ICML 2016poster85 citations
Generalization Properties and Implicit Regularization for Multiple Passes SGM
Junhong Lin, Raffaello Camoriano, Lorenzo Rosasco
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.
BibTeX
@InProceedings{pmlr-v48-lina16,
title = {Generalization Properties and Implicit Regularization for Multiple Passes SGM},
author = {Lin, Junhong and Camoriano, Raffaello and Rosasco, Lorenzo},
booktitle = {Proceedings of The 33rd International Conference on Machine Learning},
pages = {2340--2348},
year = {2016},
editor = {Balcan, Maria Florina and Weinberger, Kilian Q.},
volume = {48},
series = {Proceedings of Machine Learning Research},
address = {New York, New York, USA},
month = {20--22 Jun},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v48/lina16.pdf},
url = {https://proceedings.mlr.press/v48/lina16.html},
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.}
}