Communication-Efficient ADMM-based Federated Learning
This work addresses communication efficiency and computational costs for federated learning systems, representing an incremental improvement over existing methods.
The paper tackled communication and computational challenges in federated learning by proposing exact and inexact ADMM-based methods, achieving linear convergence under mild conditions like convexity-free operation and data distribution independence, with the inexact version significantly reducing computational burdens.
Federated learning has shown its advances over the last few years but is facing many challenges, such as how algorithms save communication resources, how they reduce computational costs, and whether they converge. To address these issues, this paper proposes exact and inexact ADMM-based federated learning. They are not only communication-efficient but also converge linearly under very mild conditions, such as convexity-free and irrelevance to data distributions. Moreover, the inexact version has low computational complexity, thereby alleviating the computational burdens significantly.