High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize

In this paper, we propose a new, simplified high probability analysis of AdaGrad for smooth, non-convex problems. More specifically, we focus on a particular accelerated gradient (AGD) template (Lan, 2020), through which we recover the original AdaGrad and its variant with averaging, and prove a con...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2022-04
Hauptverfasser: Kavis, Ali, Kfir Yehuda Levy, Cevher, Volkan
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 Kavis, Ali
Kfir Yehuda Levy
Cevher, Volkan
description In this paper, we propose a new, simplified high probability analysis of AdaGrad for smooth, non-convex problems. More specifically, we focus on a particular accelerated gradient (AGD) template (Lan, 2020), through which we recover the original AdaGrad and its variant with averaging, and prove a convergence rate of \(\mathcal O (1/ \sqrt{T})\) with high probability without the knowledge of smoothness and variance. We use a particular version of Freedman's concentration bound for martingale difference sequences (Kakade & Tewari, 2008) which enables us to achieve the best-known dependence of \(\log (1 / \delta )\) on the probability margin \(\delta\). We present our analysis in a modular way and obtain a complementary \(\mathcal O (1 / T)\) convergence rate in the deterministic setting. To the best of our knowledge, this is the first high probability result for AdaGrad with a truly adaptive scheme, i.e., completely oblivious to the knowledge of smoothness and uniform variance bound, which simultaneously has best-known dependence of \(\log( 1/ \delta)\). We further prove noise adaptation property of AdaGrad under additional noise assumptions.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2647909693</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2647909693</sourcerecordid><originalsourceid>FETCH-proquest_journals_26479096933</originalsourceid><addsrcrecordid>eNqNyr0OgjAUQOHGxESivMNNnElqiyAjEpXJaHQ3RYqWIBd7wb-nl8EHcPqGcwbMEVLOvIUvxIi5RCXnXAShmM-lw_apuVxhZzFTmalM-4YldnVOUKAFBUmliAAL2GJ9xvqhXxBXF7Smvd4Inj0Q52pjVQ6HVjdkPnrChoWqSLs_x2y6Xh2T1Gss3jtN7anEztZ9OonADyMeBZGU_11fXos-8g</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2647909693</pqid></control><display><type>article</type><title>High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize</title><source>Free E- Journals</source><creator>Kavis, Ali ; Kfir Yehuda Levy ; Cevher, Volkan</creator><creatorcontrib>Kavis, Ali ; Kfir Yehuda Levy ; Cevher, Volkan</creatorcontrib><description>In this paper, we propose a new, simplified high probability analysis of AdaGrad for smooth, non-convex problems. More specifically, we focus on a particular accelerated gradient (AGD) template (Lan, 2020), through which we recover the original AdaGrad and its variant with averaging, and prove a convergence rate of \(\mathcal O (1/ \sqrt{T})\) with high probability without the knowledge of smoothness and variance. We use a particular version of Freedman's concentration bound for martingale difference sequences (Kakade &amp; Tewari, 2008) which enables us to achieve the best-known dependence of \(\log (1 / \delta )\) on the probability margin \(\delta\). We present our analysis in a modular way and obtain a complementary \(\mathcal O (1 / T)\) convergence rate in the deterministic setting. To the best of our knowledge, this is the first high probability result for AdaGrad with a truly adaptive scheme, i.e., completely oblivious to the knowledge of smoothness and uniform variance bound, which simultaneously has best-known dependence of \(\log( 1/ \delta)\). We further prove noise adaptation property of AdaGrad under additional noise assumptions.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Algorithms ; Convergence ; Martingales ; Smoothness ; Variance</subject><ispartof>arXiv.org, 2022-04</ispartof><rights>2022. 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>776,780</link.rule.ids></links><search><creatorcontrib>Kavis, Ali</creatorcontrib><creatorcontrib>Kfir Yehuda Levy</creatorcontrib><creatorcontrib>Cevher, Volkan</creatorcontrib><title>High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize</title><title>arXiv.org</title><description>In this paper, we propose a new, simplified high probability analysis of AdaGrad for smooth, non-convex problems. More specifically, we focus on a particular accelerated gradient (AGD) template (Lan, 2020), through which we recover the original AdaGrad and its variant with averaging, and prove a convergence rate of \(\mathcal O (1/ \sqrt{T})\) with high probability without the knowledge of smoothness and variance. We use a particular version of Freedman's concentration bound for martingale difference sequences (Kakade &amp; Tewari, 2008) which enables us to achieve the best-known dependence of \(\log (1 / \delta )\) on the probability margin \(\delta\). We present our analysis in a modular way and obtain a complementary \(\mathcal O (1 / T)\) convergence rate in the deterministic setting. To the best of our knowledge, this is the first high probability result for AdaGrad with a truly adaptive scheme, i.e., completely oblivious to the knowledge of smoothness and uniform variance bound, which simultaneously has best-known dependence of \(\log( 1/ \delta)\). We further prove noise adaptation property of AdaGrad under additional noise assumptions.</description><subject>Algorithms</subject><subject>Convergence</subject><subject>Martingales</subject><subject>Smoothness</subject><subject>Variance</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNyr0OgjAUQOHGxESivMNNnElqiyAjEpXJaHQ3RYqWIBd7wb-nl8EHcPqGcwbMEVLOvIUvxIi5RCXnXAShmM-lw_apuVxhZzFTmalM-4YldnVOUKAFBUmliAAL2GJ9xvqhXxBXF7Smvd4Inj0Q52pjVQ6HVjdkPnrChoWqSLs_x2y6Xh2T1Gss3jtN7anEztZ9OonADyMeBZGU_11fXos-8g</recordid><startdate>20220406</startdate><enddate>20220406</enddate><creator>Kavis, Ali</creator><creator>Kfir Yehuda Levy</creator><creator>Cevher, Volkan</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>20220406</creationdate><title>High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize</title><author>Kavis, Ali ; Kfir Yehuda Levy ; Cevher, Volkan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26479096933</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Algorithms</topic><topic>Convergence</topic><topic>Martingales</topic><topic>Smoothness</topic><topic>Variance</topic><toplevel>online_resources</toplevel><creatorcontrib>Kavis, Ali</creatorcontrib><creatorcontrib>Kfir Yehuda Levy</creatorcontrib><creatorcontrib>Cevher, Volkan</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>Kavis, Ali</au><au>Kfir Yehuda Levy</au><au>Cevher, Volkan</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize</atitle><jtitle>arXiv.org</jtitle><date>2022-04-06</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>In this paper, we propose a new, simplified high probability analysis of AdaGrad for smooth, non-convex problems. More specifically, we focus on a particular accelerated gradient (AGD) template (Lan, 2020), through which we recover the original AdaGrad and its variant with averaging, and prove a convergence rate of \(\mathcal O (1/ \sqrt{T})\) with high probability without the knowledge of smoothness and variance. We use a particular version of Freedman's concentration bound for martingale difference sequences (Kakade &amp; Tewari, 2008) which enables us to achieve the best-known dependence of \(\log (1 / \delta )\) on the probability margin \(\delta\). We present our analysis in a modular way and obtain a complementary \(\mathcal O (1 / T)\) convergence rate in the deterministic setting. To the best of our knowledge, this is the first high probability result for AdaGrad with a truly adaptive scheme, i.e., completely oblivious to the knowledge of smoothness and uniform variance bound, which simultaneously has best-known dependence of \(\log( 1/ \delta)\). We further prove noise adaptation property of AdaGrad under additional noise assumptions.</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, 2022-04
issn 2331-8422
language eng
recordid cdi_proquest_journals_2647909693
source Free E- Journals
subjects Algorithms
Convergence
Martingales
Smoothness
Variance
title High Probability Bounds for a Class of Nonconvex Algorithms with AdaGrad Stepsize
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-28T18%3A57%3A29IST&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=High%20Probability%20Bounds%20for%20a%20Class%20of%20Nonconvex%20Algorithms%20with%20AdaGrad%20Stepsize&rft.jtitle=arXiv.org&rft.au=Kavis,%20Ali&rft.date=2022-04-06&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2647909693%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2647909693&rft_id=info:pmid/&rfr_iscdi=true