Abstract
In this paper, we present an empirical evaluation of 30 features used in touch-based continuous authentication. It is essential to identify the most significant features for each user, as behaviour is different amongst humans. Thus, a fixed feature set cannot be applied to all models. We highlight this importance by selecting features accordingly using our approach, seeking to individually select and empirically test the discriminative power of a range of features as well as feature interaction in the context of individual users. We test five different feature selection techniques: Mutual Information, Sequential Forward Selection, Sequential Floating Forward Selection, Sequential Backwards Selection, and Sequential Floating Backwards Selection. Our results show that a unique set of features can be selected for each user, while increasing or maintaining performance, i.e. up to 27 out of 30 features were removed for one user without affecting performance. We also show that distinctive features should be evaluated on a user basis, as particular features may be significant for some, while redundant for others. Moreover, for each user, the same features are selected for horizontal and vertical strokes while performance persists when using a horizontal model to predict vertical behaviour and vice versa.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.