PhD Studentship: Robust, Certified, and Scalable Federated Machine Unlearning for Privacy-Preserving AI
University of Surrey Guildford £21,805 pa
Develops next-generation federated machine unlearning algorithms that deliver verifiable, robust erasure without harming model performance or participant privacy. Research spans certified unlearning with formal guarantees, robustness to adversarial relearning, federated-specific benchmarks, and scalable unlearning for foundation models, supervised by Dr Pedro Porto Buarque de Gusmao and Dr Frank Guerin.