Touch based active user authentication using Deep Belief Networks and Random Forests

摘要

While mobile devices traditionally use authentication methods such as passwords that define a single point of entry, active authentication can provide greater security by continuously authenticating users while they use the device. By extracting features based on users' interaction with the touchscreen, we can distinguish between different users. In this research, we investigate the performances of Deep Belief Networks (DBN) and Random Forest (RF), a more traditional classification algorithm, to classify users using a dataset extracted from the touch patterns of 41 users. The dataset is separated into strokes, which are then grouped into sessions. The preliminary results show that DBNs are outperformed by the RF.