PhD defense of Chao Feng
Congratulations to Chao Feng for defending his PhD thesis “Towards Robust and Sustainable Decentralized Federated Learning” 🎉
In his dissertation, Chao Feng investigates how Decentralized Federated Learning (DFL) can be made more robust and sustainable when machine-learning models are trained collaboratively without sharing private data or relying on a central coordinator.
His work examines how poisoning attacks, network topology, data heterogeneity, hardware, communication, and energy consumption affect decentralized training. He develops and evaluates defense mechanisms for heterogeneous DFL environments, as well as methods for assessing and reducing energy consumption and greenhouse-gas emissions. Overall, the dissertation provides an integrated framework for designing decentralized learning systems that are both secure and environmentally responsible.
Chao was supervised by Prof. Burkhard Stiller as a PhD student in the Communication Systems Group.