based temporal convolution network (LFP-TCN), directly raw sequential data, aggregating global information. addition, introduce random pattern mixing augmentation generate class-unconstrained training. augmented pool enables characterizing various patterns from limited data. Finally, demonstrate practical score-level fusion, prevents long-term dependency or increased model complexity extended re-authentication time. Experiments on two public benchmark datasets show effectiveness our method its state-of-the-art performance.