In value-based deep reinforcement learning, a pruned network is a good network

Recent work has shown that deep reinforcement learning agents have difficulty in effectively using their network parameters. We leverage prior insights into the advantages of sparse training techniques and demonstrate that gradual magnitude pruning enables value-based agents to maximize parameter ef...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Obando-Ceron, Johan, Courville, Aaron, Castro, Pablo Samuel
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science - Artificial Intelligence Computer Science - Learning
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	Obando-Ceron, Johan Courville, Aaron Castro, Pablo Samuel
description	Recent work has shown that deep reinforcement learning agents have difficulty in effectively using their network parameters. We leverage prior insights into the advantages of sparse training techniques and demonstrate that gradual magnitude pruning enables value-based agents to maximize parameter effectiveness. This results in networks that yield dramatic performance improvements over traditional networks, using only a small fraction of the full network parameters.
doi_str_mv	10.48550/arxiv.2402.12479
format	Article
fullrecord	<record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2402_12479</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2402_12479</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2402_124793</originalsourceid><addsrcrecordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjEw0jM0MjG35GTw88xTKEvMKU3VTUosTk1RSElNLVAoSs3MS8svSk7NTc0rUchJTSzKy8xL11FIVCgoKs0DqspLLSnPL8pWyCwGiqXn58NFeBhY0xJzilN5oTQ3g7yba4izhy7Y5viCoszcxKLKeJAL4sEuMCasAgBb-juz</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>In value-based deep reinforcement learning, a pruned network is a good network</title><source>arXiv.org</source><creator>Obando-Ceron, Johan ; Courville, Aaron ; Castro, Pablo Samuel</creator><creatorcontrib>Obando-Ceron, Johan ; Courville, Aaron ; Castro, Pablo Samuel</creatorcontrib><description>Recent work has shown that deep reinforcement learning agents have difficulty in effectively using their network parameters. We leverage prior insights into the advantages of sparse training techniques and demonstrate that gradual magnitude pruning enables value-based agents to maximize parameter effectiveness. This results in networks that yield dramatic performance improvements over traditional networks, using only a small fraction of the full network parameters.</description><identifier>DOI: 10.48550/arxiv.2402.12479</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Learning</subject><creationdate>2024-02</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2402.12479$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2402.12479$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Obando-Ceron, Johan</creatorcontrib><creatorcontrib>Courville, Aaron</creatorcontrib><creatorcontrib>Castro, Pablo Samuel</creatorcontrib><title>In value-based deep reinforcement learning, a pruned network is a good network</title><description>Recent work has shown that deep reinforcement learning agents have difficulty in effectively using their network parameters. We leverage prior insights into the advantages of sparse training techniques and demonstrate that gradual magnitude pruning enables value-based agents to maximize parameter effectiveness. This results in networks that yield dramatic performance improvements over traditional networks, using only a small fraction of the full network parameters.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjEw0jM0MjG35GTw88xTKEvMKU3VTUosTk1RSElNLVAoSs3MS8svSk7NTc0rUchJTSzKy8xL11FIVCgoKs0DqspLLSnPL8pWyCwGiqXn58NFeBhY0xJzilN5oTQ3g7yba4izhy7Y5viCoszcxKLKeJAL4sEuMCasAgBb-juz</recordid><startdate>20240219</startdate><enddate>20240219</enddate><creator>Obando-Ceron, Johan</creator><creator>Courville, Aaron</creator><creator>Castro, Pablo Samuel</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240219</creationdate><title>In value-based deep reinforcement learning, a pruned network is a good network</title><author>Obando-Ceron, Johan ; Courville, Aaron ; Castro, Pablo Samuel</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2402_124793</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Obando-Ceron, Johan</creatorcontrib><creatorcontrib>Courville, Aaron</creatorcontrib><creatorcontrib>Castro, Pablo Samuel</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Obando-Ceron, Johan</au><au>Courville, Aaron</au><au>Castro, Pablo Samuel</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>In value-based deep reinforcement learning, a pruned network is a good network</atitle><date>2024-02-19</date><risdate>2024</risdate><abstract>Recent work has shown that deep reinforcement learning agents have difficulty in effectively using their network parameters. We leverage prior insights into the advantages of sparse training techniques and demonstrate that gradual magnitude pruning enables value-based agents to maximize parameter effectiveness. This results in networks that yield dramatic performance improvements over traditional networks, using only a small fraction of the full network parameters.</abstract><doi>10.48550/arxiv.2402.12479</doi><oa>free_for_read</oa></addata></record>
fulltext	fulltext_linktorsrc
identifier	DOI: 10.48550/arxiv.2402.12479
ispartof
issn
language	eng
recordid	cdi_arxiv_primary_2402_12479
source	arXiv.org
subjects	Computer Science - Artificial Intelligence Computer Science - Learning
title	In value-based deep reinforcement learning, a pruned network is a good network
url	https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-05T07%3A26%3A49IST&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=In%20value-based%20deep%20reinforcement%20learning,%20a%20pruned%20network%20is%20a%20good%20network&rft.au=Obando-Ceron,%20Johan&rft.date=2024-02-19&rft_id=info:doi/10.48550/arxiv.2402.12479&rft_dat=%3Carxiv_GOX%3E2402_12479%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