Online Handwriting Based Adult and Child Classification using Machine Learning Techniques

Different age group classification of handwriting such as child and adults is important in many different applications. In forensics, investigators obtain help from handwriting classification in the category of writers. There is little research conducted in this area. In this article, we proposed a novel approach to clarify changes in the development of handwritten words between adults and children and classify them based on their handwriting. In this work, we extracted various statistical and kinetic features from handwritten data and then applied sequential forward floating selection (SFFS) to select efficient features. Support vector machine (SVM) and random forest (RF) were employed to perform the classification. We developed a new dataset to evaluate our method where adult and child handwriting data was collected using a pen tablet. Each subject (child or adult) wrote the same 30 words (tasks) by using Japanese hiragana characters only. We performed classifications from three different views: i) perform classification using all tasks, ii) use information about the average of all tasks, and iii) individually each task. The results showed that the proposed approach substantially produced up to 93.0% accuracy in adults and child classification. We hope that this study provides evidence of the possibility of classifying children and adults based on handwriting data.