Generalization Properties and Implicit Regularization for Multiple Passes SGM
Machine Learning
2016-05-27 v1 Machine Learning
Abstract
We study the generalization properties of stochastic gradient methods for learning with convex loss functions and linearly parameterized functions. We show that, in the absence of penalizations or constraints, the stability and approximation properties of the algorithm can be controlled by tuning either the step-size or the number of passes over the data. In this view, these parameters can be seen to control a form of implicit regularization. Numerical results complement the theoretical findings.
Cite
@article{arxiv.1605.08375,
title = {Generalization Properties and Implicit Regularization for Multiple Passes SGM},
author = {Junhong Lin and Raffaello Camoriano and Lorenzo Rosasco},
journal= {arXiv preprint arXiv:1605.08375},
year = {2016}
}
Comments
26 pages, 4 figures. To appear in ICML 2016