Local SGD Converges Fast and Communicates Little

arXiv1805.09767MaRDI QIDQ6302069

Sebastian U. Stich

Publication date: 24 May 2018

Abstract: Mini-batch stochastic gradient descent (SGD) is state of the art in large scale distributed training. The scheme can reach a linear speedup with respect to the number of workers, but this is rarely seen in practice as the scheme often suffers from large network delays and bandwidth limits. To overcome this communication bottleneck recent works propose to reduce the communication frequency. An algorithm of this type is local SGD that runs SGD independently in parallel on different workers and averages the sequences only once in a while. This scheme shows promising results in practice, but eluded thorough theoretical analysis. We prove concise convergence rates for local SGD on convex problems and show that it converges at the same rate as mini-batch SGD in terms of number of evaluated gradients, that is, the scheme achieves linear speedup in the number of workers and mini-batch size. The number of communication rounds can be reduced up to a factor of T^{1/2}---where T denotes the number of total steps---compared to mini-batch SGD. This also holds for asynchronous implementations. Local SGD can also be used for large scale training of deep learning models. The results shown here aim serving as a guideline to further explore the theoretical and practical aspects of local SGD in these applications.

Has companion code repository: https://github.com/adelnabli/acid

Mathematics Subject Classification ID

Analysis of algorithms (68W40) Large-scale problems in mathematical programming (90C06) Parallel algorithms in computer science (68W10)

This page was built for publication: Local SGD Converges Fast and Communicates Little

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6302069)