Gradient Coding with Dynamic Clustering for Straggler Mitigation
Abstract
In distributed synchronous gradient descent (GD) the main performance bottleneck for the per-iteration completion time is the slowest \textit{straggling} workers. To speed up GD iterations in the presence of stragglers, coded distributed computation techniques are implemented by assigning redundant computations to workers. In this paper, we propose a novel gradient coding (GC) scheme that utilizes dynamic clustering, denoted by GC-DC, to speed up the gradient calculation. Under time-correlated straggling behavior, GC-DC aims at regulating the number of straggling workers in each cluster based on the straggler behavior in the previous iteration. We numerically show that GC-DC provides significant improvements in the average completion time (of each iteration) with no increase in the communication load compared to the original GC scheme.
Cite
@article{arxiv.2011.01922,
title = {Gradient Coding with Dynamic Clustering for Straggler Mitigation},
author = {Baturalp Buyukates and Emre Ozfatura and Sennur Ulukus and Deniz Gunduz},
journal= {arXiv preprint arXiv:2011.01922},
year = {2020}
}