User Identification Using Neural Network Based on Japanese Flick Input on Smartphone
Toshiki Kobayashi, Ryouta Kozakai, Yuji Watanabe · 2019
This study tries to identify users based on the personality in the characteristics of single tap and flick input when they input Japanese on smartphone. So we develop a communication chat application for smartphone in order to extract 12 touch features from Japanese input operations. We perform two experiments to obtain training data and test data for 16 subjects using smartphones installed the application. For the obtained data, user identification using a neural network of TensorFlow results in the accuracy of 80.8%.