The Effectiveness of Artificial Rhythms and Cues in Keystroke Dynamics Based User Authentication

Abstract

In keystroke dynamics based user authentication, an access system utilizes not only a valid user’s password, but also his/her typing patterns. Although high performances in terms of FAR(False Acceptance Rate) and FRR(False Rejection Rate) have been reported, most researches used a large number of valid users’ typing patterns in order to implement complex algorithms in building a classifier[1]. However, collecting sufficient typing patterns to construct a complex classifier is practically impossible. When only a handful of typing patterns are available, the only way to compensate the lack of quantity is to improve quality. To improve the quality of typing patterns, using artificial rhythms and cues were proposed[2]. In this paper, we aim at verifying the effectiveness of artificial rhythms and cues by testing hypotheses.