A machine learning approach for user authentication using touchstroke dynamics

Abstract

Touchstroke dynamics is an essential component of computer security. In recent years, we are heavily dependent on computers for communication, banking, security applications, and many other areas. This dependency has increased the chances of malicious attacks, so there is a need for high security to protect user’s secured data from unauthorized access. Currently, we are using PINs and passwords for access in computers, but these methods are not sufficient as the computer systems are accessed globally. So we propose a method for touchstroke dynamics in touchscreen mobile devices to improve security. The behavioral biometric gives a confidence measurement instead of accept/reject measurements. We have used an android mobile device for assessing the security using the touchstroke behavior of users. This provides us with confidence measurements for security purpose as compared to physiological biometric in which FRR/FAR cannot be changed by varying threshold at individual level.