An Enhanced Keystroke Biometric System and Associated Studies
Tarjani Buch, Andreea Cotoranu, Eric Jeskey, Florin Tihon, Mary Villani · 2008
The Keystroke Biometric System at Pace University identifies subjects based on long-text (about 650 keystrokes) samples. The system consistently identifies subjects using the same keyboard type (desktop or laptop) and entry mode (copy task or free text input) with degrees of accuracy ranging from 98% to 100%. The current work enhances the previously developed keystroke biometric identification system and performs additional studies. Additional identification experiments support previously documented accuracy findings, and new authentication studies indicate promise for authentication application of the system. The feature extractor component is modified to facilitate these and future experiments. Additional data samples are captured for controlled longitudinal studies to verify that reasonable accuracy can be maintained with durations of several weeks between data captures.