DATA: Differentiable ArchiTecture Approximation With Distribution Guided Sampling

Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop Differentiable ArchiTecture Approximation (DATA) with Ensemble Gumbel-Softmax (EGS) estimator and Architecture Distribution Constraint (ADC)...

Celý popis

Uložené v:

Podrobná bibliografia
Vydané v:	IEEE transactions on pattern analysis and machine intelligence Ročník 43; číslo 9; s. 2905 - 2920
Hlavní autori:	Zhang, Xinbang, Chang, Jianlong, Guo, Yiwen, Meng, Gaofeng, Xiang, Shiming, Lin, Zhouchen, Pan, Chunhong
Médium:	Journal Article
Jazyk:	English
Vydavateľské údaje:	New York IEEE 01.09.2021 The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Predmet:	Approximation Back propagation Binary codes Bridges Clustering Computer architecture distribution guided sampling ensemble gumbel-softmax Estimation Image classification Image segmentation Learning Mathematical analysis Modelling Neural architecture search(NAS) Optimization Sampling Search problems Searching Task analysis
ISSN:	0162-8828, 1939-3539, 2160-9292, 1939-3539
On-line prístup:	Získať plný text
Tagy:	Pridať tag Žiadne tagy, Buďte prvý, kto otaguje tento záznam!

Abstract	Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop Differentiable ArchiTecture Approximation (DATA) with Ensemble Gumbel-Softmax (EGS) estimator and Architecture Distribution Constraint (ADC) to automatically approximate architectures during searching and validating in a differentiable manner. Technically, the EGS estimator consists of a group of Gumbel-Softmax estimators, which is capable of converting probability vectors to binary codes and passing gradients reversely, reducing the estimation bias in a differentiable way. To narrow the distribution gap between sampled architectures and supernet, further, the ADC is introduced to reduce the variance of sampling during searching. Benefiting from such modeling, architecture probabilities and network weights in the NAS model can be jointly optimized with the standard back-propagation, yielding an end-to-end learning mechanism for searching deep neural architectures in an extended search space. Conclusively, in the validating process, a high-performance architecture that approaches to the learned one during searching is readily built. Extensive experiments on various tasks including image classification, few-shot learning, unsupervised clustering, semantic segmentation and language modeling strongly demonstrate that DATA is capable of discovering high-performance architectures while guaranteeing the required efficiency. Code is available at https://github.com/XinbangZhang/DATA-NAS .
AbstractList	Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop Differentiable ArchiTecture Approximation (DATA) with Ensemble Gumbel-Softmax (EGS) estimator and Architecture Distribution Constraint (ADC) to automatically approximate architectures during searching and validating in a differentiable manner. Technically, the EGS estimator consists of a group of Gumbel-Softmax estimators, which is capable of converting probability vectors to binary codes and passing gradients reversely, reducing the estimation bias in a differentiable way. To narrow the distribution gap between sampled architectures and supernet, further, the ADC is introduced to reduce the variance of sampling during searching. Benefiting from such modeling, architecture probabilities and network weights in the NAS model can be jointly optimized with the standard back-propagation, yielding an end-to-end learning mechanism for searching deep neural architectures in an extended search space. Conclusively, in the validating process, a high-performance architecture that approaches to the learned one during searching is readily built. Extensive experiments on various tasks including image classification, few-shot learning, unsupervised clustering, semantic segmentation and language modeling strongly demonstrate that DATA is capable of discovering high-performance architectures while guaranteeing the required efficiency. Code is available at https://github.com/XinbangZhang/DATA-NAS . Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop Differentiable ArchiTecture Approximation (DATA) with Ensemble Gumbel-Softmax (EGS) estimator and Architecture Distribution Constraint (ADC) to automatically approximate architectures during searching and validating in a differentiable manner. Technically, the EGS estimator consists of a group of Gumbel-Softmax estimators, which is capable of converting probability vectors to binary codes and passing gradients reversely, reducing the estimation bias in a differentiable way. To narrow the distribution gap between sampled architectures and supernet, further, the ADC is introduced to reduce the variance of sampling during searching. Benefiting from such modeling, architecture probabilities and network weights in the NAS model can be jointly optimized with the standard back-propagation, yielding an end-to-end learning mechanism for searching deep neural architectures in an extended search space. Conclusively, in the validating process, a high-performance architecture that approaches to the learned one during searching is readily built. Extensive experiments on various tasks including image classification, few-shot learning, unsupervised clustering, semantic segmentation and language modeling strongly demonstrate that DATA is capable of discovering high-performance architectures while guaranteeing the required efficiency. Code is available at https://github.com/XinbangZhang/DATA-NAS.Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop Differentiable ArchiTecture Approximation (DATA) with Ensemble Gumbel-Softmax (EGS) estimator and Architecture Distribution Constraint (ADC) to automatically approximate architectures during searching and validating in a differentiable manner. Technically, the EGS estimator consists of a group of Gumbel-Softmax estimators, which is capable of converting probability vectors to binary codes and passing gradients reversely, reducing the estimation bias in a differentiable way. To narrow the distribution gap between sampled architectures and supernet, further, the ADC is introduced to reduce the variance of sampling during searching. Benefiting from such modeling, architecture probabilities and network weights in the NAS model can be jointly optimized with the standard back-propagation, yielding an end-to-end learning mechanism for searching deep neural architectures in an extended search space. Conclusively, in the validating process, a high-performance architecture that approaches to the learned one during searching is readily built. Extensive experiments on various tasks including image classification, few-shot learning, unsupervised clustering, semantic segmentation and language modeling strongly demonstrate that DATA is capable of discovering high-performance architectures while guaranteeing the required efficiency. Code is available at https://github.com/XinbangZhang/DATA-NAS.
Author	Xiang, Shiming Guo, Yiwen Lin, Zhouchen Meng, Gaofeng Zhang, Xinbang Chang, Jianlong Pan, Chunhong
Author_xml	– sequence: 1 givenname: Xinbang orcidid: 0000-0002-8329-5363 surname: Zhang fullname: Zhang, Xinbang email: xinbang.zhang@nlpr.ia.ac.cn organization: Department of National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science, Beijing, China – sequence: 2 givenname: Jianlong orcidid: 0000-0002-0610-907X surname: Chang fullname: Chang, Jianlong email: jianlong.chang@huawei.com organization: Department of National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science, Beijing, China – sequence: 3 givenname: Yiwen orcidid: 0000-0002-0709-4877 surname: Guo fullname: Guo, Yiwen email: guoyiwen.ai@bytedance.com organization: Huawei Cloud & AI, National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing, China – sequence: 4 givenname: Gaofeng orcidid: 0000-0002-7103-6321 surname: Meng fullname: Meng, Gaofeng email: gfmeng@nlpr.ia.ac.cn organization: Department of National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science, Beijing, China – sequence: 5 givenname: Shiming orcidid: 0000-0002-2089-9733 surname: Xiang fullname: Xiang, Shiming email: smxiang@nlpr.ia.ac.cn organization: Department of National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science, Beijing, China – sequence: 6 givenname: Zhouchen orcidid: 0000-0003-1493-7569 surname: Lin fullname: Lin, Zhouchen email: zlin@pku.edu.cn organization: Bytedance AI Lab, Beijing, China – sequence: 7 givenname: Chunhong surname: Pan fullname: Pan, Chunhong email: chpan@nlpr.ia.ac.cn organization: School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China
BookMark	eNp9kEtPxCAYRYnR6Pj4A7qZxI2bjvBRXu4a34lGjWNcEspQxXTaEWii_16cMS5cuIFA7uG7nG203vWdQ2if4AkhWB1P76vb6wlgwBOaF0rYGhoB4bhQoGAdjTDhUEgJcgttx_iGMSkZpptoi4LkHKtyhB7Oqml1Mj7zTeOC65I3devGVbCvfupsGkI-LBah__Bzk3zfjZ99es3xmIKvh-XN5eBnbjZ-NPNF67uXXbTRmDa6vZ99Bz1dnE9Pr4qbu8vr0-qmsHl6KqiUlghhZAmGcWOcwLQxFqAupSEAM2Fy2xkzilGlatwAIbapqamVKAWu6Q46Wr2b270PLiY999G6tjWd64eooaSKU0YFz9HDP9G3fghdbqeBMaGE5IzklFylbOhjDK7R1qflp1MwvtUE62_teqldf2vXP9ozCn_QRcjCwuf_0MEK8s65X0ARSUrg9AsUBY0E
CODEN	ITPIDJ
CitedBy_id	crossref_primary_10_1016_j_engappai_2025_111476 crossref_primary_10_1016_j_isprsjprs_2022_03_007 crossref_primary_10_1016_j_neucom_2021_12_002 crossref_primary_10_1109_TNNLS_2022_3192169 crossref_primary_10_1145_3657632 crossref_primary_10_1109_TPAMI_2021_3077106 crossref_primary_10_3390_rs15092293 crossref_primary_10_23919_cje_2024_00_183 crossref_primary_10_1109_LGRS_2022_3213618
Cites_doi	10.1007/978-3-030-01246-5_2 10.24963/ijcai.2018/755 10.1016/B978-0-12-815480-9.00015-3 10.1109/CVPR.2015.7298594 10.1109/CVPR.2018.00257 10.1109/ICCV.2019.00138 10.1038/s42256-018-0006-z 10.1007/s12065-007-0002-4 10.1007/978-3-030-01264-9_8 10.1109/CVPR.2019.00720 10.1109/CVPR.2016.90 10.1109/CVPR.2019.00186 10.1109/TPAMI.2016.2577031 10.1109/CVPR.2018.00131 10.1109/ICCV.2017.626 10.1109/TPAMI.2020.3020300 10.1109/TPAMI.2017.2699184 10.1007/978-3-030-01246-5_1 10.1109/TPAMI.2015.2389824 10.1109/CVPR.2018.00907 10.1109/TPAMI.2016.2572683 10.1109/CVPR.2019.00017 10.1109/72.265960 10.1109/CVPR.2016.308 10.1109/CVPR.2017.243 10.1038/nature14539
ContentType	Journal Article
Copyright	Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2021
Copyright_xml	– notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2021
DBID	97E RIA RIE AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D 7X8
DOI	10.1109/TPAMI.2020.3020315
DatabaseName	IEEE Xplore (IEEE) IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Xplore CrossRef Computer and Information Systems Abstracts Electronics & Communications Abstracts Technology Research Database ProQuest Computer Science Collection Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Academic Computer and Information Systems Abstracts Professional MEDLINE - Academic
DatabaseTitle	CrossRef 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 Computer and Information Systems Abstracts Professional MEDLINE - Academic
DatabaseTitleList	Technology Research Database MEDLINE - Academic
Database_xml	– sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher – sequence: 2 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database
DeliveryMethod	fulltext_linktorsrc
Discipline	Engineering Computer Science
EISSN	2160-9292 1939-3539
EndPage	2920
ExternalDocumentID	10_1109_TPAMI_2020_3020315 9181426
Genre	orig-research
GrantInformation_xml	– fundername: Qualcomm funderid: 10.13039/100005144 – fundername: National Natural Science Foundation of China grantid: 91646207; 61976208 funderid: 10.13039/501100001809 – fundername: NSF China grantid: 61625301; 61731018 – fundername: Beijing Academy of Artificial Intelligence – fundername: National Key Research and Development Plan in China grantid: 2019AAA0105200 funderid: 10.13039/501100012166 – fundername: New Generation of AI grantid: 2018AAA0100400 – fundername: Major Scientific Research Project of Zhejiang Lab grantid: 2019KB0AC01; 2019KB0AB02
GroupedDBID	--- -DZ -~X .DC 0R~ 29I 4.4 53G 5GY 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFO ACGFS ACIWK ACNCT AENEX AGQYO AHBIQ AKJIK AKQYR ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 E.L EBS EJD F5P HZ~ IEDLZ IFIPE IPLJI JAVBF LAI M43 MS~ O9- OCL P2P PQQKQ RIA RIE RNS RXW TAE TN5 UHB ~02 AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D 7X8
ID	FETCH-LOGICAL-c328t-388c177a842a56aae703fac22b48a122d7a145d5a95399b0f211cfb3ab97470b3
IEDL.DBID	RIE
ISICitedReferencesCount	13
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000681124300007&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	0162-8828 1939-3539
IngestDate	Thu Oct 02 15:06:13 EDT 2025 Sun Jun 29 13:19:18 EDT 2025 Tue Nov 18 21:49:20 EST 2025 Sat Nov 29 05:15:59 EST 2025 Wed Aug 27 02:39:34 EDT 2025
IsPeerReviewed	true
IsScholarly	true
Issue	9
Language	English
License	https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c328t-388c177a842a56aae703fac22b48a122d7a145d5a95399b0f211cfb3ab97470b3
Notes	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23
ORCID	0000-0002-0610-907X 0000-0002-8329-5363 0000-0002-7103-6321 0000-0002-2089-9733 0000-0002-0709-4877 0000-0003-1493-7569
PMID	32866094
PQID	2557978651
PQPubID	85458
PageCount	16
ParticipantIDs	proquest_journals_2557978651 crossref_citationtrail_10_1109_TPAMI_2020_3020315 ieee_primary_9181426 crossref_primary_10_1109_TPAMI_2020_3020315 proquest_miscellaneous_2439635376
PublicationCentury	2000
PublicationDate	2021-09-01
PublicationDateYYYYMMDD	2021-09-01
PublicationDate_xml	– month: 09 year: 2021 text: 2021-09-01 day: 01
PublicationDecade	2020
PublicationPlace	New York
PublicationPlace_xml	– name: New York
PublicationTitle	IEEE transactions on pattern analysis and machine intelligence
PublicationTitleAbbrev	TPAMI
PublicationYear	2021
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	xie (ref74) 2016 pham (ref22) 2018 ref58 ref14 chen (ref35) 2018 melis (ref77) 2018 battaglia (ref81) 0 ref55 gumbel (ref53) 1954; 33 vinyals (ref68) 2016 ref18 liu (ref27) 2019 real (ref16) 2018 howard (ref63) 0 elsken (ref11) 2019 chang (ref41) 2019 ref46 ref45 chen (ref65) 2017; abs 1706 5587 courbariaux (ref51) 2015 ref47 ref44 sciuto (ref17) 0 bello (ref20) 2017 grave (ref80) 2017 baker (ref42) 2018 yi (ref75) 0 ref49 cai (ref50) 2019 ref7 brock (ref43) 2018 chang (ref82) 2018 ref4 ref3 ref6 elsken (ref9) 2019; 20 shin (ref29) 2018 krizhevsky (ref1) 2012 ref5 elsken (ref32) 2018 jang (ref54) 2017 li (ref33) 2019 ref34 ref36 liu (ref59) 2018 lee (ref61) 2015 vincent (ref73) 2010; 11 ref2 ref38 hsu (ref21) 2018; abs 1806 10332 ying (ref10) 2019 chen (ref66) 2019 tran (ref40) 0 józefowicz (ref48) 2015 maddison (ref52) 2017 ref70 ref72 luo (ref28) 2018 tan (ref39) 2018 yang (ref78) 2018 merity (ref76) 2018 kamath (ref13) 2018; abs 1803 6744 inan (ref79) 2017 ref23 ioffe (ref57) 2015 ref26 ref25 ref64 zilly (ref56) 2017 bello (ref19) 2017 zoph (ref24) 2017 casale (ref31) 2019; abs 1902 5116 guo (ref12) 0 finn (ref71) 2017 lecun (ref8) 2015; 521 snell (ref67) 2017 chen (ref37) 0 ravi (ref69) 2017 ref60 ref62 xie (ref30) 2019 real (ref15) 2017
References_xml	– ident: ref34 doi: 10.1007/978-3-030-01246-5_2 – volume: abs 1803 6744 year: 2018 ident: ref13 article-title: Neural architecture construction using envelopenets publication-title: CoRR – year: 2017 ident: ref19 article-title: Neural combinatorial optimization with reinforcement learning publication-title: Proc Int Conf Learn Representations Workshop Track – year: 2017 ident: ref52 article-title: The concrete distribution: A continuous relaxation of discrete random variables publication-title: Proc Int Conf Learn Representations – ident: ref44 doi: 10.24963/ijcai.2018/755 – volume: 11 start-page: 3371 year: 2010 ident: ref73 article-title: Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion publication-title: J Mach Learn Res – ident: ref14 doi: 10.1016/B978-0-12-815480-9.00015-3 – ident: ref62 doi: 10.1109/CVPR.2015.7298594 – ident: ref23 doi: 10.1109/CVPR.2018.00257 – ident: ref60 doi: 10.1109/ICCV.2019.00138 – ident: ref18 doi: 10.1038/s42256-018-0006-z – year: 0 ident: ref81 article-title: Relational inductive biases, deep learning, and graph networks – year: 2017 ident: ref69 article-title: Optimization as a model for few-shot learning publication-title: Proc Int Conf Learn Representations – start-page: 874 year: 2019 ident: ref41 article-title: DATA: differentiable architecture approximation publication-title: Proc Conf Neural Inf Process Syst – year: 0 ident: ref40 article-title: ConvNet architecture search for spatiotemporal feature learning publication-title: CoRR – start-page: 2820 year: 2018 ident: ref39 article-title: Mnasnet: Platform-aware neural architecture search for mobile publication-title: Proc IEEE Conf Comput Vis Pattern Recognit – year: 2017 ident: ref54 article-title: Categorical reparameterization with gumbel-softmax publication-title: Proc Int Conf Learn Representations – year: 2018 ident: ref43 article-title: SMASH: one-shot model architecture search through hypernetworks publication-title: Proc Int Conf Learn Representations – ident: ref47 doi: 10.1007/s12065-007-0002-4 – ident: ref64 doi: 10.1007/978-3-030-01264-9_8 – start-page: 3123 year: 2015 ident: ref51 article-title: BinaryConnect: Training deep neural networks with binary weights during propagations publication-title: Proc Conf Neural Inf Process Syst – ident: ref38 doi: 10.1109/CVPR.2019.00720 – ident: ref3 doi: 10.1109/CVPR.2016.90 – ident: ref26 doi: 10.1109/CVPR.2019.00186 – year: 0 ident: ref12 article-title: Single path one-shot neural architecture search with uniform sampling – start-page: 6638 year: 0 ident: ref37 article-title: DetNAS: Neural architecture search on object detection – year: 0 ident: ref63 article-title: MobileNets: Efficient convolutional neural networks for mobile vision applications – ident: ref2 doi: 10.1109/CVPR.2015.7298594 – year: 2019 ident: ref66 article-title: A closer look at few-shot classification publication-title: Proc Int Conf Learn Representations – volume: abs 1902 5116 year: 2019 ident: ref31 article-title: Probabilistic neural architecture search publication-title: CoRR – start-page: 4189 year: 2017 ident: ref56 article-title: Recurrent highway networks publication-title: Proc Int Conf Mach Learn – start-page: 478 year: 2016 ident: ref74 article-title: Unsupervised deep embedding for clustering analysis publication-title: Proc Int Conf Mach Learn – year: 0 ident: ref17 article-title: Evaluating the search phase of neural architecture search publication-title: Proc Int Conf Learn Representations – ident: ref4 doi: 10.1109/TPAMI.2016.2577031 – year: 2018 ident: ref78 article-title: Breaking the softmax bottleneck: A high-rank RNN language model publication-title: Proc Int Conf Learn Representations – year: 2017 ident: ref24 article-title: Neural architecture search with reinforcement learning publication-title: Proc Int Conf Learn Representations – year: 2019 ident: ref50 article-title: Proxylessnas: Direct neural architecture search on target task and hardware publication-title: Proc Int Conf Learn Representations – year: 2018 ident: ref59 article-title: Hierarchical representations for efficient architecture search publication-title: Proc Int Conf Learn Representations – year: 2018 ident: ref42 article-title: Accelerating neural architecture search using performance prediction publication-title: Proc Int Conf Learn Representations Workshop Track – start-page: 1126 year: 2017 ident: ref71 article-title: Model-agnostic meta-learning for fast adaptation of deep networks publication-title: Proc Int Conf Mach Learn – ident: ref70 doi: 10.1109/CVPR.2018.00131 – start-page: 3630 year: 2016 ident: ref68 article-title: Matching networks for one shot learning publication-title: Proc Conf Neural Inf Process Syst – volume: 20 start-page: 55:1 year: 2019 ident: ref9 article-title: Neural architecture search: A survey publication-title: J Mach Learn Res – start-page: 459 year: 2017 ident: ref20 article-title: Neural optimizer search with reinforcement learning publication-title: Proc Int Conf Mach Learn – year: 2019 ident: ref30 article-title: SNAS: stochastic neural architecture search publication-title: Proc Int Conf Learn Representations – ident: ref72 doi: 10.1109/ICCV.2017.626 – start-page: 4077 year: 2017 ident: ref67 article-title: Prototypical networks for few-shot learning publication-title: Proc Conf Neural Inf Process Syst – ident: ref49 doi: 10.1109/TPAMI.2020.3020300 – ident: ref7 doi: 10.1109/TPAMI.2017.2699184 – year: 2018 ident: ref29 article-title: Differentiable neural network architecture search publication-title: Proc Int Conf Learn Representations Workshop Track – start-page: 8713 year: 2018 ident: ref35 article-title: Searching for efficient multi-scale architectures for dense image prediction publication-title: Proc Conf Neural Inf Process Syst – year: 0 ident: ref75 article-title: Unsupervised and semi-supervised learning with categorical generative adversarial networks assisted by wasserstein distance for dermoscopy image classification – ident: ref45 doi: 10.1007/978-3-030-01246-5_1 – start-page: 1106 year: 2012 ident: ref1 article-title: Imagenet classification with deep convolutional neural networks publication-title: Proc Conf Neural Inf Process Syst – start-page: 448 year: 2015 ident: ref57 article-title: Batch normalization: Accelerating deep network training by reducing internal covariate shift publication-title: Proc 32nd Int Conf Int Conf Mach Learn – start-page: 562 year: 2015 ident: ref61 article-title: Deeply-supervised nets publication-title: Proc 18th Int Conf Artif Itell Statist – ident: ref5 doi: 10.1109/TPAMI.2015.2389824 – ident: ref25 doi: 10.1109/CVPR.2018.00907 – start-page: 11 year: 2018 ident: ref82 article-title: Structure-aware convolutional neural networks publication-title: Proc Conf Neural Inf Process Syst – ident: ref6 doi: 10.1109/TPAMI.2016.2572683 – volume: 33 year: 1954 ident: ref53 publication-title: Statistical Theory of Extreme Values and Some Practical Applications A Series of Lectures – year: 2017 ident: ref80 article-title: Improving neural language models with a continuous cache publication-title: Proc Int Conf Learn Representations – year: 2017 ident: ref79 article-title: Tying word vectors and word classifiers: A loss framework for language modeling publication-title: Proc Int Conf Learn Representations – year: 2018 ident: ref76 article-title: Regularizing and optimizing LSTM language models publication-title: Proc Int Conf Learn Representations – start-page: 4092 year: 2018 ident: ref22 article-title: Efficient neural architecture search via parameter sharing publication-title: Proc Int Conf Mach Learn – year: 2019 ident: ref11 article-title: Efficient multi-objective neural architecture search via lamarckian evolution publication-title: Proc Int Conf Learn Representations – start-page: 2902 year: 2017 ident: ref15 article-title: Large-scale evolution of image classifiers publication-title: Proc Int Conf Mach Learn – year: 2018 ident: ref77 article-title: On the state of the art of evaluation in neural language models publication-title: Proc Int Conf Learn Representations – ident: ref36 doi: 10.1109/CVPR.2019.00017 – start-page: 7827 year: 2018 ident: ref28 article-title: Neural architecture optimization publication-title: Proc Conf Neural Inf Process Syst – year: 2019 ident: ref27 article-title: DARTS: differentiable architecture search publication-title: Proc Int Conf Learn Representations – volume: abs 1706 5587 year: 2017 ident: ref65 article-title: Rethinking atrous convolution for semantic image segmentation publication-title: CoRR – volume: abs 1806 10332 year: 2018 ident: ref21 article-title: MONAS: multi-objective neural architecture search using reinforcement learning publication-title: CoRR – start-page: 2342 year: 2015 ident: ref48 article-title: An empirical exploration of recurrent network architectures publication-title: Proc 32nd Int Conf Int Conf Mach Learn – year: 2018 ident: ref32 article-title: Simple and efficient architecture search for convolutional neural networks publication-title: Proc Int Conf Learn Representations Workshop Track – start-page: 7105 year: 2019 ident: ref10 article-title: Nas-bench-101: Towards reproducible neural architecture search publication-title: Proc 36th Int Conf Mach Learn – ident: ref46 doi: 10.1109/72.265960 – ident: ref55 doi: 10.1109/CVPR.2016.308 – start-page: 4780 year: 2018 ident: ref16 article-title: Regularized evolution for image classifier architecture search publication-title: Proc AAAI – ident: ref58 doi: 10.1109/CVPR.2017.243 – volume: 521 start-page: 436 year: 2015 ident: ref8 article-title: Deep learning publication-title: Nature doi: 10.1038/nature14539 – year: 2019 ident: ref33 article-title: Random search and reproducibility for neural architecture search publication-title: Proc UAI
SSID	ssj0014503
Score	2.440202
Snippet	Neural architecture search (NAS) is inherently subject to the gap of architectures during searching and validating. To bridge this gap effectively, we develop...
SourceID	proquest crossref ieee
SourceType	Aggregation Database Enrichment Source Index Database Publisher
StartPage	2905
SubjectTerms	Approximation Back propagation Binary codes Bridges Clustering Computer architecture distribution guided sampling ensemble gumbel-softmax Estimation Image classification Image segmentation Learning Mathematical analysis Modelling Neural architecture search(NAS) Optimization Sampling Search problems Searching Task analysis
Title	DATA: Differentiable ArchiTecture Approximation With Distribution Guided Sampling
URI	https://ieeexplore.ieee.org/document/9181426 https://www.proquest.com/docview/2557978651 https://www.proquest.com/docview/2439635376
Volume	43
WOSCitedRecordID	wos000681124300007&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: PRVIEE databaseName: IEEE Electronic Library (IEL) customDbUrl: eissn: 2160-9292 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014503 issn: 0162-8828 databaseCode: RIE dateStart: 19790101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LS-RAEC5UPOjB54rji17Ym2ZNOp3ujrfgcw8ryo44t9Cv7A5IRpwZ8edb3cmEFUXwFkgnJF1VqfpSjw_gB3cukSpWkZapiZijPNJO8MgaVmW5FaziJpBNiOtrORjkN3Nw1PXCOOdC8Zn76Q9DLt-OzNT_KjvO0R2hR5mHeSFE06vVZQxYFliQMYJBC0cYMWuQifPj_k3x-xdCQYoI1SfeEk9Xk1LJeZyzN_4oEKy8-yoHV3Ox-rWHXIOVNqQkRaMD6zDn6g1YndE1kNZ6N2D5v9mDm3B7VvSLE3LWEqSgoesHR8Ig2n6TVyCFnzf-MmyaG8n9cPIPl487iixyOR1aZ8kf5avS67_f4O7ivH96FbX8CpHBt59EqZQmEUJJRlXGlXJo_ZUylGomVUKpFQq31WYq9-NrdVwhWDSVTpX2ICTW6RYs1KPabQOx3FrGtBFKMCYoSr-qUAcMYt5UxFb0IJntcmna4eOeA-OhDCAkzssgpNILqWyF1IPD7prHZvTGp6s3vSy6la0YerA3E2bZWue4RBglED3zLOnB9-402pVPlqjajaa4BiM1DMbw-7vz8Z13YYn6-pZQb7YHC5OnqduHRfM8GY6fDlBFB_IgqOgr977f6A
linkProvider	IEEE
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1bS9xAFD54Kdg-eG1xvbQj9K1Gk8lkJvEtaK2iLhZT9C3MLXVBssXdFX--ZybZUFEE3wI5CcmcOTnny7l8AN-5tVEqQxmoNNYBs5QHygoeGM2qJDOCVVx7sgnR76c3N9nlDOx2vTDWWl98Zvfcoc_lm6GeuF9l-xm6I_QoszCfMEajpluryxmwxPMgYwyDNo5AYtoiE2b7xWV-cYpgkCJGdam3yBHWxDTlPMzYM4_kKVZefJe9szleet9jLsNiG1SSvNkFKzBj61VYmhI2kNZ-V-HTf9MH1-D3UV7kB-SopUhBU1d3lvhRtEWTWSC5mzj-OGjaG8n1YHyL4qOOJIv8mgyMNeRKurr0-u9n-HP8szg8CVqGhUDj24-DOE11JIRMGZUJl9Ki_VdSU6pYKiNKjZC4rCaRmRtgq8IK4aKuVCyVgyGhir_AXD2s7ToQw41hTGkhBWOCov6rCneBRtQbi9CIHkTTVS51O37csWDclR6GhFnplVQ6JZWtknrwo7vmXzN8403pNaeLTrJVQw-2psosW_sclQikBOJnnkQ92OlOo2W5dIms7XCCMhirYTiGX-CN1-_8DRZOiovz8vy0f7YJH6mrdvHVZ1swN76f2G34oB_Gg9H9V79RnwB-1-JH
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%3A+Differentiable+ArchiTecture+Approximation+With+Distribution+Guided+Sampling&rft.jtitle=IEEE+transactions+on+pattern+analysis+and+machine+intelligence&rft.au=Zhang%2C+Xinbang&rft.au=Chang%2C+Jianlong&rft.au=Guo%2C+Yiwen&rft.au=Meng%2C+Gaofeng&rft.date=2021-09-01&rft.pub=IEEE&rft.issn=0162-8828&rft.volume=43&rft.issue=9&rft.spage=2905&rft.epage=2920&rft_id=info:doi/10.1109%2FTPAMI.2020.3020315&rft_id=info%3Apmid%2F32866094&rft.externalDocID=9181426
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0162-8828&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0162-8828&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0162-8828&client=summon