Asynchronous SGD on Graphs: a Unified Framework for Asynchronous Decentralized and Federated Optimization
This work addresses the problem of communication bottlenecks in distributed optimization for researchers and practitioners, offering a general solution that is incremental in nature by unifying existing approaches.
The paper tackles the challenge of combining decentralized and asynchronous communication techniques in distributed machine learning by introducing Asynchronous SGD on Graphs (AGRAF SGD), a unified framework that covers algorithms like SGD, Decentralized SGD, Local SGD, and FedBuff, and provides convergence rates under milder assumptions while recovering or improving upon best-known results.
Decentralized and asynchronous communications are two popular techniques to speedup communication complexity of distributed machine learning, by respectively removing the dependency over a central orchestrator and the need for synchronization. Yet, combining these two techniques together still remains a challenge. In this paper, we take a step in this direction and introduce Asynchronous SGD on Graphs (AGRAF SGD) -- a general algorithmic framework that covers asynchronous versions of many popular algorithms including SGD, Decentralized SGD, Local SGD, FedBuff, thanks to its relaxed communication and computation assumptions. We provide rates of convergence under much milder assumptions than previous decentralized asynchronous works, while still recovering or even improving over the best know results for all the algorithms covered.