– This paper aims to design a compact scheme of behavioural biometric-based user authentication, develop an adaptive mechanism that selects an appropriate classifier in an adaptive way and conduct a study to explore the effect of this mechanism.
– As a study, the proposed adaptive mechanism was implemented using a cost-based metric, which enables mobile phones to adopt a less costly classifier in an adaptive way to build the user normal-behaviour model and detect behavioural anomalies.
– The user study with 50 participants indicates that our proposed mechanism can positively affect the authentication performance by maintaining the authentication accuracy at a relatively high and stable level.
– The authentication accuracy can be further improved by incorporating other appropriate classifiers (e.g. neural networks) and considering other touch-gesture-related features (e.g. the speed of a touch).
– This work explores the effect of adaptive mechanism on behavioural biometric-based user authentication. The results should be of interest for software developers and security specialists in deciding whether to implement such a mechanism for enhancing authentication performance on mobile phones.
– The user study with 50 participants indicates that this mechanism can positively affect the authentication performance by maintaining the authentication accuracy at a relatively high and stable level. To the best of our knowledge, our work is an early work discussing the implementation of an adaptive mechanism on a mobile phone.
