English

Improved SVRG for quadratic functions

Machine Learning 2021-06-16 v2 Computation Machine Learning

Abstract

We analyse an iterative algorithm to minimize quadratic functions whose Hessian matrix HH is the expectation of a random symmetric d×dd\times d matrix. The algorithm is a variant of the stochastic variance reduced gradient (SVRG). In several applications, including least-squares regressions, ridge regressions, linear discriminant analysis and regularized linear discriminant analysis, the running time of each iteration is proportional to dd. Under smoothness and convexity conditions, the algorithm has linear convergence. When applied to quadratic functions, our analysis improves the state-of-the-art performance of SVRG up to a logarithmic factor. Furthermore, for well-conditioned quadratic problems, our analysis improves the state-of-the-art running times of accelerated SVRG, and is better than the known matching lower bound, by a logarithmic factor. Our theoretical results are backed with numerical experiments.

Keywords

Cite

@article{arxiv.2006.01017,
  title  = {Improved SVRG for quadratic functions},
  author = {Nabil Kahale},
  journal= {arXiv preprint arXiv:2006.01017},
  year   = {2021}
}

Comments

14 pages

R2 v1 2026-06-23T15:57:55.886Z