Asynchronous Federated Optimization
This addresses the need for more efficient and scalable training on edge devices in federated learning, representing an incremental improvement over existing methods.
The paper tackles the problem of improving flexibility and scalability in federated learning by proposing a new asynchronous federated optimization algorithm, achieving near-linear convergence to a global optimum for strongly convex and some non-convex problems, with empirical results showing quick convergence and tolerance to staleness.
Federated learning enables training on a massive number of edge devices. To improve flexibility and scalability, we propose a new asynchronous federated optimization algorithm. We prove that the proposed approach has near-linear convergence to a global optimum, for both strongly convex and a restricted family of non-convex problems. Empirical results show that the proposed algorithm converges quickly and tolerates staleness in various applications.