Data-Aware Adaptive Pruning Model Compression Algorithm Based on a Group Attention Mechanism and Reinforcement Learning

The success of convolutional neural networks (CNNs) benefits from the stacking of convolutional layers, which improves the model's receptive field for image data but also causes a decrease in inference speed. To improve the inference speed of large convolutional network models without sacrifici...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE access Ročník 10; s. 82396 - 82406
Hlavní autori: Yang, Zhi, Zhai, Yuan, Xiang, Yi, Wu, Jianquan, Shi, Jinliang, Wu, Ying
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Piscataway IEEE 2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Predmet:
ISSN:2169-3536, 2169-3536
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract The success of convolutional neural networks (CNNs) benefits from the stacking of convolutional layers, which improves the model's receptive field for image data but also causes a decrease in inference speed. To improve the inference speed of large convolutional network models without sacrificing performance indicators too much, a data-aware adaptive pruning algorithm is proposed. The algorithm consists of two parts, namely, a channel pruning method based on the attention mechanism and a data-aware pruning policy based on reinforcement learning. Experimental results on the CIFAR-100 dataset show that the performance of the proposed pruning algorithm is reduced by only 2.05%, 1.93% and 5.66% after pruning the VGG19, ResNet56 and EfficientNet networks, respectively, but the speedup ratios are 3.63, 3.35, and 1.14, respectively, and the comprehensive pruning performance is the best. In addition, the generalization ability of the reconstruction model is evaluated on the ImageNet dataset and FGVC Aircraft dataset, and the performance of the proposed algorithm is the best, which shows that the proposed algorithm learns data-related information in the pruning process, that is, it is a data-aware algorithm.
AbstractList The success of convolutional neural networks (CNNs) benefits from the stacking of convolutional layers, which improves the model's receptive field for image data but also causes a decrease in inference speed. To improve the inference speed of large convolutional network models without sacrificing performance indicators too much, a data-aware adaptive pruning algorithm is proposed. The algorithm consists of two parts, namely, a channel pruning method based on the attention mechanism and a data-aware pruning policy based on reinforcement learning. Experimental results on the CIFAR-100 dataset show that the performance of the proposed pruning algorithm is reduced by only 2.05%, 1.93% and 5.66% after pruning the VGG19, ResNet56 and EfficientNet networks, respectively, but the speedup ratios are 3.63, 3.35, and 1.14, respectively, and the comprehensive pruning performance is the best. In addition, the generalization ability of the reconstruction model is evaluated on the ImageNet dataset and FGVC Aircraft dataset, and the performance of the proposed algorithm is the best, which shows that the proposed algorithm learns data-related information in the pruning process, that is, it is a data-aware algorithm.
Author Shi, Jinliang
Zhai, Yuan
Wu, Ying
Xiang, Yi
Yang, Zhi
Wu, Jianquan
Author_xml – sequence: 1
  givenname: Zhi
  orcidid: 0000-0003-1046-8654
  surname: Yang
  fullname: Yang, Zhi
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
– sequence: 2
  givenname: Yuan
  surname: Zhai
  fullname: Zhai, Yuan
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
– sequence: 3
  givenname: Yi
  surname: Xiang
  fullname: Xiang, Yi
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
– sequence: 4
  givenname: Jianquan
  surname: Wu
  fullname: Wu, Jianquan
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
– sequence: 5
  givenname: Jinliang
  surname: Shi
  fullname: Shi, Jinliang
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
– sequence: 6
  givenname: Ying
  orcidid: 0000-0002-9683-5998
  surname: Wu
  fullname: Wu, Ying
  email: wuying1992@cqust.edu.cn
  organization: School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, China
BookMark eNp9kU9v1DAQxSNUJErpJ-jFEucs_hMnzjGEUiptBaJwtsb2ZOtVEgfHS8W3JyEFIQ74Yut53m9G815mZ2MYMcuuGN0xRus3Tdte39_vOOV8J5hSjNXPsnPOyjoXUpRnf71fZJfzfKTLUYskq_Ps8R0kyJtHiEgaB1Py35F8iqfRjwdyFxz2pA3DFHGefRhJ0x9C9OlhIG9hRkcWCchNDKeJNCnhmNaiO7QPMPp5IDA68hn92IVocVi-yR4hruxX2fMO-hkvn-6L7Ov76y_th3z_8ea2bfa5LahKOROutka5mltjpeuoYA4Y5UqqCiXw0kiQYAxYwUpRUqgclIZKZbpKyLoUF9ntxnUBjnqKfoD4Qwfw-pcQ4kFDTN72qGXBrVv25yy1RYUKys4ZXMjCmIqDWFivN9YUw7cTzkkfwymOy_iaV1QUQhVy7VhvVTaGeY7YaesTrItJEXyvGdVrbHqLTa-x6afYFq_4x_t74v-7rjaXR8Q_jloxURVM_ATj2KbX
CODEN IAECCG
CitedBy_id crossref_primary_10_1109_ACCESS_2024_3457863
crossref_primary_10_1109_ACCESS_2025_3533419
crossref_primary_10_1109_ACCESS_2022_3229192
crossref_primary_10_1109_ACCESS_2024_3498904
Cites_doi 10.1109/CVPR.2016.280
10.1007/s11263-015-0816-y
10.1109/ICCV.2017.298
10.3390/electronics9081209
10.1007/s11042-017-4440-4
10.1109/ICCV.2017.541
10.1109/CVPR.2016.90
10.1109/ICCV.2017.155
10.1109/CVPR42600.2020.00160
10.1109/ICME.2017.8019465
10.1109/CVPR.2019.00290
10.1609/aaai.v33i01.33015676
10.1109/CVPR.2018.00890
10.3390/app10196866
10.1111/mice.12449
10.1109/MSP.2017.2765695
10.3389/fnins.2019.00405
10.1109/CVPR.2019.01152
10.1016/j.imavis.2021.104143
ContentType Journal Article
Copyright Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2022
Copyright_xml – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2022
DBID 97E
ESBDL
RIA
RIE
AAYXX
CITATION
7SC
7SP
7SR
8BQ
8FD
JG9
JQ2
L7M
L~C
L~D
DOA
DOI 10.1109/ACCESS.2022.3188119
DatabaseName IEEE Xplore (IEEE)
IEEE Xplore Open Access Journals
IEEE All-Society Periodicals Package (ASPP) 1998–Present
IEEE Electronic Library (IEL)
CrossRef
Computer and Information Systems Abstracts
Electronics & Communications Abstracts
Engineered Materials Abstracts
METADEX
Technology Research Database
Materials Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
DOAJ Directory of Open Access Journals
DatabaseTitle CrossRef
Materials Research Database
Engineered Materials Abstracts
Technology Research Database
Computer and Information Systems Abstracts – Academic
Electronics & Communications Abstracts
ProQuest Computer Science Collection
Computer and Information Systems Abstracts
Advanced Technologies Database with Aerospace
METADEX
Computer and Information Systems Abstracts Professional
DatabaseTitleList
Materials Research Database

Database_xml – sequence: 1
  dbid: DOA
  name: DOAJ Directory of Open Access Journals
  url: https://www.doaj.org/
  sourceTypes: Open Website
– sequence: 2
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
EISSN 2169-3536
EndPage 82406
ExternalDocumentID oai_doaj_org_article_542cd188dc0c47e8a6fdbeac33bb72a3
10_1109_ACCESS_2022_3188119
9813741
Genre orig-research
GrantInformation_xml – fundername: Chongqing Talent Plan, Cooperation Project between Chongqing Municipal Undergraduate Universities and Institutes affiliated to the Chinese Academy of Sciences, in 2021
  grantid: HZ2021015
  funderid: 10.13039/501100002367
– fundername: National Natural Science and Foundation of China
  grantid: 61903055
  funderid: 10.13039/501100001809
GroupedDBID 0R~
4.4
5VS
6IK
97E
AAJGR
ABAZT
ABVLG
ACGFS
ADBBV
AGSQL
ALMA_UNASSIGNED_HOLDINGS
BCNDV
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
EBS
EJD
ESBDL
GROUPED_DOAJ
IPLJI
JAVBF
KQ8
M43
M~E
O9-
OCL
OK1
RIA
RIE
RNS
AAYXX
CITATION
7SC
7SP
7SR
8BQ
8FD
JG9
JQ2
L7M
L~C
L~D
ID FETCH-LOGICAL-c408t-13d9cb8d92cbc5df031da1028587e5a26b5a5abbac316360a7da6b058bf735963
IEDL.DBID DOA
ISICitedReferencesCount 6
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000841773400001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 2169-3536
IngestDate Tue Oct 14 15:00:10 EDT 2025
Mon Jun 30 04:53:45 EDT 2025
Sat Nov 29 06:32:14 EST 2025
Tue Nov 18 22:19:02 EST 2025
Wed Aug 27 02:14:03 EDT 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Language English
License https://creativecommons.org/licenses/by/4.0/legalcode
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c408t-13d9cb8d92cbc5df031da1028587e5a26b5a5abbac316360a7da6b058bf735963
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ORCID 0000-0003-1046-8654
0000-0002-9683-5998
OpenAccessLink https://doaj.org/article/542cd188dc0c47e8a6fdbeac33bb72a3
PQID 2703438456
PQPubID 4845423
PageCount 11
ParticipantIDs ieee_primary_9813741
crossref_primary_10_1109_ACCESS_2022_3188119
crossref_citationtrail_10_1109_ACCESS_2022_3188119
doaj_primary_oai_doaj_org_article_542cd188dc0c47e8a6fdbeac33bb72a3
proquest_journals_2703438456
PublicationCentury 2000
PublicationDate 20220000
2022-00-00
20220101
2022-01-01
PublicationDateYYYYMMDD 2022-01-01
PublicationDate_xml – year: 2022
  text: 20220000
PublicationDecade 2020
PublicationPlace Piscataway
PublicationPlace_xml – name: Piscataway
PublicationTitle IEEE access
PublicationTitleAbbrev Access
PublicationYear 2022
Publisher IEEE
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Publisher_xml – name: IEEE
– name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
References ref13
ref15
ref14
Maji (ref22) 2013
ref11
ref10
ref2
ref1
ref16
ref19
ref18
Li (ref8)
Peng (ref17)
ref24
ref20
ref21
Srivastava (ref26) 2014; 15
Meng (ref12) 2020
ref27
ref7
ref9
ref3
ref6
ref5
Tan (ref25)
Simonyan (ref23)
Han (ref4)
References_xml – ident: ref9
  doi: 10.1109/CVPR.2016.280
– ident: ref21
  doi: 10.1007/s11263-015-0816-y
– ident: ref16
  doi: 10.1109/ICCV.2017.298
– ident: ref19
  doi: 10.3390/electronics9081209
– ident: ref24
  doi: 10.1007/s11042-017-4440-4
– ident: ref15
  doi: 10.1109/ICCV.2017.541
– volume-title: arXiv:1306.5151
  year: 2013
  ident: ref22
  article-title: Fine-grained visual classification of aircraft
– ident: ref20
  doi: 10.1109/CVPR.2016.90
– ident: ref10
  doi: 10.1109/ICCV.2017.155
– ident: ref13
  doi: 10.1109/CVPR42600.2020.00160
– ident: ref1
  doi: 10.1109/ICME.2017.8019465
– start-page: 5113
  volume-title: Proc. Int. Conf. Mach. Learn.
  ident: ref17
  article-title: Collaborative channel pruning for deep networks
– start-page: 1135
  volume-title: Proc. Adv. Neural Inf. Process. Syst.
  ident: ref4
  article-title: Learning both weights and connections for efficient neural network
– ident: ref11
  doi: 10.1109/CVPR.2019.00290
– start-page: 1
  volume-title: Proc. Int. Conf. Learn. Represent.
  ident: ref8
  article-title: Pruning filters for efficient ConvNets
– ident: ref6
  doi: 10.1609/aaai.v33i01.33015676
– ident: ref3
  doi: 10.1109/CVPR.2018.00890
– start-page: 6105
  volume-title: Proc. Int. Conf. Mach. Learn.
  ident: ref25
  article-title: EfficientNet: Rethinking model scaling for convolutional neural networks
– ident: ref2
  doi: 10.3390/app10196866
– ident: ref7
  doi: 10.1111/mice.12449
– ident: ref27
  doi: 10.1109/MSP.2017.2765695
– volume-title: arXiv:2009.14410
  year: 2020
  ident: ref12
  article-title: Pruning filter in filter
– ident: ref18
  doi: 10.3389/fnins.2019.00405
– volume: 15
  start-page: 1929
  issue: 1
  year: 2014
  ident: ref26
  article-title: Dropout: A simple way to prevent neural networks from overfitting
  publication-title: J. Mach. Learn. Res.
– ident: ref5
  doi: 10.1109/CVPR.2019.01152
– start-page: 1
  volume-title: Proc. Int. Conf. Learn. Represent.
  ident: ref23
  article-title: Very deep convolutional networks for large-scale image recognition
– ident: ref14
  doi: 10.1016/j.imavis.2021.104143
SSID ssj0000816957
Score 2.2755334
Snippet The success of convolutional neural networks (CNNs) benefits from the stacking of convolutional layers, which improves the model's receptive field for image...
The success of convolutional neural networks (CNNs) benefits from the stacking of convolutional layers, which improves the model’s receptive field for image...
SourceID doaj
proquest
crossref
ieee
SourceType Open Website
Aggregation Database
Enrichment Source
Index Database
Publisher
StartPage 82396
SubjectTerms Adaptive algorithms
Aircraft performance
Algorithms
Artificial neural networks
attention mechanism
Computational modeling
Convolutional neural networks
Datasets
Filtering algorithms
Inference
Machine learning
Model compression
pruning algorithm
Reinforcement learning
Training data
SummonAdditionalLinks – databaseName: IEEE Electronic Library (IEL)
  dbid: RIE
  link: http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Lb9QwELZKxQEOvApi24J84NjQxI_YOaYLFReqCoHUmzV-pKy0zVbZLP37eBw3qgRC4hZZduT48zw88XxDyIfGlVH1Rs-tKr0vhCh9AUKGouQqqK6uO1F3qdiEurjQV1fN5R45mXNhQgjp8ln4iI_pX77fuB2Gyk4bXXGFWeqPlKqnXK05noIFJBqpMrFQVTan7XIZvyEeARmLJ1OtK2TTeWB8Ekd_LqryhyZO5uX8-f9N7AV5lt1I2k64vyR7oX9Fnj4gFzwgd59ghKK9gyHQ1sMt6jV6OewwEEKxBNqaojKY7sH2tF1fb4bV-POGnkXD5mlsApoiU7Qdx-lSJP0aMFF4tb2h0Hv6LSTaVZcijDQztV6_Jj_OP39ffilymYXCiVJjMXrfOKt9w5x10ndRzD2g3yG1ChJYbSVIsBYcr5BdDJSH2pZS205xGQX4DdnvN314Syh0inXRBYmLEQTUla2CY_EIB54rKyQsCLtff-MyBzmWwlibdBYpGzOBZhA0k0FbkJN50O1EwfHv7mcI7NwV-bNTQ0TMZHE0UjDnY38ft6pQQUPdeRttEOfWKgZ8QQ4Q5fklGeAFOb7fJibL-tawqDQF19ETPfz7qCPyBCc4BW6Oyf447MI78tj9Glfb4X3axr8BtILx0Q
  priority: 102
  providerName: IEEE
Title Data-Aware Adaptive Pruning Model Compression Algorithm Based on a Group Attention Mechanism and Reinforcement Learning
URI https://ieeexplore.ieee.org/document/9813741
https://www.proquest.com/docview/2703438456
https://doaj.org/article/542cd188dc0c47e8a6fdbeac33bb72a3
Volume 10
WOSCitedRecordID wos000841773400001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAON
  databaseName: DOAJ Directory of Open Access Journals
  customDbUrl:
  eissn: 2169-3536
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0000816957
  issn: 2169-3536
  databaseCode: DOA
  dateStart: 20130101
  isFulltext: true
  titleUrlDefault: https://www.doaj.org/
  providerName: Directory of Open Access Journals
– providerCode: PRVHPJ
  databaseName: ROAD: Directory of Open Access Scholarly Resources
  customDbUrl:
  eissn: 2169-3536
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0000816957
  issn: 2169-3536
  databaseCode: M~E
  dateStart: 20130101
  isFulltext: true
  titleUrlDefault: https://road.issn.org
  providerName: ISSN International Centre
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV1Lb9QwELZQxQEOCCiILaXygSNRY8eO7WO6tOLSqkIg9WaNHykrbdMqm9Ibv52x465WQoILlxwsx4k9D8-M7O8j5KPxNbpejNxYHUIlRB0qEDJWdaOi6tu2F22fySbUxYW-ujKXO1Rf6UzYDA88L9yxFNwHpnXAQYWKGto-OPQWTeOc4pBxPmtldpKp7IM1a41UBWaI1ea4Wy5xRpgQco55qtYsYevsbEUZsb9QrPzhl_Nmc_aSvChRIu3mv3tFnsThNXm-gx24Tx4-wwRV9wBjpF2Au-S26OV4n-ocNDGcrWmy9fmY60C79fXtuJp-3NAT3LcCxSagufBEu2mazzzS85juAa82NxSGQL_GjKrqcwGRFiDW6zfk-9npt-WXqrAoVF7UOnHNB-OdDoZ752Xo0YoDpLBCahUl8NZJkOAcLilL4GGgArSultr1qpFon2_J3nA7xHeEQq94jxFGo0QU0DLHoueYoUFolBMSFoQ_Lqj1BWI8MV2sbU41amNnKdgkBVuksCCfti_dzQgbf-9-kiS17ZrgsXMDKo0tSmP_pTQLsp_kvB3EaIZzYgty-Ch3W0x5Yzn6RNFoDDQP_sen35NnaTpzFeeQ7E3jffxAnvqf02ozHmUtxuf5r9OjfBfxNy3y-HA
linkProvider Directory of Open Access Journals
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Lb9QwELaqggQceBXElgI-cGxo4tixc0wXqiLaVYWK1Js1fqRdaZutdrP079fjuFElEBK3yLIjx5_n4YnnG0I-1zYPqjd4bkXuXMZ57jLgwmd5Kb1sq6rlVRuLTcjZTF1c1GdbZH_MhfHex8tn_gs-xn_5bmk3GCo7qFVRSsxSf4SVs1K21hhRwRIStZCJWqjI64NmOg1fEQ6BjIWzqVIF8uk8MD-RpT-VVflDF0cDc_Ti_6b2kjxPjiRtBuRfkS3fvSbPHtAL7pDbr9BD1tzCytPGwQ1qNnq22mAohGIRtAVFdTDchO1os7hcrub91TU9DKbN0dAENMamaNP3w7VIeuoxVXi-vqbQOfrTR-JVG2OMNHG1Xr4hv46-nU-Ps1RoIbM8V1iO3tXWKFcza6xwbRB0B-h5CCW9AFYZAQKMAVsWyC8G0kFlcqFMK0sRRPgt2e6WnX9HKLSStcEJCYvhOVSFKbxl4RAHrpSGC5gQdr_-2iYWciyGsdDxNJLXegBNI2g6gTYh--Ogm4GE49_dDxHYsSsyaMeGgJhOAqkFZ9aF_i5sVi69gqp1JlihsjRGMignZAdRHl-SAJ6QvfttopO0rzULapOXKviiu38f9Yk8OT4_PdEn32c_3pOnONkhjLNHtvvVxn8gj-3vfr5efYxb-g5t9fUa
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Data-Aware+Adaptive+Pruning+Model+Compression+Algorithm+Based+on+a+Group+Attention+Mechanism+and+Reinforcement+Learning&rft.jtitle=IEEE+access&rft.au=Yang%2C+Zhi&rft.au=Zhai%2C+Yuan&rft.au=Xiang%2C+Yi&rft.au=Wu%2C+Jianquan&rft.date=2022&rft.pub=IEEE&rft.eissn=2169-3536&rft.volume=10&rft.spage=82396&rft.epage=82406&rft_id=info:doi/10.1109%2FACCESS.2022.3188119&rft.externalDocID=9813741
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2169-3536&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2169-3536&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2169-3536&client=summon