Transfer Knowledge from Head to Tail: Uncertainty Calibration under Long-tailed Distribution

How to estimate the uncertainty of a given model is a crucial problem. Current calibration techniques treat different classes equally and thus implicitly assume that the distribution of training data is balanced, but ignore the fact that real-world data often follows a long-tailed distribution. In t...

Full description

Saved in:
Bibliographic Details
Published in:Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) pp. 19978 - 19987
Main Authors: Chen, Jiahao, Su, Bing
Format: Conference Proceeding
Language:English
Published: IEEE 01.06.2023
Subjects:
ISSN:1063-6919
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract How to estimate the uncertainty of a given model is a crucial problem. Current calibration techniques treat different classes equally and thus implicitly assume that the distribution of training data is balanced, but ignore the fact that real-world data often follows a long-tailed distribution. In this paper, we explore the problem of calibrating the model trained from a long-tailed distribution. Due to the difference between the imbalanced training distribution and balanced test distribution, existing calibration methods such as temperature scaling can not generalize well to this problem. Specific calibration methods for domain adaptation are also not applicable because they rely on unlabeled target domain instances which are not available. Models trained from a long-tailed distribution tend to be more over-confident to head classes. To this end, we propose a novel knowledge-transferring-based calibration method by estimating the importance weights for samples of tail classes to realize long-tailed calibration. Our method models the distribution of each class as a Gaussian distribution and views the source statistics of head classes as a prior to calibrate the target distributions of tail classes. We adaptively transfer knowledge from head classes to get the target probability density of tail classes. The importance weight is estimated by the ratio of the target probability density over the source probability density. Extensive experiments on CIFAR-10-LT, MNIST-LT, CIFAR-100-LT, and ImageNet-LT datasets demonstrate the effectiveness of our method.
AbstractList How to estimate the uncertainty of a given model is a crucial problem. Current calibration techniques treat different classes equally and thus implicitly assume that the distribution of training data is balanced, but ignore the fact that real-world data often follows a long-tailed distribution. In this paper, we explore the problem of calibrating the model trained from a long-tailed distribution. Due to the difference between the imbalanced training distribution and balanced test distribution, existing calibration methods such as temperature scaling can not generalize well to this problem. Specific calibration methods for domain adaptation are also not applicable because they rely on unlabeled target domain instances which are not available. Models trained from a long-tailed distribution tend to be more over-confident to head classes. To this end, we propose a novel knowledge-transferring-based calibration method by estimating the importance weights for samples of tail classes to realize long-tailed calibration. Our method models the distribution of each class as a Gaussian distribution and views the source statistics of head classes as a prior to calibrate the target distributions of tail classes. We adaptively transfer knowledge from head classes to get the target probability density of tail classes. The importance weight is estimated by the ratio of the target probability density over the source probability density. Extensive experiments on CIFAR-10-LT, MNIST-LT, CIFAR-100-LT, and ImageNet-LT datasets demonstrate the effectiveness of our method.
Author Su, Bing
Chen, Jiahao
Author_xml – sequence: 1
  givenname: Jiahao
  surname: Chen
  fullname: Chen, Jiahao
  email: nicelemon666@gmail.com
  organization: Gaoling School of Artificial Intelligence, Renmin University of China
– sequence: 2
  givenname: Bing
  surname: Su
  fullname: Su, Bing
  email: subingats@gmail.com
  organization: Gaoling School of Artificial Intelligence, Renmin University of China
BookMark eNotjNFKwzAUQKMoOOf-YA_5gc7cZE1zfZPqnDhQZPNJGDdNOiJdKmmH7O_t0KfzcDjnml3ENnrGpiBmAAJvy4-391wWEmdSSDUTgKDO2AQLNCoXSoBEc85GILTKNAJesUnXfQkhlATQaEbsc50odrVP_CW2P413O8_r1O750pPjfcvXFJo7vomVTz2F2B95SU2wifrQRn6IbkhXbdxlgx1y_hC6PgV7OOkbdllT0_nJP8dss3hcl8ts9fr0XN6vsiDFvM9sRUo7r2xeKARSdSWdk2puKoNUGC3nJ29ljSgdAFW1yjWQ1YRoDVg1ZtO_b_Deb79T2FM6bkEMdzBC_QLfKldQ
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IH
CBEJK
RIE
RIO
DOI 10.1109/CVPR52729.2023.01913
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan (POP) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP) 1998-present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE/IET Electronic Library (IEL) (UW System Shared)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
EISBN 9798350301298
EISSN 1063-6919
EndPage 19987
ExternalDocumentID 10204180
Genre orig-research
GrantInformation_xml – fundername: Research Funds of Renmin University of China
  grantid: 21XNLG05
  funderid: 10.13039/501100004260
– fundername: National Natural Science Foundation of China
  grantid: 61976206,61832017
  funderid: 10.13039/501100001809
– fundername: Fundamental Research Funds for the Central Universities
  funderid: 10.13039/501100012226
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-i204t-bca36de3b57391a3fc2dd2348c89a78624a36db2f992d11acf3561ab6a99b81b3
IEDL.DBID RIE
ISICitedReferencesCount 28
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001062531304029&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 02:56:33 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i204t-bca36de3b57391a3fc2dd2348c89a78624a36db2f992d11acf3561ab6a99b81b3
PageCount 10
ParticipantIDs ieee_primary_10204180
PublicationCentury 2000
PublicationDate 2023-June
PublicationDateYYYYMMDD 2023-06-01
PublicationDate_xml – month: 06
  year: 2023
  text: 2023-June
PublicationDecade 2020
PublicationTitle Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online)
PublicationTitleAbbrev CVPR
PublicationYear 2023
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0003211698
Score 2.388829
Snippet How to estimate the uncertainty of a given model is a crucial problem. Current calibration techniques treat different classes equally and thus implicitly...
SourceID ieee
SourceType Publisher
StartPage 19978
SubjectTerms Adaptation models
continual
Head
low-shot
meta
or long-tail learning
Tail
Temperature distribution
Training
Training data
Transfer
Uncertainty
Title Transfer Knowledge from Head to Tail: Uncertainty Calibration under Long-tailed Distribution
URI https://ieeexplore.ieee.org/document/10204180
WOSCitedRecordID wos001062531304029&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/eLvHCXMwlV3LSsNAFB20uHBVHxXfzMLt1GQmycy4rZaCpRRpSxdCmVekIIm0qeDfO3eSVjcu3IW8mZvhnrk55x6E7mjkUbiMHEnyVBJo6E20zhVxVgrloO4f2O6zIR-NxHwux41YPWhhnHOBfOa6sBn-5dvSbKBU5mc4jZJY-BX6PudZLdbaFVSYX8pkUjTyuDiS973Z-CWlHj12wSO868EMmBj8MlEJOaTf_ufTj1DnR42Hx7s8c4z2XHGC2g18xM3kXJ-i15B3_BX4eVsow6AewQMfR1yVeKKW7w946s8ONIDqC4MyS9ffAAY12QoPy-KNAKvU3_oRmuo2flgdNO0_TXoD0pgnkKV_zYpoo1hmHdMpZzJWLDfUWsoSYYRUHGQhcFzTXEpq41iZnHkopXSmpNQey7Iz1CrKwp0jrKjQNou0UsYmueLSOu5Tu8kkS7mOkwvUgdFafNT9MRbbgbr8Y_8VOoSA1ISra9SqVht3gw7MZ7Vcr25DVL8Bl2KkCw
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3LSsNAFB2kCrqqj4pvZ-F2ajKT17itlkpjKdKWLoQyr0ihJNKmgn_v3Ela3bhwF_JmboZ75uacexC6o55F4dwzJMhCTqChN5EyE8RonggDdX_Hdp-k8WCQTKd8WIvVnRbGGOPIZ6YNm-5fvi7UGkpldoZTL_ATu0LfDYOAepVca1tSYXYxE_GkFsj5Hr_vTIavIbX4sQ0u4W0LZ8DG4JeNissi3eY_n3-IWj96PDzcZpojtGPyY9SsASSup-fqBL25zGOvwP1NqQyDfgT3bCRxWeCRmC8e8Nie7YgA5RcGbZasvgIMerIlTov8nQCv1N76Edrq1o5YLTTuPo06PVLbJ5C5fc2SSCVYpA2TYcy4L1imqNaUBYlKuIhBGALHJc04p9r3hcqYBVNCRoJzadEsO0WNvMjNGcKCJlJHnhRC6SATMdcmtsldRZyFsfSDc9SC0Zp9VB0yZpuBuvhj_y3a741e0ln6POhfogMITkW_ukKNcrk212hPfZbz1fLGRfgbTQOnUg
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%3Abook&rft.genre=proceeding&rft.title=Proceedings+%28IEEE+Computer+Society+Conference+on+Computer+Vision+and+Pattern+Recognition.+Online%29&rft.atitle=Transfer+Knowledge+from+Head+to+Tail%3A+Uncertainty+Calibration+under+Long-tailed+Distribution&rft.au=Chen%2C+Jiahao&rft.au=Su%2C+Bing&rft.date=2023-06-01&rft.pub=IEEE&rft.eissn=1063-6919&rft.spage=19978&rft.epage=19987&rft_id=info:doi/10.1109%2FCVPR52729.2023.01913&rft.externalDocID=10204180