Explainable AI(XAI) for Touch Stroke Dynamics

The proposed research aims to investigate the most important features of touch stroke dynamics for user authentication through saliency mapping and feature importance analysis. The study will initially
Carry out a literature review.
Understand - Why does touch stroke dynamics work?
Understand the application of XAI in touch stroke dynamics.
Critically evaluate XAI tools for the above purpose.
Consider appropriate datasets.

The study will focus on saliency maps to identify the most relevant regions of touch strokes for authentication. Additionally, feature importance analysis will be conducted to rank the importance of each behavioral feature. The proposed research will provide a deeper understanding of touch stroke dynamics for user authentication and can contribute to the development of more accurate and efficient biometric authentication systems.

UH will focus on joint development activities as well as run workshops with research groups working in the current topic of XAI at the University of Kent and Queen Mary University as well as online workshops with MMU colleagues.

Building trustworthy AI is recently the focus of researchers, Govenments and commercial organisation.This is particular the case with the use digital devices as mobile phones and tabs where a user can be authenticated continuously. It is vital that machine learning alogirthms in such biometrics be trusted. XAI helps to build the trust which is what the project aims to achieve.