Asynchronous Stochastic Gradient Descent with Variance Reduction for Non-Convex Optimization
read the original abstract
We provide the first theoretical analysis on the convergence rate of the asynchronous stochastic variance reduced gradient (SVRG) descent algorithm on non-convex optimization. Recent studies have shown that the asynchronous stochastic gradient descent (SGD) based algorithms with variance reduction converge with a linear convergent rate on convex problems. However, there is no work to analyze asynchronous SGD with variance reduction technique on non-convex problem. In this paper, we study two asynchronous parallel implementations of SVRG: one is on a distributed memory system and the other is on a shared memory system. We provide the theoretical analysis that both algorithms can obtain a convergence rate of $O(1/T)$, and linear speed up is achievable if the number of workers is upper bounded. V1,v2,v3 have been withdrawn due to reference issue, please refer the newest version v4.
This paper has not been read by Pith yet.
Forward citations
Cited by 1 Pith paper
-
Distributed Inexact Successive Convex Approximation ADMM: Analysis-Part I
The paper develops two variants of a distributed inexact SCA-ADMM algorithm and proves first-order convergence rate guarantees under mild assumptions for non-convex problems with robustness to errors and delays.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.