Biometric system with online adaptation based on graphs and additional data sources
User authentication can assist in the security of digital systems. Biometrics is a way to implement user authentication by utilizing unique individual characteristics. However, users’ biometric characteristics can change over time, making the biometric reference obtained during enrollment outdated. Adaptive biometric systems address this issue by employing adaptation strategies that can adjust the biometric reference over time. In the literature related to adaptive biometric systems, there are various proposed adaptation strategies. However, one of the problems encountered in the studies conducted in this research is that some samples of genuine users are not used in the adaptation of the biometric reference. This work addresses a study on biometric systems with online adaptation based on graphs and using additional data sources to incorporate more genuine samples during adaptation. Graphs had previously been used in adaptive biometric systems in an offline scenario, but the current work investigates how to perform this adaptation online with the use of additional data sources. The research focused on the keystroke dynamics biometric modality.