Transfer Learning of Keystroke Dynamics for Cross-Device User Authentication
2026-08-17 • Machine Learning
Machine LearningHuman-Computer Interaction
AI summaryⓘ
The authors studied how people’s typing patterns, known as keystroke dynamics, can be used to verify a user’s identity across different devices like phones and tablets. Since typing behaviors can change between devices, they used a method called inductive transfer learning to adapt the patterns learned on one device to another. They combined this adapted information with limited new data from the second device to train a system that can tell users apart more accurately. Their tests showed that this approach reduces errors compared to previous methods.
Keystroke dynamicsBehavioral biometricsUser authenticationCross-device authenticationInductive transfer learningDistribution driftBinary classifierEqual error rateFeature extractionBBMAS dataset
Authors
Nuwan Kaluarachchi, Sevvandi Kandanaarachchi, Kristen Moore, Arathi Arakala, Conrad Sanderson
Abstract
Keystroke dynamics (typing patterns) can be used as a behavioural biometric modality for user authentication, with applications such as fraud prevention. While the modality has been shown to work well for single device authentication, its application to cross-device scenarios is more challenging. Dynamics learned on one device (eg., phone) may not be directly applicable to authentication on a secondary device with a different form factor (eg., tablet) due to changes in typing patterns that can lead to distribution drifts. To address this, we propose a cross-device user authentication system based on inductive transfer learning, where keystroke dynamics learned on one device are adapted to a secondary device. The adapted data is then combined with necessarily limited training data for the secondary device, which is used to robustly train a binary classifier. Furthermore, an extended set of keystroke features is used to better capture discriminative dynamics. Experiments on the BBMAS dataset show that proposed system achieves an equal error rate of 14.2% for the cross-device scenario, surpassing state-of-the-art methods.