Dynamic Recognition for User Age-Group Classification Using Hand-Writing Based Finger on Smartphones

Abstract

The way people interact with handheld devices such as smartphone and tablet tends heavily dependent on age and experience. It can argued that the automatic identification of an age group or a level of user’s experience based on the way they are using the devices could contribute greatly to providing adaptive usage environment for each user. This study aims to investigate the effectiveness of employing the dynamic features generated by users of smartphones and tablets to automatically recognise their age group. To achieve that we created a database of over 2520 trials from 42 participants of elderly (60+) and younger users (20-39) using finger based handwriting of 10 different English words. The user recognition consists of three stages: collecting touch hand writing data, extracting features, and classification. Handwriting on touchscreen data was collected on two sizes of smartphones devices based finger. The features used were force pressure (FP), movement time (MT), and signature precision (SP). In the training dataset, the feature’s average for each trial of 6 across 10 words was calculated. A KNN classification is used to classify user age. The study considered number of users in the training dataset for 100%, 50%, and one user (i.e. 1%). The results revealed there were high classification accuracy on small smartphone compared to mini-tablet. The classification accuracy using the combined features for all users on the training dataset was (82%) on small smartphone and (77%) on mini-tablet. We found that the classification of younger users (95%) were more accurate than the elderly users (55%). The study provides an evidence of the possibility of classifying user age group based on hand writing words on touchscreen based finger.