LSTM-based free-text keystroke dynamics for continuous authentication
Benoit Azanguezet Q., Junie Toukem T., Elie Fute Tagne · International Journal of Computer Applications · 2025
We live in a fully networked world, with the ability to access digital information systems anytime, anywhere, using a variety of technological devices.This raises concerns about the security of systems and the protection of users' personal information.The login/password pair is the most widely used authentication method in today's information systems.However, the system becomes vulnerable if a third party obtains this information.Keystroke dynamics can be used as an additional layer of security to continuously check whether the person using the system is legitimate.In this research, we propose a continuous authentication model that uses the temporal and textual characteristics of a user's keystroke dynamics based on a one-way LSTM.We have tested this approach on data collected from 8 users.The final result obtained by this model is promising, with an accuracy of 96.67%.