Vibration Sensitive Keystroke Analysis
Maciej Lopatka, Maria-Hendrike Peetz, Merijn van Erp, H.P. Stehouwer, Menno M. van Zaanen · UvA-DARE (University of Amsterdam) · 2009
We present a novel method for performing non-invasive biometric analysis on habitual keystroke patterns using a vibration-based feature space. With the increasing availability of 3-D accelerometer chips in laptop computers, conventional methods using time vectors may be augmented using a distinct feature space. In these preliminary results, we demonstrate the efficacy of a vibration-based feature space for both authentication and identification paradigms. Discussion of the generation of this feature space, the learning of typing profiles, and evaluation in several experimental paradigms is provided. Several suggestions for the further development of vibration information in keystroke dynamics are also discussed.