Show simple item record

dc.contributor.authorKovalev, Dmitry
dc.contributor.authorGower, Robert M.
dc.contributor.authorRichtarik, Peter
dc.contributor.authorRogozin, Alexander
dc.date.accessioned2020-03-12T05:11:51Z
dc.date.available2020-03-12T05:11:51Z
dc.date.issued2020-02-26
dc.identifier.urihttp://hdl.handle.net/10754/662101
dc.description.abstractSince the late 1950's when quasi-Newton methods first appeared, they have become one of the most widely used and efficient algorithmic paradigms for unconstrained optimization. Despite their immense practical success, there is little theory that shows why these methods are so efficient. We provide a semi-local rate of convergence for the randomized BFGS method which can be significantly better than that of gradient descent, finally giving theoretical evidence supporting the superior empirical performance of the method.
dc.publisherarXiv
dc.relation.urlhttps://arxiv.org/pdf/2002.11337
dc.rightsArchived with thanks to arXiv
dc.titleFast Linear Convergence of Randomized BFGS
dc.typePreprint
dc.contributor.departmentComputer Science
dc.contributor.departmentComputer Science Program
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.eprint.versionPre-print
dc.contributor.institutionT´el´ecom ParisTech, Paris, France
dc.contributor.institutionMoscow Institute of Physics and Technology, Dolgoprudny, Russia
dc.identifier.arxivid2002.11337
kaust.personKovalev, Dmitry
kaust.personRichtarik, Peter
kaust.personRogozin, Alexander
refterms.dateFOA2020-03-12T05:12:26Z


Files in this item

Thumbnail
Name:
Preprintfile1.pdf
Size:
2.763Mb
Format:
PDF
Description:
Pre-print

This item appears in the following Collection(s)

Show simple item record