Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation

Deep reinforcement learning (DRL) is a promising approach for adaptive robot control, but its current application to robotics is currently hindered by high sample requirements. We propose two novel data augmentation techniques for DRL in order to reuse more efficiently observed data. The first one c...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Lin, Yijiong, Huang, Jiancong, Zimmer, Matthieu, Rojas, Juan, Weng, Paul
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science - Artificial Intelligence Computer Science - Robotics
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	Lin, Yijiong Huang, Jiancong Zimmer, Matthieu Rojas, Juan Weng, Paul
description	Deep reinforcement learning (DRL) is a promising approach for adaptive robot control, but its current application to robotics is currently hindered by high sample requirements. We propose two novel data augmentation techniques for DRL in order to reuse more efficiently observed data. The first one called Kaleidoscope Experience Replay exploits reflectional symmetries, while the second called Goal-augmented Experience Replay takes advantage of lax goal definitions. Our preliminary experimental results show a large increase in learning speed.
doi_str_mv	10.48550/arxiv.1910.09959
format	Article
fullrecord	<record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1910_09959</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1910_09959</sourcerecordid><originalsourceid>FETCH-LOGICAL-a679-3ff411beebfd2f3b5a50e644a32cce411ac2a1cd437e1ca955d46617743fe43</originalsourceid><addsrcrecordid>eNotj8tOwzAQRb1hgQofwIr5gZQ4thO8rEp5SEGVSvfRxJkpIzVO5QZK_562sLrSPdKRjlJ3Op_aR-fyB0w_8j3V_nTk3jt_rZbr4YCp28P7kAg-sN9tCRbMEoRiOIJEWJFEHlKgnuIINWGKEjdwkPETnnBEmH1tzghHGeKNumLc7un2fydq9bxYz1-zevnyNp_VGZaVzwyz1bolarkr2LQOXU6ltWiKEOiEMBSoQ2dNRTqgd66zZamryhomaybq_k966Wl2SXpMx-bc1Vy6zC-l3Uln</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation</title><source>arXiv.org</source><creator>Lin, Yijiong ; Huang, Jiancong ; Zimmer, Matthieu ; Rojas, Juan ; Weng, Paul</creator><creatorcontrib>Lin, Yijiong ; Huang, Jiancong ; Zimmer, Matthieu ; Rojas, Juan ; Weng, Paul</creatorcontrib><description>Deep reinforcement learning (DRL) is a promising approach for adaptive robot control, but its current application to robotics is currently hindered by high sample requirements. We propose two novel data augmentation techniques for DRL in order to reuse more efficiently observed data. The first one called Kaleidoscope Experience Replay exploits reflectional symmetries, while the second called Goal-augmented Experience Replay takes advantage of lax goal definitions. Our preliminary experimental results show a large increase in learning speed.</description><identifier>DOI: 10.48550/arxiv.1910.09959</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Robotics</subject><creationdate>2019-10</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>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1910.09959$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1910.09959$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Lin, Yijiong</creatorcontrib><creatorcontrib>Huang, Jiancong</creatorcontrib><creatorcontrib>Zimmer, Matthieu</creatorcontrib><creatorcontrib>Rojas, Juan</creatorcontrib><creatorcontrib>Weng, Paul</creatorcontrib><title>Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation</title><description>Deep reinforcement learning (DRL) is a promising approach for adaptive robot control, but its current application to robotics is currently hindered by high sample requirements. We propose two novel data augmentation techniques for DRL in order to reuse more efficiently observed data. The first one called Kaleidoscope Experience Replay exploits reflectional symmetries, while the second called Goal-augmented Experience Replay takes advantage of lax goal definitions. Our preliminary experimental results show a large increase in learning speed.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Robotics</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj8tOwzAQRb1hgQofwIr5gZQ4thO8rEp5SEGVSvfRxJkpIzVO5QZK_562sLrSPdKRjlJ3Op_aR-fyB0w_8j3V_nTk3jt_rZbr4YCp28P7kAg-sN9tCRbMEoRiOIJEWJFEHlKgnuIINWGKEjdwkPETnnBEmH1tzghHGeKNumLc7un2fydq9bxYz1-zevnyNp_VGZaVzwyz1bolarkr2LQOXU6ltWiKEOiEMBSoQ2dNRTqgd66zZamryhomaybq_k966Wl2SXpMx-bc1Vy6zC-l3Uln</recordid><startdate>20191018</startdate><enddate>20191018</enddate><creator>Lin, Yijiong</creator><creator>Huang, Jiancong</creator><creator>Zimmer, Matthieu</creator><creator>Rojas, Juan</creator><creator>Weng, Paul</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20191018</creationdate><title>Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation</title><author>Lin, Yijiong ; Huang, Jiancong ; Zimmer, Matthieu ; Rojas, Juan ; Weng, Paul</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a679-3ff411beebfd2f3b5a50e644a32cce411ac2a1cd437e1ca955d46617743fe43</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Robotics</topic><toplevel>online_resources</toplevel><creatorcontrib>Lin, Yijiong</creatorcontrib><creatorcontrib>Huang, Jiancong</creatorcontrib><creatorcontrib>Zimmer, Matthieu</creatorcontrib><creatorcontrib>Rojas, Juan</creatorcontrib><creatorcontrib>Weng, Paul</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Lin, Yijiong</au><au>Huang, Jiancong</au><au>Zimmer, Matthieu</au><au>Rojas, Juan</au><au>Weng, Paul</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation</atitle><date>2019-10-18</date><risdate>2019</risdate><abstract>Deep reinforcement learning (DRL) is a promising approach for adaptive robot control, but its current application to robotics is currently hindered by high sample requirements. We propose two novel data augmentation techniques for DRL in order to reuse more efficiently observed data. The first one called Kaleidoscope Experience Replay exploits reflectional symmetries, while the second called Goal-augmented Experience Replay takes advantage of lax goal definitions. Our preliminary experimental results show a large increase in learning speed.</abstract><doi>10.48550/arxiv.1910.09959</doi><oa>free_for_read</oa></addata></record>
fulltext	fulltext_linktorsrc
identifier	DOI: 10.48550/arxiv.1910.09959
ispartof
issn
language	eng
recordid	cdi_arxiv_primary_1910_09959
source	arXiv.org
subjects	Computer Science - Artificial Intelligence Computer Science - Robotics
title	Towards More Sample Efficiency in Reinforcement Learning with Data Augmentation
url	https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-04T18%3A42%3A21IST&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=Towards%20More%20Sample%20Efficiency%20in%20Reinforcement%20Learning%20with%20Data%20Augmentation&rft.au=Lin,%20Yijiong&rft.date=2019-10-18&rft_id=info:doi/10.48550/arxiv.1910.09959&rft_dat=%3Carxiv_GOX%3E1910_09959%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