Delving Deep Into Label Smoothing
Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification...
Saved in:
| Published in | IEEE transactions on image processing Vol. 30; pp. 5984 - 5996 |
|---|---|
| Main Authors | , , , , , , |
| Format | Journal Article |
| Language | English |
| Published |
New York
IEEE
2021
The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Subjects | |
| Online Access | Get full text |
Cover
Loading…
| Abstract | Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/ |
|---|---|
| AbstractList | Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/ Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/.Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/. |
| Author | Zhang, Chang-Bin Li, Zhen Cheng, Ming-Ming Wei, Yunchao Han, Qi Jiang, Peng-Tao Hou, Qibin |
| Author_xml | – sequence: 1 givenname: Chang-Bin orcidid: 0000-0003-0043-8240 surname: Zhang fullname: Zhang, Chang-Bin organization: TKLNDST, CS, Nankai University, Tianjin, China – sequence: 2 givenname: Peng-Tao orcidid: 0000-0002-1786-4943 surname: Jiang fullname: Jiang, Peng-Tao organization: TKLNDST, CS, Nankai University, Tianjin, China – sequence: 3 givenname: Qibin orcidid: 0000-0002-8388-8708 surname: Hou fullname: Hou, Qibin organization: Department of Electrical and Computer Engineering, National University of Singapore, Singapore – sequence: 4 givenname: Yunchao orcidid: 0000-0002-2812-8781 surname: Wei fullname: Wei, Yunchao organization: Institute of Information Science, Beijing Jiaotong University, Beijing, China – sequence: 5 givenname: Qi surname: Han fullname: Han, Qi organization: TKLNDST, CS, Nankai University, Tianjin, China – sequence: 6 givenname: Zhen surname: Li fullname: Li, Zhen organization: TKLNDST, CS, Nankai University, Tianjin, China – sequence: 7 givenname: Ming-Ming orcidid: 0000-0001-5550-8758 surname: Cheng fullname: Cheng, Ming-Ming email: cmm@nankai.edu.cn organization: TKLNDST, CS, Nankai University, Tianjin, China |
| BookMark | eNp9kE1LAzEQhoNUrK3eBS8rXrxsnXxuc5TWj0JBwXoO2TjRLdtN3WwF_70pLR568DQD87zDyzMgvSY0SMgFhRGloG8Xs5cRA0ZHHMZaC3ZETqkWNAcQrJd2kEVeUKH7ZBDjEoAKSdUJ6XNBlaKanpKrKdbfVfORTRHX2azpQja3JdbZ6yqE7jNdzsixt3XE8_0ckreH-8XkKZ8_P84md_PccSa6nDHpnVBOS1AlZR6ZQD8uvCwZANdj4f17IUvgCGgd5eglOqtoKcADKMaH5Gb3d92Grw3Gzqyq6LCubYNhEw2TQkqtpeAJvT5Al2HTNqndllJKczXWiVI7yrUhxha9cVVnuyo0XWur2lAwW4kmSTRbiWYvMQXhILhuq5Vtf_6LXO4iFSL-4VookcrwX09heZ8 |
| CODEN | IIPRE4 |
| CitedBy_id | crossref_primary_10_23919_JSEE_2023_000069 crossref_primary_10_1109_TCAD_2022_3218239 crossref_primary_10_1088_1741_2552_adbec0 crossref_primary_10_1109_TIM_2024_3372220 crossref_primary_10_1109_TMM_2022_3181439 crossref_primary_10_1016_j_eswa_2023_119632 crossref_primary_10_1007_s44196_022_00144_y crossref_primary_10_1016_j_patcog_2024_110938 crossref_primary_10_1109_JBHI_2022_3180989 crossref_primary_10_1016_j_compind_2022_103786 crossref_primary_10_1109_TCE_2023_3293841 crossref_primary_10_1002_ima_22933 crossref_primary_10_1109_TCSVT_2023_3311142 crossref_primary_10_1109_TGRS_2023_3336357 crossref_primary_10_1109_TPAMI_2022_3166451 crossref_primary_10_7717_peerj_cs_2005 crossref_primary_10_1016_j_asoc_2024_111925 crossref_primary_10_1177_13694332241291247 crossref_primary_10_1109_TIP_2022_3197931 crossref_primary_10_1016_j_knosys_2023_110384 crossref_primary_10_1016_j_inffus_2023_02_012 crossref_primary_10_1016_j_patcog_2024_110545 crossref_primary_10_1016_j_neucom_2023_03_035 crossref_primary_10_1016_j_ins_2023_119014 crossref_primary_10_1109_ACCESS_2024_3378105 crossref_primary_10_1016_j_future_2024_107551 crossref_primary_10_1093_bioinformatics_btaf014 crossref_primary_10_1016_j_engappai_2024_107968 crossref_primary_10_1007_s13042_025_02537_9 crossref_primary_10_1145_3585279 crossref_primary_10_1145_3627817 crossref_primary_10_1109_ACCESS_2024_3359367 crossref_primary_10_1109_ACCESS_2025_3540164 crossref_primary_10_3390_fractalfract7020132 crossref_primary_10_1016_j_compag_2024_108951 crossref_primary_10_1109_TAI_2023_3297086 crossref_primary_10_1016_j_jrmge_2024_09_025 crossref_primary_10_1016_j_knosys_2023_110916 crossref_primary_10_1109_TTE_2023_3313986 crossref_primary_10_1016_j_cageo_2024_105708 crossref_primary_10_1109_TCSVT_2024_3359681 crossref_primary_10_1007_s13042_025_02591_3 crossref_primary_10_1016_j_engappai_2023_107605 crossref_primary_10_2139_ssrn_4149531 crossref_primary_10_1016_j_patcog_2023_110225 crossref_primary_10_1038_s41598_024_52360_5 crossref_primary_10_1016_j_asoc_2023_110191 crossref_primary_10_1016_j_apm_2024_115844 crossref_primary_10_1016_j_neucom_2024_127637 crossref_primary_10_3390_rs15143578 crossref_primary_10_1109_TII_2023_3256400 crossref_primary_10_1109_TAFFC_2022_3220291 crossref_primary_10_1007_s11263_024_02080_0 crossref_primary_10_3390_cancers14215312 crossref_primary_10_1109_TIP_2023_3321461 crossref_primary_10_1016_j_eswa_2024_126174 crossref_primary_10_2139_ssrn_4051352 crossref_primary_10_1016_j_patcog_2022_109063 crossref_primary_10_1109_ACCESS_2024_3485479 crossref_primary_10_1088_1361_6501_ad6e0e crossref_primary_10_1016_j_knosys_2024_112877 crossref_primary_10_1108_IJICC_07_2023_0198 crossref_primary_10_1109_TPAMI_2023_3311336 crossref_primary_10_1145_3674975 crossref_primary_10_1109_TAFFC_2024_3378698 crossref_primary_10_1007_s11227_024_06628_2 crossref_primary_10_1016_j_compeleceng_2024_109156 crossref_primary_10_1016_j_neucom_2025_129833 crossref_primary_10_1109_TCSVT_2023_3341728 crossref_primary_10_1186_s40537_024_01039_9 crossref_primary_10_3390_s24072168 crossref_primary_10_1109_JAS_2024_124629 crossref_primary_10_1371_journal_pdig_0000432 crossref_primary_10_1016_j_neucom_2024_128182 crossref_primary_10_1007_s11432_023_3979_9 crossref_primary_10_1016_j_neunet_2024_106215 crossref_primary_10_1109_ACCESS_2024_3362636 crossref_primary_10_1109_TNNLS_2023_3325541 crossref_primary_10_1007_s11227_023_05381_2 crossref_primary_10_1109_JSTARS_2022_3208107 crossref_primary_10_1371_journal_pone_0261307 crossref_primary_10_1016_j_sigpro_2023_109003 crossref_primary_10_1111_phor_12489 crossref_primary_10_1016_j_ipm_2023_103327 crossref_primary_10_1088_1361_6501_ad1f2d crossref_primary_10_1109_TAFFC_2023_3285231 crossref_primary_10_1049_ccs2_12079 crossref_primary_10_1109_TIM_2023_3235461 crossref_primary_10_1121_10_0011741 crossref_primary_10_1109_TCSVT_2024_3406443 crossref_primary_10_3934_nhm_2023070 crossref_primary_10_1007_s00330_023_09687_w crossref_primary_10_1109_TIM_2024_3453330 crossref_primary_10_1007_s00371_022_02721_w crossref_primary_10_1016_j_ress_2022_108972 crossref_primary_10_1016_j_ress_2025_110909 crossref_primary_10_2139_ssrn_4201034 crossref_primary_10_32604_cmc_2024_058586 crossref_primary_10_1007_s40962_024_01335_3 crossref_primary_10_1016_j_jvcir_2024_104062 crossref_primary_10_1109_ACCESS_2023_3300692 crossref_primary_10_1016_j_ymssp_2025_112582 crossref_primary_10_1109_JSEN_2023_3295363 crossref_primary_10_32604_iasc_2023_029037 crossref_primary_10_1038_s41467_024_45999_1 crossref_primary_10_1016_j_eswa_2023_119544 crossref_primary_10_1016_j_biosystemseng_2024_09_008 crossref_primary_10_1088_1361_6501_adbf34 crossref_primary_10_1016_j_patcog_2024_110725 crossref_primary_10_1109_TGRS_2022_3174352 crossref_primary_10_1109_TGRS_2024_3419111 crossref_primary_10_1145_3644391 crossref_primary_10_7717_peerj_cs_1288 crossref_primary_10_1016_j_eswa_2023_121981 crossref_primary_10_1088_1741_2552_ac8dc5 crossref_primary_10_1109_LRA_2025_3540379 |
| Cites_doi | 10.1109/CVPR42600.2020.01009 10.1109/TIP.2019.2921876 10.1109/CVPR.2018.00474 10.1109/TPAMI.2020.3029801 10.1109/CVPR.2017.634 10.1109/34.134056 10.1109/ICCV.2019.00971 10.1109/ICCV.2019.00053 10.1109/TNNLS.2017.2677468 10.1109/TPAMI.2015.2456899 10.1109/TIP.2016.2531289 10.1109/CVPR.2019.01150 10.1109/TNNLS.2019.2938782 10.1109/CVPR.2018.00582 10.1109/CVPR42600.2020.00396 10.1109/TIP.2019.2917781 10.1007/s11263-009-0275-4 10.1109/ICCV.2019.00524 10.1109/CVPR.2019.00265 10.1109/TNNLS.2016.2545112 10.1109/TCSVT.2016.2607345 10.1109/TPAMI.2019.2938758 10.1109/ACCESS.2019.2960566 10.1109/CVPR.2016.308 10.1109/TIP.2018.2883743 10.1609/aaai.v33i01.33015565 10.1109/TNNLS.2018.2851924 10.1109/TIP.2020.3044220 10.1109/ICCV.2019.00822 10.1109/TIP.2019.2947792 10.1109/ICCV.2019.00381 10.1109/CVPR.2016.514 10.1007/s11432-020-3097-4 10.1007/s11263-019-01265-2 10.1109/TNNLS.2017.2699783 10.1109/TIP.2017.2774041 10.1109/TNNLS.2018.2875470 10.1109/CVPR.2016.90 10.1109/CVPR.2019.00718 10.1109/TNNLS.2018.2792062 10.1109/CVPR.2018.00745 10.1109/TPAMI.2017.2723400 10.1109/TIP.2018.2877939 10.1007/978-3-030-01258-8_5 10.1109/TIP.2019.2950508 10.1109/CVPR.2016.91 10.1109/CVPR.2017.243 10.1109/CVPR.2009.5206848 10.1109/TNNLS.2018.2852721 10.1109/ICCV.2019.00972 10.1109/TIP.2015.2423557 10.1109/TPAMI.2016.2577031 10.1109/ICCVW.2013.77 10.1109/ICVGIP.2008.47 10.1109/TIP.2020.2989544 10.1109/ICCV.2019.00041 |
| 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/TIP.2021.3089942 |
| DatabaseName | IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Electronic Library (IEL) 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://proxy.k.utb.cz/login?url=https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Applied Sciences Engineering |
| EISSN | 1941-0042 |
| EndPage | 5996 |
| ExternalDocumentID | 10_1109_TIP_2021_3089942 9464693 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: NSFC grantid: 61922046 funderid: 10.13039/501100001809 – fundername: Science and Technology Innovation Project from Chinese Ministry of Education – fundername: Fundamental Research Funds for the Central Universities through Nankai University grantid: 63213090 funderid: 10.13039/501100012226 – fundername: National Key Research and Development Program of China grantid: 2018AAA0100400 funderid: 10.13039/501100012166 |
| GroupedDBID | --- -~X .DC 0R~ 29I 4.4 53G 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABFSI ABQJQ ABVLG ACGFO ACGFS ACIWK AENEX AETIX AGQYO AGSQL AHBIQ AI. AIBXA AKJIK AKQYR ALLEH ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 E.L EBS EJD F5P HZ~ H~9 ICLAB IFIPE IFJZH IPLJI JAVBF LAI M43 MS~ O9- OCL P2P RIA RIE RNS TAE TN5 VH1 AAYOK AAYXX CITATION RIG 7SC 7SP 8FD JQ2 L7M L~C L~D 7X8 |
| ID | FETCH-LOGICAL-c324t-225fc46c9506b12fe24ef87f5b2003984ffd75b03e0eac13ef5eca61b40f00623 |
| IEDL.DBID | RIE |
| ISSN | 1057-7149 1941-0042 |
| IngestDate | Fri Jul 11 09:19:31 EDT 2025 Mon Jun 30 10:19:55 EDT 2025 Thu Apr 24 22:53:19 EDT 2025 Tue Jul 01 02:03:26 EDT 2025 Wed Aug 27 02:26:42 EDT 2025 |
| IsPeerReviewed | true |
| IsScholarly | true |
| 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-c324t-225fc46c9506b12fe24ef87f5b2003984ffd75b03e0eac13ef5eca61b40f00623 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23 |
| ORCID | 0000-0003-0043-8240 0000-0002-2812-8781 0000-0002-1786-4943 0000-0001-5550-8758 0000-0002-8388-8708 |
| PMID | 34166191 |
| PQID | 2546693689 |
| PQPubID | 85429 |
| PageCount | 13 |
| ParticipantIDs | crossref_citationtrail_10_1109_TIP_2021_3089942 crossref_primary_10_1109_TIP_2021_3089942 proquest_miscellaneous_2545599543 ieee_primary_9464693 proquest_journals_2546693689 |
| ProviderPackageCode | CITATION AAYXX |
| PublicationCentury | 2000 |
| PublicationDate | 20210000 2021-00-00 20210101 |
| PublicationDateYYYYMMDD | 2021-01-01 |
| PublicationDate_xml | – year: 2021 text: 20210000 |
| PublicationDecade | 2020 |
| PublicationPlace | New York |
| PublicationPlace_xml | – name: New York |
| PublicationTitle | IEEE transactions on image processing |
| PublicationTitleAbbrev | TIP |
| 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 | li (ref28) 2020; 31 ref57 zhang (ref69) 2017 ref59 ref15 ref53 ref52 ref54 ref10 liu (ref77) 2016 ref16 ref18 maaten (ref56) 2008; 9 goodfellow (ref70) 2015 devries (ref12) 2017 ref50 arazo (ref51) 2019 ref45 ref48 ref42 ref41 ref44 ref43 ref49 ref7 müller (ref55) 2019 ref9 ref4 ref3 ref6 ref5 ref40 kurakin (ref71) 2017 ref79 ref35 ref78 ref37 ref36 ref75 ref31 ref74 ref30 guo (ref80) 2017 ref33 ref76 ref32 ref2 ghiasi (ref14) 2018 reed (ref11) 2015 ref39 ref38 shu (ref47) 2019 krizhevsky (ref8) 0 ref73 ref72 tan (ref22) 2019; 97 xiao (ref68) 2015 ref24 ref67 ref23 ref26 ref25 ref64 ref20 ref63 ref66 ref65 ren (ref46) 2018 hinton (ref17) 2015 paszke (ref58) 2017 zhang (ref13) 2018 ref27 ref29 maji (ref21) 2013 ref60 ref62 ref61 wah (ref19) 2011 furlanello (ref34) 2018 simonyan (ref1) 2015 |
| References_xml | – ident: ref23 doi: 10.1109/CVPR42600.2020.01009 – ident: ref66 doi: 10.1109/TIP.2019.2921876 – ident: ref6 doi: 10.1109/CVPR.2018.00474 – ident: ref30 doi: 10.1109/TPAMI.2020.3029801 – ident: ref4 doi: 10.1109/CVPR.2017.634 – volume: 97 start-page: 6105 year: 2019 ident: ref22 article-title: EfficientNet: Rethinking model scaling for convolutional neural networks publication-title: Mach Learn Res – ident: ref40 doi: 10.1109/34.134056 – ident: ref72 doi: 10.1109/ICCV.2019.00971 – ident: ref38 doi: 10.1109/ICCV.2019.00053 – ident: ref52 doi: 10.1109/TNNLS.2017.2677468 – start-page: 4696 year: 2019 ident: ref55 article-title: When does label smoothing help? publication-title: Proc Adv Neural Inform Process Syst – ident: ref48 doi: 10.1109/TPAMI.2015.2456899 – start-page: 21 year: 2016 ident: ref77 article-title: SSD: Single shot multibox detector publication-title: Proc Eur Conf Comput Vis – year: 2015 ident: ref11 article-title: Training deep neural networks on noisy labels with bootstrapping publication-title: Proc Int Conf Learn Represent Workshop – ident: ref62 doi: 10.1109/TIP.2016.2531289 – ident: ref50 doi: 10.1109/CVPR.2019.01150 – ident: ref42 doi: 10.1109/TNNLS.2019.2938782 – volume: 9 start-page: 2579 year: 2008 ident: ref56 article-title: Visualizing data using t-SNE publication-title: J Mach Learn Res – volume: 31 start-page: 2294 year: 2020 ident: ref28 article-title: Reconstruction regularized deep metric learning for multi-label image classification publication-title: IEEE Trans Neural Netw Learn Syst – ident: ref44 doi: 10.1109/CVPR.2018.00582 – ident: ref57 doi: 10.1109/CVPR42600.2020.00396 – start-page: 10727 year: 2018 ident: ref14 article-title: DropBlock: A regularization method for convolutional networks publication-title: Proc Adv Neural Inform Process Syst – ident: ref75 doi: 10.1109/TIP.2019.2917781 – ident: ref79 doi: 10.1007/s11263-009-0275-4 – ident: ref45 doi: 10.1109/ICCV.2019.00524 – start-page: 312 year: 2019 ident: ref51 article-title: Unsupervised label noise modeling and loss correction publication-title: Proc Int Conf Mech Learn – ident: ref29 doi: 10.1109/CVPR.2019.00265 – start-page: 1321 year: 2017 ident: ref80 article-title: On calibration of modern neural networks publication-title: Proc Int Conf Mach Learn – ident: ref61 doi: 10.1109/TNNLS.2016.2545112 – start-page: 2691 year: 2015 ident: ref68 article-title: Learning from massive noisy labeled data for image classification publication-title: Proc IEEE Conf Comput Vis Pattern Recognit (CVPR) – start-page: 1602 year: 2018 ident: ref34 article-title: Born-again neural networks publication-title: Proc Int Conf Mech Learn – ident: ref64 doi: 10.1109/TCSVT.2016.2607345 – year: 2018 ident: ref13 article-title: Mixup: Beyond empirical risk minimization publication-title: Proc Int Conf Learn Represent – year: 2017 ident: ref58 article-title: Automatic differentiation in PyTorch publication-title: Proc Adv Neural Inform Process Syst Workshop – ident: ref7 doi: 10.1109/TPAMI.2019.2938758 – ident: ref15 doi: 10.1109/ACCESS.2019.2960566 – ident: ref10 doi: 10.1109/CVPR.2016.308 – ident: ref37 doi: 10.1109/TIP.2018.2883743 – ident: ref25 doi: 10.1609/aaai.v33i01.33015565 – ident: ref33 doi: 10.1109/TNNLS.2018.2851924 – ident: ref32 doi: 10.1109/TIP.2020.3044220 – year: 2015 ident: ref1 article-title: Very deep convolutional networks for large-scale image recognition publication-title: Proc Int Conf Learn Represent – ident: ref31 doi: 10.1109/ICCV.2019.00822 – start-page: 4334 year: 2018 ident: ref46 article-title: Learning to reweight examples for robust deep learning publication-title: Proc Int Conf Mech Learn – ident: ref74 doi: 10.1109/TIP.2019.2947792 – ident: ref24 doi: 10.1109/ICCV.2019.00381 – ident: ref26 doi: 10.1109/CVPR.2016.514 – ident: ref59 doi: 10.1007/s11432-020-3097-4 – ident: ref16 doi: 10.1007/s11263-019-01265-2 – year: 2011 ident: ref19 article-title: The caltech-UCSD birds-200-2011 dataset – ident: ref41 doi: 10.1109/TNNLS.2017.2699783 – ident: ref65 doi: 10.1109/TIP.2017.2774041 – ident: ref53 doi: 10.1109/TNNLS.2018.2875470 – ident: ref2 doi: 10.1109/CVPR.2016.90 – ident: ref54 doi: 10.1109/CVPR.2019.00718 – ident: ref43 doi: 10.1109/TNNLS.2018.2792062 – ident: ref5 doi: 10.1109/CVPR.2018.00745 – ident: ref67 doi: 10.1109/TPAMI.2017.2723400 – ident: ref39 doi: 10.1109/TIP.2018.2877939 – year: 2015 ident: ref17 article-title: Distilling the knowledge in a neural network publication-title: Proc Adv Neural Inform Process Syst Workshop – year: 2017 ident: ref71 article-title: Adversarial machine learning at scale publication-title: Proc Int Conf Learn Represent – ident: ref27 doi: 10.1007/978-3-030-01258-8_5 – ident: ref35 doi: 10.1109/TIP.2019.2950508 – ident: ref73 doi: 10.1109/CVPR.2016.91 – ident: ref3 doi: 10.1109/CVPR.2017.243 – year: 2017 ident: ref69 article-title: Understanding deep learning requires rethinking generalization publication-title: Proc Int Conf Learn Represent – year: 0 ident: ref8 article-title: Learning multiple layers of features from tiny images – ident: ref9 doi: 10.1109/CVPR.2009.5206848 – ident: ref63 doi: 10.1109/TNNLS.2018.2852721 – year: 2013 ident: ref21 article-title: A database for fine-grained aircraft recognition publication-title: Proc IEEE Conf Comp Vis Pattern Recognit – ident: ref78 doi: 10.1109/ICCV.2019.00972 – ident: ref60 doi: 10.1109/TIP.2015.2423557 – year: 2017 ident: ref12 article-title: Improved regularization of convolutional neural networks with cutout publication-title: arXiv 1708 04552 – ident: ref76 doi: 10.1109/TPAMI.2016.2577031 – year: 2015 ident: ref70 article-title: Explaining and harnessing adversarial examples publication-title: Proc Int Conf Learn Represent – ident: ref20 doi: 10.1109/ICCVW.2013.77 – ident: ref18 doi: 10.1109/ICVGIP.2008.47 – start-page: 1919 year: 2019 ident: ref47 article-title: Meta-weight-net: Learning an explicit mapping for sample weighting publication-title: Proc Adv Neural Inform Process Syst – ident: ref36 doi: 10.1109/TIP.2020.2989544 – ident: ref49 doi: 10.1109/ICCV.2019.00041 |
| SSID | ssj0014516 |
| Score | 2.6897995 |
| Snippet | Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the... |
| SourceID | proquest crossref ieee |
| SourceType | Aggregation Database Enrichment Source Index Database Publisher |
| StartPage | 5984 |
| SubjectTerms | Artificial neural networks Cats Classification knowledge distillation Labels Noise measurement noisy labels online label smoothing Predictive models Regularization Robustness Smoothing Smoothing methods soft labels Source code Training |
| Title | Delving Deep Into Label Smoothing |
| URI | https://ieeexplore.ieee.org/document/9464693 https://www.proquest.com/docview/2546693689 https://www.proquest.com/docview/2545599543 |
| Volume | 30 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV3dT9swED8VnsYDH4WJDoZSaS-TSOskthM_IqBqp4GQBhJvUeycX1aSaqQv_PWcnTTa2DTxFukcx_H5fHe-8-8AvqBkpYwKHWZJhiEnjRwWJhMhckvagKVofS2Cm1s5f-DfHsXjAM77uzCI6JPPcOIefSy_rM3aHZVNFZfkzSVbsEWOW3tXq48YuIKzPrIp0jAls38TkmRqer-4I0cwjiaJi3FxV8CG9m4aior-0Ea-vMpfe7JXNLM9uNkMsc0v-TlZN3piXt6gN773H_Zht7M4g4t2iRzAAKsh7HXWZ9DJ9vMQdn6DJjyE8RUu3VlDcIW4ChZVUwffC43L4MdTTcwlyhE8zK7vL-dhV04hNGQ1NSFJrjVcGiWY1FFsMeZos9QK7RLUVMatLVOhWYKMduMoQSvQFDLSnFl31TL5CNtVXeExBDxKpShFyrUynDpUMWYYFSyx5OWauBzBdDOtuemwxl3Ji2XufQ6mcuJJ7niSdzwZwdf-jVWLs_GftoduXvt23ZSO4HTDubwTxOfcwf0TTWZqBOOeTCLk4iJFhfXat_G4azz59O-eT-CD-3578nIK282vNX4mW6TRZ34RvgJqLNVw |
| linkProvider | IEEE |
| linkToHtml | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjR1NT9sw9Imxw8aBbnyIQrcFaZdJpHUS24mP0xhqtxZNWpG4RbHzfKEkCNILv55nJ43Yh6bdIj3bcvz8vvy-AD6iZKWMCh1mSYYhJ4kcFiYTIXJL0oClaH0vgsWlnF7xb9fiegvO-lwYRPTBZzh2n96XX9Zm7Z7KJopLsuaSF_CS5L6I22yt3mfgWs5636ZIw5QU_41TkqnJcvaDTME4GifOy8VdCxvi3rQZFf0ij3yDlT-4shc1FwNYbDbZRpjcjNeNHpvH3-o3_u9fvIHdTucMPreX5C1sYbUHg07_DDrqftiDnWfFCffh9BxX7rUhOEe8C2ZVUwfzQuMq-HlbE3oJcgBXF1-XX6Zh11AhNKQ3NSHRrjVcGiWY1FFsMeZos9QK7ULUVMatLVOhWYKM-HGUoBVoChlpzqxLtkwOYbuqKzyCgEepFKVIuVaG04IqxgyjgiWW7FwTl0OYbI41N121cdf0YpV7q4OpnHCSO5zkHU6G8KmfcddW2vjH2H13rv247kiHMNpgLu9I8SF3Bf8JJjM1hNMeTETkPCNFhfXaj_GV13hy_PeVP8Cr6XIxz-ezy-8n8NrtpX2HGcF2c7_Gd6SZNPq9v5BPv17Yug |
| 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=Delving+Deep+Into+Label+Smoothing&rft.jtitle=IEEE+transactions+on+image+processing&rft.au=Zhang%2C+Chang-Bin&rft.au=Jiang%2C+Peng-Tao&rft.au=Hou%2C+Qibin&rft.au=Wei%2C+Yunchao&rft.date=2021&rft.pub=IEEE&rft.issn=1057-7149&rft.volume=30&rft.spage=5984&rft.epage=5996&rft_id=info:doi/10.1109%2FTIP.2021.3089942&rft_id=info%3Apmid%2F34166191&rft.externalDocID=9464693 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1057-7149&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1057-7149&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1057-7149&client=summon |