MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition

Real-world training data usually exhibits long-tailed distribution, where several majority classes have a significantly larger number of samples than the remaining minority classes. This imbalance degrades the performance of typical supervised learning algorithms designed for balanced training sets....

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) S. 5208 - 5217
Hauptverfasser: Li, Shuang, Gong, Kaixiong, Liu, Chi Harold, Wang, Yulin, Qiao, Feng, Cheng, Xinjing
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.06.2021
Schlagworte:
ISSN:1063-6919
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract Real-world training data usually exhibits long-tailed distribution, where several majority classes have a significantly larger number of samples than the remaining minority classes. This imbalance degrades the performance of typical supervised learning algorithms designed for balanced training sets. In this paper, we address this issue by augmenting minority classes with a recently proposed implicit semantic data augmentation (ISDA) algorithm [37], which produces diversified augmented samples by translating deep features along many semantically meaningful directions. Importantly, given that ISDA estimates the class-conditional statistics to obtain semantic directions, we find it ineffective to do this on minority classes due to the insufficient training data. To this end, we propose a novel approach to learn transformed semantic directions with meta-learning automatically. In specific, the augmentation strategy during training is dynamically optimized, aiming to minimize the loss on a small balanced validation set, which is approximated via a meta update step. Extensive empirical results on CIFAR-LT-10/100, ImageNet-LT, and iNaturalist 2017/2018 validate the effectiveness of our method.
AbstractList Real-world training data usually exhibits long-tailed distribution, where several majority classes have a significantly larger number of samples than the remaining minority classes. This imbalance degrades the performance of typical supervised learning algorithms designed for balanced training sets. In this paper, we address this issue by augmenting minority classes with a recently proposed implicit semantic data augmentation (ISDA) algorithm [37], which produces diversified augmented samples by translating deep features along many semantically meaningful directions. Importantly, given that ISDA estimates the class-conditional statistics to obtain semantic directions, we find it ineffective to do this on minority classes due to the insufficient training data. To this end, we propose a novel approach to learn transformed semantic directions with meta-learning automatically. In specific, the augmentation strategy during training is dynamically optimized, aiming to minimize the loss on a small balanced validation set, which is approximated via a meta update step. Extensive empirical results on CIFAR-LT-10/100, ImageNet-LT, and iNaturalist 2017/2018 validate the effectiveness of our method.
Author Wang, Yulin
Liu, Chi Harold
Li, Shuang
Gong, Kaixiong
Qiao, Feng
Cheng, Xinjing
Author_xml – sequence: 1
  givenname: Shuang
  surname: Li
  fullname: Li, Shuang
  email: shuangli@bit.edu.cn
  organization: Beijing Institute of Technology
– sequence: 2
  givenname: Kaixiong
  surname: Gong
  fullname: Gong, Kaixiong
  email: kxgong@bit.edu.cn
  organization: Beijing Institute of Technology
– sequence: 3
  givenname: Chi Harold
  surname: Liu
  fullname: Liu, Chi Harold
  email: liuchi02@gmail.com
  organization: Beijing Institute of Technology
– sequence: 4
  givenname: Yulin
  surname: Wang
  fullname: Wang, Yulin
  email: wang-yl19@mails.tsinghua.edu.cn
  organization: Tsinghua University
– sequence: 5
  givenname: Feng
  surname: Qiao
  fullname: Qiao, Feng
  email: feng.qiao@inceptio.ai
  organization: Inceptio Tech
– sequence: 6
  givenname: Xinjing
  surname: Cheng
  fullname: Cheng, Xinjing
  email: cnorbot@gmail.com
  organization: Inceptio Tech
BookMark eNotjM1Kw0AURkdRsK19Al3MCyTO752MuxJsFSJKW7stN9OZMpJMJEkXvr0WXX2Hw-GbkqvUJU_IPWc558w-lLv3tQIlTS6Y4DljmpsLMuUAWinNrLgkE85AZmC5vSHzYfhkjEnBOdhiQlavfsTN4nR8pGeiG99iGqOjv6r1acQxdomGrqdVl47ZFmPjD3QXhxM2dO1dd0zxnNyS64DN4Of_OyMfy6dt-ZxVb6uXclFlUTA5Zso5GaxlxmENAUVd60MhlcRCOTS1DTYYFMAt2HCAoAuHQgL3AbSQKtRyRu7-fqP3fv_Vxxb7773VxoDW8gecqE6H
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IH
CBEJK
RIE
RIO
DOI 10.1109/CVPR46437.2021.00517
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan (POP) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE/IET Electronic Library
IEEE Proceedings Order Plans (POP) 1998-present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE/IET Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
EISBN 1665445092
9781665445092
EISSN 1063-6919
EndPage 5217
ExternalDocumentID 9577655
Genre orig-research
GroupedDBID 6IE
6IH
6IL
6IN
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
IJVOP
OCL
RIE
RIL
RIO
ID FETCH-LOGICAL-i203t-4cc3f9907cab6fa2bb5d8343a84ca7b9f9f7a261969fd6f58ca2361ef65234fb3
IEDL.DBID RIE
ISICitedReferencesCount 132
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000739917305041&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:24:15 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i203t-4cc3f9907cab6fa2bb5d8343a84ca7b9f9f7a261969fd6f58ca2361ef65234fb3
PageCount 10
ParticipantIDs ieee_primary_9577655
PublicationCentury 2000
PublicationDate 2021-June
PublicationDateYYYYMMDD 2021-06-01
PublicationDate_xml – month: 06
  year: 2021
  text: 2021-June
PublicationDecade 2020
PublicationTitle Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online)
PublicationTitleAbbrev CVPR
PublicationYear 2021
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0003211698
Score 2.591175
Snippet Real-world training data usually exhibits long-tailed distribution, where several majority classes have a significantly larger number of samples than the...
SourceID ieee
SourceType Publisher
StartPage 5208
SubjectTerms Benchmark testing
Computer vision
Semantics
Supervised learning
Training
Training data
Visualization
Title MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition
URI https://ieeexplore.ieee.org/document/9577655
WOSCitedRecordID wos000739917305041&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
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LTwIxEG6AePCECsZ3evDowi59ezNE9ICEABJupO22ZBPZJbDr77ddVoyJF0-d9NJkmnbmm5lvBoD7CFspiEF-fLsIsAxV4MyICaLIUi4JjZUsW-YP2WjEFwsxroGHAxfGGFMWn5mOF8tcfpzpwofKuoIwRgmpg7pb91ytQzwFOSRDBa_YcVEouv35eIJ9XsqhwF7UKbtR_ZqhUpqQQfN_h5-A9g8XD44PVuYU1Ex6BpqV8wirp7lrgZc3k8vpU7F6hF6CU7N2Sks0dFvrimCUQueiwmGWroKZdN9BDOfJrpAfcPJdRpSlbfA-eJ71X4NqSkKQ9EKUB1hrZJ1NYVoqamVPKRJzhJHkWEumhBWWSY-TqLAxtYRr6RuuGEsdBsVWoXPQSLPUXAAoHZ6hGmPmvBYcasUJp4JFEqFQCR7ZS9Dyellu9o0wlpVKrv7evgbHXvH7uqob0Mi3hbkFR_ozT3bbu_L2vgBKFJqR
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LTwIxEG4QTfSECsa3PXh0YZc-tvVmjIhxIQSQcCNttyWbyK4B1t9vu6wYEy_emrk0mUk7883MNwPAbYCN4EQjt76de1j40rNuRHtBYCgThMZSFCPzo7DfZ9MpH1TA3ZYLo7Uums900x2LWn6cqdylylqchCElZAfsus1ZJVtrm1FBFstQzkp-XODz1uNkMMSuMmVxYDtoFvOofm1RKZxIp_a_6w9B44eNBwdbP3MEKjo9BrUyfITl41zVwXNPr8XoIZ_fQ3eCI72waksUtKJFSTFKoQ1SYZSlc28s7IcQw0myysU7HH43EmVpA7x1nsaPXa_ck-AlbR-tPawUMtarhEpIakRbShIzhJFgWIlQcsNNKBxSotzE1BCmhBu5og21KBQbiU5ANc1SfQqgsIiGKoxDG7dgX0lGGOVhIBDyJWeBOQN1p5fZx2YUxqxUyfnf4huw3x33oln00n-9AAfOCJsuq0tQXS9zfQX21Oc6WS2vC0t-AYkDndo
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=proceeding&rft.title=Proceedings+%28IEEE+Computer+Society+Conference+on+Computer+Vision+and+Pattern+Recognition.+Online%29&rft.atitle=MetaSAug%3A+Meta+Semantic+Augmentation+for+Long-Tailed+Visual+Recognition&rft.au=Li%2C+Shuang&rft.au=Gong%2C+Kaixiong&rft.au=Liu%2C+Chi+Harold&rft.au=Wang%2C+Yulin&rft.date=2021-06-01&rft.pub=IEEE&rft.eissn=1063-6919&rft.spage=5208&rft.epage=5217&rft_id=info:doi/10.1109%2FCVPR46437.2021.00517&rft.externalDocID=9577655