Cooperative SGD: A unified Framework for the Design and Analysis of Communication-Efficient SGD Algorithms

State-of-the-art distributed machine learning suffers from significant delays due to frequent communication and synchronizing between worker nodes. Emerging communication-efficient SGD algorithms that limit synchronization between locally trained models have been shown to be effective in speeding-up distributed SGD. However, a rigorous convergence analysis and comparative study of different communication-reduction strategies remains a largely open problem. This paper presents a new framework called Coooperative SGD that subsumes existing communication-efficient SGD algorithms such as federated-averaging, elastic-averaging and decentralized SGD. By analyzing Cooperative SGD, we provide novel convergence guarantees for existing algorithms. Moreover this framework enables us to design new communication-efficient SGD algorithms that strike the best balance between reducing communication overhead and achieving fast error convergence.