Age, gender and operating-hand estimation on smart mobile devices

In recent years, studies have explored the possibility of extracting ancillary information, such as gender, age, height, weight, etc., from primary biometric traits, namely, iris, fingerprint and face, however, the estimation of soft biometrics from mobile biometric data associated with behavioral modalities, e.g., touch and phone movement, is still a much less explored area. This paper investigates the possibility to estimate soft attributes on smart mobile devices. In particular, we design a scheme to estimate age, gender, and operating-hand using information originated from keystrokes, when the user enters her secret PIN/password. The experimental analysis of the devised scheme on the publicly available keystroke dataset 'TDAS' shows promising results. The proposed method attained the highest accuracy of 87.7%, 82.8%, 95.5%, respectively, for age, gender, and operating-hand estimation via timing-based keystroke features.