A Recognition System Using K-nearest Neighbors (K-NN) Classification of Keyboard Dynamics Patterns

Security in access granting systems in modern life has urged administrators to rethink about present authentication methods. In this paper merit of keystroke dynamics and k-nearest neighbor (K-NN) as a stochastic classifier are examined. The keystroke patterns from a pre-collected dataset are used as a working dataset for this research. K-NN is a straightforward and easy to implement classifier which relies on the majority vote of k nearest neighbor patterns labels to the test pattern regarding the proper distance measurement criterion. A nearly promising results achieved in this research and best equal error rate equivalent to 11.76% is evaluated in k=40.

Keywords:

Access control mechanism , Biometrics , Keyboard dynamics , K-nearest neighbors (K-NN) , Equal error rate (EER) ,