Title: Threshold selection for keystroke dynamics identification system

Authors: Onsiri Silasai; Sucha Smanchat; Sirapat Boonkrong

Addresses: Faculty of Information Technology and Digital Innovation, King Mongkut's University of Technology North Bangkok, Bangkok, Thailand ' Faculty of Information Technology and Digital Innovation, King Mongkut's University of Technology North Bangkok, Bangkok, Thailand ' Institute of Digital Arts and Science, Suranaree University of Technology, Nakhon Ratchasima, Thailand

Abstract: Keystroke dynamics is the timing information captured when typing on a computer keyboard. It includes hold time and inter-key time. In authentication and identification systems, a threshold is an essential element used in the decision-making process to determine whether a user be granted access or not. Therefore, the threshold selection process is vital. In this work, 15 users were asked to type long texts using a word processor twice a day for 10 days. Two scenarios were used to determine the ability to identify users. EER and accuracy were used to confirm the result and find the most appropriate threshold. The result showed that the highest count thresholds were 0.20 and 0.30. When confirmed by using EER and accuracy, the optimal threshold is 0.20 with an EER of 0.08% and an accuracy of 87.60%. Additionally, our proposed method outperforms those that use free texts to create typing patterns.

Keywords: keystroke dynamics; threshold selection; user identification.

DOI: 10.1504/IJBM.2025.147205

International Journal of Biometrics, 2025 Vol.17 No.4, pp.387 - 405

Received: 11 Mar 2023
Accepted: 11 May 2024
Published online: 11 Jul 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article