On the Role of Data Homogeneity in Multi-Agent Non-convex Stochastic Optimization

arXiv2208.13162MaRDI QIDQ6409042FDOQ6409042

Hoi-To Wai, Qiang Li

Publication date: 28 August 2022

Abstract: This paper studies the role of data homogeneity on multi-agent optimization. Concentrating on the decentralized stochastic gradient (DSGD) algorithm, we characterize the transient time, defined as the minimum number of iterations required such that DSGD can achieve comparable performance as its centralized counterpart. When the Hessians for the objective functions are identical at different agents, we show that the transient time of DSGD is $O(n^{4/3}/h{o}^{8/3})$ for smooth (possibly non-convex) objective functions, where $n$ is the number of agents and $ho$ is the spectral gap of connectivity graph. This is improved over the bound of $O(n^2/h{o}^4)$ without the Hessian homogeneity assumption. Our analysis leverages a property that the objective function is twice continuously differentiable. Numerical experiments are presented to illustrate the essence of data homogeneity to fast convergence of DSGD.

This page was built for publication: On the Role of Data Homogeneity in Multi-Agent Non-convex Stochastic Optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6409042)