Controlling Participation in Federated Learning with Feedback
Michael Cummins, Guner Dilsad Er, Michael Muehlebach
TL;DR
This work addresses the high communication cost in federated learning by introducing FedBack, an ADMM-based framework that deterministically controls per-client participation through an integral feedback mechanism. By modeling participation as a discrete-time dynamical system and applying an integral controller to adjust participation thresholds, FedBack achieves target participation rates while maintaining global convergence to stationary points. Theoretical results establish stability of the participation dynamics and global convergence under standard assumptions, and empirical results on MNIST and CIFAR-10 demonstrate substantial reductions in participation events and communication overhead with robust performance. Overall, FedBack provides a practical, scalable approach to bandwidth-constrained FL with potential extensions to other proximal and event-triggered optimization frameworks.
Abstract
We address the problem of client participation in federated learning, where traditional methods typically rely on a random selection of a small subset of clients for each training round. In contrast, we propose FedBack, a deterministic approach that leverages control-theoretic principles to manage client participation in ADMM-based federated learning. FedBack models client participation as a discrete-time dynamical system and employs an integral feedback controller to adjust each client's participation rate individually, based on the client's optimization dynamics. We provide global convergence guarantees for our approach by building on the recent federated learning research. Numerical experiments on federated image classification demonstrate that FedBack achieves up to 50\% improvement in communication and computational efficiency over algorithms that rely on a random selection of clients.