Asynchronous Stochastic Gradient Descent with Variance Reduction for Non-Convex Optimization
This work addresses the lack of convergence guarantees for asynchronous SGD with variance reduction in non-convex settings, which is important for distributed machine learning practitioners.
The paper provides the first theoretical analysis of asynchronous stochastic variance reduced gradient (SVRG) descent for non-convex optimization, showing that two parallel implementations achieve a convergence rate of O(1/T) with linear speedup under certain conditions.
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 V1,v2,v3 have been withdrawn due to reference issue, please refer the newest version v4.