Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View

Q-learning has long been one of the most popular reinforcement learning algorithms, and theoretical analysis of Q-learning has been an active research topic for decades. Although researches on asymptotic convergence analysis of Q-learning have a long tradition, non-asymptotic convergence has only re...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lim, Han-Dong, Lee, Donghwan
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Lim, Han-Dong
Lee, Donghwan
description Q-learning has long been one of the most popular reinforcement learning algorithms, and theoretical analysis of Q-learning has been an active research topic for decades. Although researches on asymptotic convergence analysis of Q-learning have a long tradition, non-asymptotic convergence has only recently come under active study. The main goal of this paper is to investigate new finite-time analysis of asynchronous Q-learning under Markovian observation models via a control system viewpoint. In particular, we introduce a discrete-time time-varying switching system model of Q-learning with diminishing step-sizes for our analysis, which significantly improves recent development of the switching system analysis with constant step-sizes, and leads to \(\mathcal{O}\left( \sqrt{\frac{\log k}{k}} \right)\) convergence rate that is comparable to or better than most of the state of the art results in the literature. In the mean while, a technique using the similarly transformation is newly applied to avoid the difficulty in the analysis posed by diminishing step-sizes. The proposed analysis brings in additional insights, covers different scenarios, and provides new simplified templates for analysis to deepen our understanding on Q-learning via its unique connection to discrete-time switching systems.
doi_str_mv 10.48550/arxiv.2207.12217
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2207_12217</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2207_12217</sourcerecordid><originalsourceid>FETCH-LOGICAL-a677-6b2e4b2ec5c80d9ce36a32758c724221edaca7c012876c6453923f81147009b73</originalsourceid><addsrcrecordid>eNotj89KxDAYxHPxIKsP4Mm8QGr-tE17LNVVYUFki9eSTb9uP2iTJemq9endXT0MAwMzzI-QO8GTtMgy_mDCN34mUnKdCCmFvib7NTqcgTU4Aa2cGZeIkfqeVnFxdgje-WOk72wEExy6PT26DgJ9xOnUi8M52c5wYFv8AdoHP9Hauzn4kTUD-AAzWvqB8HVDrnozRrj99xVp1k9N_cI2b8-vdbVhJtea5TsJ6Uk2swXvSgsqN0rqrLBapqfD0BlrtOVCFjq3eZqpUqq-ECLVnJc7rVbk_m_2QtoeAk4mLO2ZuL0Qq19zW1Fi</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View</title><source>arXiv.org</source><creator>Lim, Han-Dong ; Lee, Donghwan</creator><creatorcontrib>Lim, Han-Dong ; Lee, Donghwan</creatorcontrib><description>Q-learning has long been one of the most popular reinforcement learning algorithms, and theoretical analysis of Q-learning has been an active research topic for decades. Although researches on asymptotic convergence analysis of Q-learning have a long tradition, non-asymptotic convergence has only recently come under active study. The main goal of this paper is to investigate new finite-time analysis of asynchronous Q-learning under Markovian observation models via a control system viewpoint. In particular, we introduce a discrete-time time-varying switching system model of Q-learning with diminishing step-sizes for our analysis, which significantly improves recent development of the switching system analysis with constant step-sizes, and leads to \(\mathcal{O}\left( \sqrt{\frac{\log k}{k}} \right)\) convergence rate that is comparable to or better than most of the state of the art results in the literature. In the mean while, a technique using the similarly transformation is newly applied to avoid the difficulty in the analysis posed by diminishing step-sizes. The proposed analysis brings in additional insights, covers different scenarios, and provides new simplified templates for analysis to deepen our understanding on Q-learning via its unique connection to discrete-time switching systems.</description><identifier>DOI: 10.48550/arxiv.2207.12217</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Learning</subject><creationdate>2022-07</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,782,887</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2207.12217$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2207.12217$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Lim, Han-Dong</creatorcontrib><creatorcontrib>Lee, Donghwan</creatorcontrib><title>Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View</title><description>Q-learning has long been one of the most popular reinforcement learning algorithms, and theoretical analysis of Q-learning has been an active research topic for decades. Although researches on asymptotic convergence analysis of Q-learning have a long tradition, non-asymptotic convergence has only recently come under active study. The main goal of this paper is to investigate new finite-time analysis of asynchronous Q-learning under Markovian observation models via a control system viewpoint. In particular, we introduce a discrete-time time-varying switching system model of Q-learning with diminishing step-sizes for our analysis, which significantly improves recent development of the switching system analysis with constant step-sizes, and leads to \(\mathcal{O}\left( \sqrt{\frac{\log k}{k}} \right)\) convergence rate that is comparable to or better than most of the state of the art results in the literature. In the mean while, a technique using the similarly transformation is newly applied to avoid the difficulty in the analysis posed by diminishing step-sizes. The proposed analysis brings in additional insights, covers different scenarios, and provides new simplified templates for analysis to deepen our understanding on Q-learning via its unique connection to discrete-time switching systems.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj89KxDAYxHPxIKsP4Mm8QGr-tE17LNVVYUFki9eSTb9uP2iTJemq9endXT0MAwMzzI-QO8GTtMgy_mDCN34mUnKdCCmFvib7NTqcgTU4Aa2cGZeIkfqeVnFxdgje-WOk72wEExy6PT26DgJ9xOnUi8M52c5wYFv8AdoHP9Hauzn4kTUD-AAzWvqB8HVDrnozRrj99xVp1k9N_cI2b8-vdbVhJtea5TsJ6Uk2swXvSgsqN0rqrLBapqfD0BlrtOVCFjq3eZqpUqq-ECLVnJc7rVbk_m_2QtoeAk4mLO2ZuL0Qq19zW1Fi</recordid><startdate>20220725</startdate><enddate>20220725</enddate><creator>Lim, Han-Dong</creator><creator>Lee, Donghwan</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20220725</creationdate><title>Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View</title><author>Lim, Han-Dong ; Lee, Donghwan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a677-6b2e4b2ec5c80d9ce36a32758c724221edaca7c012876c6453923f81147009b73</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Lim, Han-Dong</creatorcontrib><creatorcontrib>Lee, Donghwan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Lim, Han-Dong</au><au>Lee, Donghwan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View</atitle><date>2022-07-25</date><risdate>2022</risdate><abstract>Q-learning has long been one of the most popular reinforcement learning algorithms, and theoretical analysis of Q-learning has been an active research topic for decades. Although researches on asymptotic convergence analysis of Q-learning have a long tradition, non-asymptotic convergence has only recently come under active study. The main goal of this paper is to investigate new finite-time analysis of asynchronous Q-learning under Markovian observation models via a control system viewpoint. In particular, we introduce a discrete-time time-varying switching system model of Q-learning with diminishing step-sizes for our analysis, which significantly improves recent development of the switching system analysis with constant step-sizes, and leads to \(\mathcal{O}\left( \sqrt{\frac{\log k}{k}} \right)\) convergence rate that is comparable to or better than most of the state of the art results in the literature. In the mean while, a technique using the similarly transformation is newly applied to avoid the difficulty in the analysis posed by diminishing step-sizes. The proposed analysis brings in additional insights, covers different scenarios, and provides new simplified templates for analysis to deepen our understanding on Q-learning via its unique connection to discrete-time switching systems.</abstract><doi>10.48550/arxiv.2207.12217</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2207.12217
ispartof
issn
language eng
recordid cdi_arxiv_primary_2207_12217
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Learning
title Finite-Time Analysis of Asynchronous Q-learning under Diminishing Step-Size from Control-Theoretic View
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-04T04%3A07%3A08IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Finite-Time%20Analysis%20of%20Asynchronous%20Q-learning%20under%20Diminishing%20Step-Size%20from%20Control-Theoretic%20View&rft.au=Lim,%20Han-Dong&rft.date=2022-07-25&rft_id=info:doi/10.48550/arxiv.2207.12217&rft_dat=%3Carxiv_GOX%3E2207_12217%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true