Abstract In response to the need for more secure authentication methods, new paradigms have been developed, such as continuous authentication (CA). We present a hybrid CA method combining ML-enhanced risk-based authentication (MLE-RBA) and keystroke dynamics. While this paper focuses on architectural design, earlier experiments indicate that MLE-RBA can significantly improve threat detection over earlier RBA approaches. The combination with keystroke biometrics is expected to further strengthen real-time detection of session hijacking attempts without compromising usability.
Matiushin et al. (Mon,) studied this question.