Dr Xiyue Zhang
BSc, PhD
Expertise
I develop methods to make AI systems more trustworthy by verifying and testing that they are reliable, robust, fair, and secure before they are used in the real world.
Current positions
Lecturer
School of Computer Science
Contact
Press and media
Many of our academics speak to the media as experts in their field of research. If you are a journalist, please contact the University’s Media and PR Team:
Biography
Modern AI systems are increasingly being used in applications where failures can have significant consequences. Her research focuses on developing methods to verify and test AI systems, providing strong evidence that they behave reliably under a wide range of conditions. By combining ideas from formal verification, software testing, and machine learning, she seeks to bridge the gap between rapid advances in AI capabilities and the assurance needed for their safe and responsible deployment.
Research interests
Xiyue Zhang’s research focuses on developing trustworthy AI systems through rigorous verification and testing methods. Her work sits at the intersection of formal methods, artificial intelligence, and software engineering, aiming to improve the reliability, robustness, fairness, and security of modern AI systems. She is particularly interested in scalable assurance techniques for deep learning, foundation models, and AI-enabled intelligent systems, with applications to safety-critical and high-assurance domains.
Publications
Recent publications
01/06/2026Privacy-Preserving Robustness Verification for Neural Networks
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence
Risk-Averse Certification of Bayesian Neural Networks
PREMAP
Journal of Machine Learning Research
Protecting Deep Learning Model Copyrights With Adversarial Example-Free Reuse Detection
IEEE Transactions on Neural Networks and Learning Systems
Runtime Backdoor Detection for Federated Learning via Representational Dissimilarity Analysis
IEEE Transactions on Dependable and Secure Computing