Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy

Structural re-parameterization (Rep) methods has achieved significant performance improvement on traditional convolutional network. Most current Rep methods rely on prior knowledge to select the reparameterization operations. However, the performance of architecture is limited by the type of operati...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2022-07
Hauptverfasser: Yu, Xinyi, Wang, Xiaowei, Rong, Jintao, Zhang, Mingyang, Ou, Linlin
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 Yu, Xinyi
Wang, Xiaowei
Rong, Jintao
Zhang, Mingyang
Ou, Linlin
description Structural re-parameterization (Rep) methods has achieved significant performance improvement on traditional convolutional network. Most current Rep methods rely on prior knowledge to select the reparameterization operations. However, the performance of architecture is limited by the type of operations and prior knowledge. To break this restriction, in this work, an improved re-parameterization search space is designed, which including more type of re-parameterization operations. Concretely, the performance of convolutional networks can be further improved by the search space. To effectively explore this search space, an automatic re-parameterization enhancement strategy is designed based on neural architecture search (NAS), which can search a excellent re-parameterization architecture. Besides, we visualize the output features of the architecture to analyze the reasons for the formation of the re-parameterization architecture. On public datasets, we achieve better results. Under the same training conditions as ResNet, we improve the accuracy of ResNet-50 by 1.82% on ImageNet-1k.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2650101111</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2650101111</sourcerecordid><originalsourceid>FETCH-proquest_journals_26501011113</originalsourceid><addsrcrecordid>eNqNjEEKwjAURIMgKNo7fHAdSFNbXasVVwrWfQn1R6PV1J9Uqae3Fg_gbOYt5k2PDWUUhXw-lXLAAucuQgiZzGQcR0NWp1qbwuDdwx55pUjd0COZt_LG3mFXIXXkIENFxRm0JUiVa7i3fIVVaRvYon9ZusJCOTxCa60MYfG1VAnp05Z1x9RA5ts3PDVj1teqdBj8esQm6_Sw3PCK7KNG5_OLranVXS6TWIQibBP9t_oAxzNNIw</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2650101111</pqid></control><display><type>article</type><title>Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy</title><source>Free E- Journals</source><creator>Yu, Xinyi ; Wang, Xiaowei ; Rong, Jintao ; Zhang, Mingyang ; Ou, Linlin</creator><creatorcontrib>Yu, Xinyi ; Wang, Xiaowei ; Rong, Jintao ; Zhang, Mingyang ; Ou, Linlin</creatorcontrib><description>Structural re-parameterization (Rep) methods has achieved significant performance improvement on traditional convolutional network. Most current Rep methods rely on prior knowledge to select the reparameterization operations. However, the performance of architecture is limited by the type of operations and prior knowledge. To break this restriction, in this work, an improved re-parameterization search space is designed, which including more type of re-parameterization operations. Concretely, the performance of convolutional networks can be further improved by the search space. To effectively explore this search space, an automatic re-parameterization enhancement strategy is designed based on neural architecture search (NAS), which can search a excellent re-parameterization architecture. Besides, we visualize the output features of the architecture to analyze the reasons for the formation of the re-parameterization architecture. On public datasets, we achieve better results. Under the same training conditions as ResNet, we improve the accuracy of ResNet-50 by 1.82% on ImageNet-1k.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Networks ; Parameterization ; Searching</subject><ispartof>arXiv.org, 2022-07</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>Yu, Xinyi</creatorcontrib><creatorcontrib>Wang, Xiaowei</creatorcontrib><creatorcontrib>Rong, Jintao</creatorcontrib><creatorcontrib>Zhang, Mingyang</creatorcontrib><creatorcontrib>Ou, Linlin</creatorcontrib><title>Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy</title><title>arXiv.org</title><description>Structural re-parameterization (Rep) methods has achieved significant performance improvement on traditional convolutional network. Most current Rep methods rely on prior knowledge to select the reparameterization operations. However, the performance of architecture is limited by the type of operations and prior knowledge. To break this restriction, in this work, an improved re-parameterization search space is designed, which including more type of re-parameterization operations. Concretely, the performance of convolutional networks can be further improved by the search space. To effectively explore this search space, an automatic re-parameterization enhancement strategy is designed based on neural architecture search (NAS), which can search a excellent re-parameterization architecture. Besides, we visualize the output features of the architecture to analyze the reasons for the formation of the re-parameterization architecture. On public datasets, we achieve better results. Under the same training conditions as ResNet, we improve the accuracy of ResNet-50 by 1.82% on ImageNet-1k.</description><subject>Networks</subject><subject>Parameterization</subject><subject>Searching</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNjEEKwjAURIMgKNo7fHAdSFNbXasVVwrWfQn1R6PV1J9Uqae3Fg_gbOYt5k2PDWUUhXw-lXLAAucuQgiZzGQcR0NWp1qbwuDdwx55pUjd0COZt_LG3mFXIXXkIENFxRm0JUiVa7i3fIVVaRvYon9ZusJCOTxCa60MYfG1VAnp05Z1x9RA5ts3PDVj1teqdBj8esQm6_Sw3PCK7KNG5_OLranVXS6TWIQibBP9t_oAxzNNIw</recordid><startdate>20220703</startdate><enddate>20220703</enddate><creator>Yu, Xinyi</creator><creator>Wang, Xiaowei</creator><creator>Rong, Jintao</creator><creator>Zhang, Mingyang</creator><creator>Ou, Linlin</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>20220703</creationdate><title>Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy</title><author>Yu, Xinyi ; Wang, Xiaowei ; Rong, Jintao ; Zhang, Mingyang ; Ou, Linlin</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26501011113</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Networks</topic><topic>Parameterization</topic><topic>Searching</topic><toplevel>online_resources</toplevel><creatorcontrib>Yu, Xinyi</creatorcontrib><creatorcontrib>Wang, Xiaowei</creatorcontrib><creatorcontrib>Rong, Jintao</creatorcontrib><creatorcontrib>Zhang, Mingyang</creatorcontrib><creatorcontrib>Ou, Linlin</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>Yu, Xinyi</au><au>Wang, Xiaowei</au><au>Rong, Jintao</au><au>Zhang, Mingyang</au><au>Ou, Linlin</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy</atitle><jtitle>arXiv.org</jtitle><date>2022-07-03</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>Structural re-parameterization (Rep) methods has achieved significant performance improvement on traditional convolutional network. Most current Rep methods rely on prior knowledge to select the reparameterization operations. However, the performance of architecture is limited by the type of operations and prior knowledge. To break this restriction, in this work, an improved re-parameterization search space is designed, which including more type of re-parameterization operations. Concretely, the performance of convolutional networks can be further improved by the search space. To effectively explore this search space, an automatic re-parameterization enhancement strategy is designed based on neural architecture search (NAS), which can search a excellent re-parameterization architecture. Besides, we visualize the output features of the architecture to analyze the reasons for the formation of the re-parameterization architecture. On public datasets, we achieve better results. Under the same training conditions as ResNet, we improve the accuracy of ResNet-50 by 1.82% on ImageNet-1k.</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, 2022-07
issn 2331-8422
language eng
recordid cdi_proquest_journals_2650101111
source Free E- Journals
subjects Networks
Parameterization
Searching
title Efficient Re-parameterization Operations Search for Easy-to-Deploy Network Based on Directional Evolutionary Strategy
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-28T23%3A22%3A48IST&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=Efficient%20Re-parameterization%20Operations%20Search%20for%20Easy-to-Deploy%20Network%20Based%20on%20Directional%20Evolutionary%20Strategy&rft.jtitle=arXiv.org&rft.au=Yu,%20Xinyi&rft.date=2022-07-03&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2650101111%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2650101111&rft_id=info:pmid/&rfr_iscdi=true