Solving Ridge Regression using Sketched Preconditioned SVRG

Abstract

We develop a novel preconditioning method for ridge regression, based on recent linear sketching methods. By equipping Stochastic Variance Reduced Gradient (SVRG) with this preconditioning process, we obtain a significant speed-up relative to fast stochastic methods such as SVRG, SDCA and SAG.

References in corpus (3)

• Sketching as a Tool for Numerical Linear Algebra
• Randomized Block Krylov Methods for Stronger and Faster Approximate Singular Value Decomposition
• Un-regularizing: approximate proximal point and faster stochastic algorithms for empirical risk minimization