A Randomized Coordinate Descent Method with Volume Sampling

We analyze the coordinate descent method with a new coordinate selection strategy, called volume sampling. This strategy prescribes selecting subsets of variables of certain size proportionally to the determinants of principal submatrices of the matrix, that bounds the curvature of the objective fun...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2020-04
Hauptverfasser: Rodomanov, Anton, Kropotov, Dmitry
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Rodomanov, Anton
Kropotov, Dmitry
description We analyze the coordinate descent method with a new coordinate selection strategy, called volume sampling. This strategy prescribes selecting subsets of variables of certain size proportionally to the determinants of principal submatrices of the matrix, that bounds the curvature of the objective function. In the particular case, when the size of the subsets equals one, volume sampling coincides with the well-known strategy of sampling coordinates proportionally to their Lipschitz constants. For the coordinate descent with volume sampling, we establish the convergence rates both for convex and strongly convex problems. Our theoretical results show that, by increasing the size of the subsets, it is possible to accelerate the method up to the factor which depends on the spectral gap between the corresponding largest eigenvalues of the curvature matrix. Several numerical experiments confirm our theoretical conclusions.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2206814269</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2206814269</sourcerecordid><originalsourceid>FETCH-proquest_journals_22068142693</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwdlQISsxLyc_NrEpNUXDOzy9KycxLLElVcEktTk7NK1HwTS3JyE9RKM8syVAIy88pzU1VCE7MLcjJzEvnYWBNS8wpTuWF0twMym6uIc4eugVF-YWlqcUl8Vn5pUV5QKl4IyMDMwtDEyMzS2PiVAEAzhs23A</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2206814269</pqid></control><display><type>article</type><title>A Randomized Coordinate Descent Method with Volume Sampling</title><source>Free E- Journals</source><creator>Rodomanov, Anton ; Kropotov, Dmitry</creator><creatorcontrib>Rodomanov, Anton ; Kropotov, Dmitry</creatorcontrib><description>We analyze the coordinate descent method with a new coordinate selection strategy, called volume sampling. This strategy prescribes selecting subsets of variables of certain size proportionally to the determinants of principal submatrices of the matrix, that bounds the curvature of the objective function. In the particular case, when the size of the subsets equals one, volume sampling coincides with the well-known strategy of sampling coordinates proportionally to their Lipschitz constants. For the coordinate descent with volume sampling, we establish the convergence rates both for convex and strongly convex problems. Our theoretical results show that, by increasing the size of the subsets, it is possible to accelerate the method up to the factor which depends on the spectral gap between the corresponding largest eigenvalues of the curvature matrix. Several numerical experiments confirm our theoretical conclusions.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Convex analysis ; Curvature ; Descent ; Economic models ; Eigenvalues ; Iterative methods ; Mathematical analysis ; Matrix methods ; Randomization ; Sampling</subject><ispartof>arXiv.org, 2020-04</ispartof><rights>2020. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>780,784</link.rule.ids></links><search><creatorcontrib>Rodomanov, Anton</creatorcontrib><creatorcontrib>Kropotov, Dmitry</creatorcontrib><title>A Randomized Coordinate Descent Method with Volume Sampling</title><title>arXiv.org</title><description>We analyze the coordinate descent method with a new coordinate selection strategy, called volume sampling. This strategy prescribes selecting subsets of variables of certain size proportionally to the determinants of principal submatrices of the matrix, that bounds the curvature of the objective function. In the particular case, when the size of the subsets equals one, volume sampling coincides with the well-known strategy of sampling coordinates proportionally to their Lipschitz constants. For the coordinate descent with volume sampling, we establish the convergence rates both for convex and strongly convex problems. Our theoretical results show that, by increasing the size of the subsets, it is possible to accelerate the method up to the factor which depends on the spectral gap between the corresponding largest eigenvalues of the curvature matrix. Several numerical experiments confirm our theoretical conclusions.</description><subject>Convex analysis</subject><subject>Curvature</subject><subject>Descent</subject><subject>Economic models</subject><subject>Eigenvalues</subject><subject>Iterative methods</subject><subject>Mathematical analysis</subject><subject>Matrix methods</subject><subject>Randomization</subject><subject>Sampling</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwdlQISsxLyc_NrEpNUXDOzy9KycxLLElVcEktTk7NK1HwTS3JyE9RKM8syVAIy88pzU1VCE7MLcjJzEvnYWBNS8wpTuWF0twMym6uIc4eugVF-YWlqcUl8Vn5pUV5QKl4IyMDMwtDEyMzS2PiVAEAzhs23A</recordid><startdate>20200429</startdate><enddate>20200429</enddate><creator>Rodomanov, Anton</creator><creator>Kropotov, Dmitry</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20200429</creationdate><title>A Randomized Coordinate Descent Method with Volume Sampling</title><author>Rodomanov, Anton ; Kropotov, Dmitry</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_22068142693</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Convex analysis</topic><topic>Curvature</topic><topic>Descent</topic><topic>Economic models</topic><topic>Eigenvalues</topic><topic>Iterative methods</topic><topic>Mathematical analysis</topic><topic>Matrix methods</topic><topic>Randomization</topic><topic>Sampling</topic><toplevel>online_resources</toplevel><creatorcontrib>Rodomanov, Anton</creatorcontrib><creatorcontrib>Kropotov, Dmitry</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Rodomanov, Anton</au><au>Kropotov, Dmitry</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>A Randomized Coordinate Descent Method with Volume Sampling</atitle><jtitle>arXiv.org</jtitle><date>2020-04-29</date><risdate>2020</risdate><eissn>2331-8422</eissn><abstract>We analyze the coordinate descent method with a new coordinate selection strategy, called volume sampling. This strategy prescribes selecting subsets of variables of certain size proportionally to the determinants of principal submatrices of the matrix, that bounds the curvature of the objective function. In the particular case, when the size of the subsets equals one, volume sampling coincides with the well-known strategy of sampling coordinates proportionally to their Lipschitz constants. For the coordinate descent with volume sampling, we establish the convergence rates both for convex and strongly convex problems. Our theoretical results show that, by increasing the size of the subsets, it is possible to accelerate the method up to the factor which depends on the spectral gap between the corresponding largest eigenvalues of the curvature matrix. Several numerical experiments confirm our theoretical conclusions.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2020-04
issn 2331-8422
language eng
recordid cdi_proquest_journals_2206814269
source Free E- Journals
subjects Convex analysis
Curvature
Descent
Economic models
Eigenvalues
Iterative methods
Mathematical analysis
Matrix methods
Randomization
Sampling
title A Randomized Coordinate Descent Method with Volume Sampling
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-11T17%3A53%3A18IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=A%20Randomized%20Coordinate%20Descent%20Method%20with%20Volume%20Sampling&rft.jtitle=arXiv.org&rft.au=Rodomanov,%20Anton&rft.date=2020-04-29&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2206814269%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2206814269&rft_id=info:pmid/&rfr_iscdi=true