Recent Advances in the Development of a Long-Text-Input Keystroke Biometric Authentication System for Arbitrary Text Input

This study focuses on the development and evaluation of a new classification algorithm that halves the previously reported best error rate. Using keystroke data from 119 users, closed system performance was obtained as a function of the number of keystrokes per sample. The applications of interest are authenticating online student test takers and computer users in security sensitive environments. The authentication process operates on keystroke data windows as short as 1/2 minute. Performance was obtained on 119 test users compared to the previous maximum of 30. For each population size, the performance increases, and the equal error rate decreases, as the number of keystrokes per sample increases. Performance on 14, 30, and 119 users was 99.6%, 98.3%, and 96.3%, respectively, on 755-keystroke samples, indicating the potential of this approach. Because the mean population performance does not give the complete picture, the varied performance over the population of users was analyzed.