34
10

Order Optimal One-Shot Distributed Learning

Abstract

We consider distributed statistical optimization in one-shot setting, where there are mm machines each observing nn i.i.d. samples. Based on its observed samples, each machine then sends an O(log(mn))O(\log(mn))-length message to a server, at which a parameter minimizing an expected loss is to be estimated. We propose an algorithm called Multi-Resolution Estimator (MRE) whose expected error is no larger than O~(m1/max(d,2)n1/2)\tilde{O}\big(m^{-{1}/{\max(d,2)}} n^{-1/2}\big), where dd is the dimension of the parameter space. This error bound meets existing lower bounds up to poly-logarithmic factors, and is thereby order optimal. The expected error of MRE, unlike existing algorithms, tends to zero as the number of machines (mm) goes to infinity, even when the number of samples per machine (nn) remains upper bounded by a constant. This property of the MRE algorithm makes it applicable in new machine learning paradigms where mm is much larger than nn.

View on arXiv
Comments on this paper