User-Age Classification Using Touch Gestures on Smartphones

Abstract

In this paper we investigated the possibility of classifying users’ age-group using gesture-based features on smartphones. The features used were gesture accuracy, speed, movement time, and finger/force pressure. Nearest Neighbour classification was used to classify a given user’s age-group. The 50 participants involved in this research included 25 elderly and 25 younger users. User-dependent and user-independent age-group classification scenarios were considered. On each scenario, two types of analysis were considered; using a single-feature and combined-features to represent a user-age group. The results revealed that classification accuracy was relatively higher for the younger age group than the elderly age group. Also, a higher classification accuracy was achieved on the small smartphone than on mini-tablets. The results also showed that the classification accuracy increases when combining the gesture features in to a single representation as opposed to using a single gesture feature.