Doubly Accelerated Stochastic Variance Reduced Dual Averaging Method for Regularized Empirical Risk Minimization

We consider a composite convex optimization problem associated with regularized empirical risk minimization, which often arises in machine learning.