Authentication using Dynamics Keystrokes and Quantum Machine Learning

Abstract

Authenticating a user based on his/her typing pattern is known as the keystroke dynamics. Here, Authentication is based on user typing data and user typing data cannot be copy by anyone including machine. Authenticating a user at the time of login is called static keystroke dynamics, whereas authenticating a user after login is called free text authentication. To date, free-text/dynamics/continuous authentication statistical and classical machine learning algorithms have been used. However, in this study, we solve the problem of authentication using classical, and quantum algorithms. The given dataset contained two types of information 1) text and 2) typing rhythm. We use typing rhythm data to solve the authentication problem. Out of classical, and quantum machine learning algorithms, the best performance was achieved by the QSVM algorithms. QSVM is able to solve with 100% accuracy and 0% EER. Among classical fusion of MLP and RF is able to solve the problem with 99.3% accuracy and EER of 0.007%.