A study on user-specific threshold configuration for keystroke dynamics in the context of adaptive biometric systems

Keystroke dynamics is a biometric modality which can be applied as an additional authentication factor. Some studies have shown that keystroke data can change over time and, consequently, a biometric system which does not update the biometric reference acquired at enrolment time may face performance degradation. Adaptive biometric systems can be applied in this case and automatically adapt the biometric reference. An important aspect of these systems is how the thresholds are defined. Thresholds can be used for classification and to define which samples will be used for adaptation. A few studies have worked on how to adapt the thresholds. This paper studies user-specific thresholds for keystroke dynamics in adaptive biometric systems.