Authentication via Keystroke Dynamics by K-means and Fractal Dimension

The majority of computer systems employ a login ID and password as the principal method for secure access. Keystroke dynamics is a behavioral biometric that is used for user authentication. Although there are some identity verification methods in keystroke dynamics, their accuracy are variable and depends on the dataset and its acquisition methodology. We present the use of a novel feature extraction process beside a fractal dimension technique to obtain higher accuracy in classifying keystroke dynamics patterns for authenticating. The keystroke patterns of the paper are selected from a dataset prepared by Killourhy. To classify keystroke patterns, K-Means algorithm is used to better distinguish patterns of different classes. The results show that the total classification rate of K-means has reached 58% on the dataset. Moreover, using fractal dimension technique beside K-means algorithm would increase the accuracy from 58% to 69.6%.

keystroke dynamics, authentication, fractal dimension, box counting method, k-means