Minimizing Finite Sums with the Stochastic Average Gradient - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles Mathematical Programming Year : 2017

Minimizing Finite Sums with the Stochastic Average Gradient


We propose the stochastic average gradient (SAG) method for optimizing the sum of a finite number of smooth convex functions. Like stochastic gradient (SG) methods, the SAG method's iteration cost is independent of the number of terms in the sum. However, by incorporating a memory of previous gradient values the SAG method achieves a faster convergence rate than black-box SG methods. The convergence rate is improved from O(1/k^{1/2}) to O(1/k) in general, and when the sum is strongly-convex the convergence rate is improved from the sub-linear O(1/k) to a linear convergence rate of the form O(p^k) for p < 1. Further, in many cases the convergence rate of the new method is also faster than black-box deterministic gradient methods, in terms of the number of gradient evaluations. Numerical experiments indicate that the new algorithm often dramatically outperforms existing SG and deterministic gradient methods, and that the performance may be further improved through the use of non-uniform sampling strategies.
Fichier principal
Vignette du fichier
sagMP.pdf (926.31 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-00860051 , version 1 (10-09-2013)
hal-00860051 , version 2 (10-05-2016)



Mark Schmidt, Nicolas Le Roux, Francis Bach. Minimizing Finite Sums with the Stochastic Average Gradient. Mathematical Programming, 2017, 162 (1-2), pp.83-112. ⟨10.1007/s10107-016-1030-6⟩. ⟨hal-00860051v2⟩
4007 View
14628 Download



Gmail Facebook Twitter LinkedIn More