Computer Science Seminar: Gradient Leakage Defense for Federated Learning
Wednesday, 11 November 2026 at 14:00
Computer Science Seminar by Prof Xianghua Xie
Event details
Speaker: Prof Xianghua Xie, Swansea University
Title: Gradient Leakage Defense for Federated Learning
Abstract: Federated Learning (FL) is a widely adopted privacy-preserving machine learning approach where private data remains local, enabling secure computations and the exchange of local model gradients between local clients and third-party parameter servers. However, recent findings reveal that privacy may be compromised and sensitive information potentially recovered from shared gradients. In this talk, we offer detailed analysis and a novel perspective on understanding the gradient leakage problem. These theoretical works lead to a new gradient leakage defense technique that secures arbitrary model architectures using a private key-lock module. Only the locked gradient is transmitted to the parameter server for global model aggregation. Our proposed learning method is resistant to gradient leakage attacks, and the key-lock module is designed and trained to ensure that, without the private information of the key-lock module: a) reconstructing private training data from the shared gradient is infeasible; and b) the global model’s inference performance is significantly compromised. We discuss the theoretical underpinnings of why gradients can leak private information and provide theoretical proof of our method’s effectiveness. We conducted extensive empirical evaluations with many models on several popular benchmarks, demonstrating the robustness of our proposed approach in both maintaining model performance and defending against gradient leakage attacks.
Bio: Xianghua Xie is currently leading a research team on Computer Vision and Machine Learning (http://csvision.swan.ac.uk) in the Department of Computer Science, Swansea University. He is also head of the Artificial Intelligence research group and the Intelligent Robotics research group in the department. He was a recipient of an RCUK Academic Fellowship between September 2007 and March 2012. He was appointed as a Senior Lecturer from October 2012, then an Associate Professor in April 2013, and a full Professor from March 2019. Prior to his position at Swansea, He was a Research Associate at the Computer Vision Group, Department of Computer Science, University of Bristol, where he completed both his PhD (2006) and MSc (2002) degrees.
He has strong research interests in the areas of Pattern Recognition and Machine Intelligence and their applications to real-world problems. He has been an investigator on several research projects funded by external bodies, such as EPSRC, Leverhulme, NISCHR, and WORD. Among his research works, those of significant importance include detecting abnormal patterns in complex visual and medical data, assisted diagnosis using automated image analysis, fully automated volumetric image segmentation, registration, and motion analysis, machine understanding of human action, efficient deep learning, and deep learning on irregular domains. He has published over 200 fully refereed research publications and (co-)edited several conference proceedings. He is an associate editor of Pattern Recognition, Robotics, and IET Computer Vision, as well as an editorial member of a number of other international journals. He has chaired and co-chaired several international conferences, e.g. BMVC2015, BMVC2019 and the newly established AIiH conference series.
Organiser
Computer Science
Location
Innovation Centre, Flexi Space