Enhancing Keystroke Biometric Authentication Using Deep Learning Techniques

Ensuring the data security and integrity of data is a critical challenge for organizations. The need for effective intrusion detection and user authentication in critical infrastructure systems has become especially relevant recently. Detecting insiders is challenging because they are difficult to detect and identify, so advanced techniques are required to prevent their activities. This paper discusses an approach that uses a Siamese neural network to authenticate users by transforming their biometric typing data or behavioral characteristics into images. To achieve this goal, the paper considers several methods for transforming numerical (tabular) keystroke data into images. The transformed keystroke data is then used to train Siamese neural networks for image similarity detection to distinguish a real user from a possible insider. The experiments conducted in this study show that transforming time series data into images is a robust and effective approach to improve the accuracy of intrusion detection and user authentication. This paper highlights the importance of comparing different methods for transforming numerical keystroke data into images to improve the accuracy of intrusion detection and user authentication. The proposed methodology can be useful for intrusion detection and user authentication, improving the security and reliability of critical infrastructure systems.