Timezone: »
Poster
Communication trade-offs for Local-SGD with large step size
Aymeric Dieuleveut · Kumar Kshitij Patel
Wed Dec 11 05:00 PM -- 07:00 PM (PST) @ East Exhibition Hall B + C #155
Synchronous mini-batch SGD is state-of-the-art for large-scale distributed machine learning. However, in practice, its convergence is bottlenecked by slow communication rounds between worker nodes. A natural solution to reduce communication is to use the \emph{``local-SGD''} model in which the workers train their model independently and synchronize every once in a while. This algorithm improves the computation-communication trade-off but its convergence is not understood very well. We propose a non-asymptotic error analysis, which enables comparison to \emph{one-shot averaging} i.e., a single communication round among independent workers, and \emph{mini-batch averaging} i.e., communicating at every step. We also provide adaptive lower bounds on the communication frequency for large step-sizes ($ t^{-\alpha} $, $ \alpha\in (1/2 , 1 ) $) and show that \emph{Local-SGD} reduces communication by a factor of $O\Big(\frac{\sqrt{T}}{P^{3/2}}\Big)$, with $T$ the total number of gradients and $P$ machines.
Author Information
Aymeric Dieuleveut (Ecole Polytechnique, IPParis)
Kumar Kshitij Patel (Toyota Technological Institute at Chicago)
More from the Same Authors
-
2022 : Distributed Online and Bandit Convex Optimization »
Kumar Kshitij Patel · Aadirupa Saha · Nati Srebro · Lingxiao Wang -
2022 : On Convexity and Linear Mode Connectivity in Neural Networks »
David Yunis · Kumar Kshitij Patel · Pedro Savarese · Gal Vardi · Jonathan Frankle · Matthew Walter · Karen Livescu · Michael Maire -
2022 : Quadratic minimization: from conjugate gradients to an adaptive heavy-ball method with Polyak step-sizes »
Baptiste Goujaud · Adrien Taylor · Aymeric Dieuleveut -
2022 : Poster Session 2 »
Jinwuk Seok · Bo Liu · Ryotaro Mitsuboshi · David Martinez-Rubio · Weiqiang Zheng · Ilgee Hong · Chen Fan · Kazusato Oko · Bo Tang · Miao Cheng · Aaron Defazio · Tim G. J. Rudner · Gabriele Farina · Vishwak Srinivasan · Ruichen Jiang · Peng Wang · Jane Lee · Nathan Wycoff · Nikhil Ghosh · Yinbin Han · David Mueller · Liu Yang · Amrutha Varshini Ramesh · Siqi Zhang · Kaifeng Lyu · David Yunis · Kumar Kshitij Patel · Fangshuo Liao · Dmitrii Avdiukhin · Xiang Li · Sattar Vakili · Jiaxin Shi -
2022 Poster: Towards Optimal Communication Complexity in Distributed Non-Convex Optimization »
Kumar Kshitij Patel · Lingxiao Wang · Blake Woodworth · Brian Bullins · Nati Srebro -
2022 Poster: FLamby: Datasets and Benchmarks for Cross-Silo Federated Learning in Realistic Healthcare Settings »
Jean Ogier du Terrail · Samy-Safwan Ayed · Edwige Cyffers · Felix Grimberg · Chaoyang He · Regis Loeb · Paul Mangold · Tanguy Marchand · Othmane Marfoq · Erum Mushtaq · Boris Muzellec · Constantin Philippenko · Santiago Silva · Maria Teleńczuk · Shadi Albarqouni · Salman Avestimehr · Aurélien Bellet · Aymeric Dieuleveut · Martin Jaggi · Sai Praneeth Karimireddy · Marco Lorenzi · Giovanni Neglia · Marc Tommasi · Mathieu Andreux -
2021 Poster: Federated-EM with heterogeneity mitigation and variance reduction »
Aymeric Dieuleveut · Gersende Fort · Eric Moulines · Geneviève Robin -
2021 Poster: Preserved central model for faster bidirectional compression in distributed settings »
Constantin Philippenko · Aymeric Dieuleveut -
2021 Poster: A Stochastic Newton Algorithm for Distributed Convex Optimization »
Brian Bullins · Kshitij Patel · Ohad Shamir · Nathan Srebro · Blake Woodworth -
2020 Poster: Debiasing Averaged Stochastic Gradient Descent to handle missing values »
Aude Sportisse · Claire Boyer · Aymeric Dieuleveut · Julie Josse -
2020 Poster: Minibatch vs Local SGD for Heterogeneous Distributed Learning »
Blake Woodworth · Kumar Kshitij Patel · Nati Srebro -
2019 Poster: Unsupervised Scalable Representation Learning for Multivariate Time Series »
Jean-Yves Franceschi · Aymeric Dieuleveut · Martin Jaggi