Show simple item record

dc.contributor.authorKovalev, Dmitry
dc.contributor.authorHorvath, Samuel
dc.contributor.authorRichtarik, Peter
dc.date.accessioned2019-05-29T11:47:26Z
dc.date.available2019-05-29T11:47:26Z
dc.date.issued2019-01-24
dc.identifier.urihttp://hdl.handle.net/10754/653122
dc.description.abstractThe stochastic variance-reduced gradient method (SVRG) and its acceleratedvariant (Katyusha) have attracted enormous attention in the machine learningcommunity in the last few years due to their superior theoretical propertiesand empirical behaviour on training supervised machine learning models via theempirical risk minimization paradigm. A key structural element in both of thesemethods is the inclusion of an outer loop at the beginning of which a full passover the training data is made in order to compute the exact gradient, which isthen used to construct a variance-reduced estimator of the gradient. In thiswork we design {\em loopless variants} of both of these methods. In particular,we remove the outer loop and replace its function by a coin flip performed ineach iteration designed to trigger, with a small probability, the computationof the gradient. We prove that the new methods enjoy the same superiortheoretical convergence properties as the original methods. However, wedemonstrate through numerical experiments that our methods have substantiallysuperior practical behavior.
dc.publisherarXiv
dc.relation.urlhttps://arxiv.org/pdf/1901.08689
dc.rightsArchived with thanks to arXiv
dc.titleDon't Jump Through Hoops and Remove Those Loops: SVRG and Katyusha are Better Without the Outer Loop
dc.typePreprint
dc.contributor.departmentComputer Science
dc.contributor.departmentComputer Science Program
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.contributor.departmentStatistics
dc.contributor.departmentStatistics Program
dc.eprint.versionPre-print
dc.contributor.institutionUniversity of Edinburgh, United Kingdom
dc.contributor.institutionMoscow Institute of Physics and Technology, Russian Federation
dc.identifier.arxividarXiv:1901.08689
kaust.personKovalev, Dmitry
kaust.personHorvath, Samuel
kaust.personRichtarik, Peter
refterms.dateFOA2019-05-29T11:47:38Z


Files in this item

Thumbnail
Name:
1901.08689.pdf
Size:
1.141Mb
Format:
PDF
Description:
Preprint

This item appears in the following Collection(s)

Show simple item record