Touch keystroke dynamics for demographic classification

• Published in: Pattern recognition letters 2022-06, Vol.158, p.63-70
• Main Authors: Cascone, Lucia, Nappi, Michele, Narducci, Fabio, Pero, Chiara
• Format: Article
• Language: English
• Subjects: Algorithms; Biometric recognition systems; Classification; Datasets; Demographic analysis; Demographics; Demography; Machine learning; Smartphones; Soft biometrics; System effectiveness; Touch dynamics; User experience
• ISSN: 0167-8655; 1872-7344
• DOI: 10.1016/j.patrec.2022.04.023

•Soft biometrics from smartphone touch dynamics.•Analysis from touch dynamics can classify users.•The impact of biometric biased experimentation on classification accuracy.•Results on public datasets confirm the feasibility of the proposal. Soft biometric traits are not fully distinctive in recognition tasks, but they can effectively added to biometric recognition systems to improve the overall performance. In this work, the focus is on the analysis of touch keystroke dynamics of smartphone’s users for demographic classification in age, gender and user experience. Starting from the data collected in three publicly available datasets and using traditional lightweight machine learning classification algorithms, the results reported in this work shows that an effective demographic analysis can be achieved as well as continuous authentication could be improved. Moreover, the study emphasize a critical issue affecting the experimental protocols in soft biometric analysis, discussing how sensibly the performance of a system can increase on a not wise splitting of the samples in the datasets.