License: Creative Commons Attribution 3.0 Unported license (CC BY 3.0)
When quoting this document, please refer to the following
DOI: 10.4230/LIPIcs.ITCS.2017.3
URN: urn:nbn:de:0030-drops-81850
URL: http://dagstuhl.sunsite.rwth-aachen.de/volltexte/2017/8185/
Go to the corresponding LIPIcs Volume Portal

Allen-Zhu, Zeyuan ; Orecchia, Lorenzo

Linear Coupling: An Ultimate Unification of Gradient and Mirror Descent

pdf-format:
LIPIcs-ITCS-2017-3.pdf (0.6 MB)

Abstract

First-order methods play a central role in large-scale machine learning. Even though many variations exist, each suited to a particular problem, almost all such methods fundamentally rely on two types of algorithmic steps: gradient descent, which yields primal progress, and mirror descent, which yields dual progress.

We observe that the performances of gradient and mirror descent are complementary, so that faster algorithms can be designed by "linearly coupling" the two. We show how to reconstruct Nesterov's accelerated gradient methods using linear coupling, which gives a cleaner interpretation than Nesterov's original proofs. We also discuss the power of linear coupling by extending it to many other settings that Nesterov's methods cannot apply to.

BibTeX - Entry

@InProceedings{allenzhu_et_al:LIPIcs:2017:8185,
  author =	{Zeyuan Allen-Zhu and Lorenzo Orecchia},
  title =	{{Linear Coupling: An Ultimate Unification of Gradient and Mirror Descent}},
  booktitle =	{8th Innovations in Theoretical Computer Science Conference (ITCS 2017)},
  pages =	{3:1--3:22},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-029-3},
  ISSN =	{1868-8969},
  year =	{2017},
  volume =	{67},
  editor =	{Christos H. Papadimitriou},
  publisher =	{Schloss Dagstuhl--Leibniz-Zentrum fuer Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{http://drops.dagstuhl.de/opus/volltexte/2017/8185},
  URN =		{urn:nbn:de:0030-drops-81850},
  doi =		{10.4230/LIPIcs.ITCS.2017.3},
  annote =	{Keywords: linear coupling, gradient descent, mirror descent, acceleration}
}

Keywords: linear coupling, gradient descent, mirror descent, acceleration
Collection: 8th Innovations in Theoretical Computer Science Conference (ITCS 2017)
Issue Date: 2017
Date of publication: 28.11.2017

DROPS-Home | Fulltext Search | Imprint | Privacy Published by LZI