CASIA OpenIR  > 多模态人工智能系统全国重点实验室
Towards Prior Gap and Representation Gap for Long-tailed Recognition, Pattern Recognition
Zhang Ming-Liang1,2; Zhang Xu-Yao1,2; Wang Chang1,2; Liu Cheng-Lin1,2
Source PublicationPattern Recognition
ISSN0031-3203
2023-01
Volume133Issue:109012Pages:109012
Subtype模式识别,机器学习
Abstract

Most deep learning models are elaborately designed for balanced datasets, and thus they inevitably suffer performance degradation in practical long-tailed recognition tasks, especially to the minority classes. There are two crucial issues in learning from imbalanced datasets: skew decision boundary and unrepresentative feature space. In this work, we establish a theoretical framework to analyze the sources of these two issues from Bayesian perspective, and find that they are closely related to the prior gap and the representation gap, respectively. Under this framework, we show that existing long-tailed recognition methods manage to remove either the prior gap or the presentation gap. Different from these methods, we propose to simultaneously remove the two gaps to achieve more accurate long-tailed recognition. Specifically, we propose the prior calibration strategy to remove the prior gap and introduce three strategies (representative feature extraction, optimization strategy adjustment and effective sample modeling) to mitigate the representation gap. Extensive experiments on five benchmark datasets validate the superiority of our method against the state-of-the-art competitors.

KeywordLong-tailed learning Prior gap Representation gap Image recognition
DOI10.1016/j.patcog.2022.109012
Indexed BySCI
Language英语
IS Representative Paper
Sub direction classification模式识别基础
planning direction of the national heavy laboratory其他
Paper associated data
Citation statistics
Cited Times:7[WOS]   [WOS Record]     [Related Records in WOS]
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/55699
Collection多模态人工智能系统全国重点实验室
Corresponding AuthorZhang Ming-Liang
Affiliation1.National Laboratory of Pattern Recognition (NLPR), Institute of Automation of Chinese Academy of Sciences
2.School of Artificial Intelligence, University of Chinese Academy of Sciences
First Author AffilicationChinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
Corresponding Author AffilicationChinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
Recommended Citation
GB/T 7714
Zhang Ming-Liang,Zhang Xu-Yao,Wang Chang,et al. Towards Prior Gap and Representation Gap for Long-tailed Recognition, Pattern Recognition[J]. Pattern Recognition,2023,133(109012):109012.
APA Zhang Ming-Liang,Zhang Xu-Yao,Wang Chang,&Liu Cheng-Lin.(2023).Towards Prior Gap and Representation Gap for Long-tailed Recognition, Pattern Recognition.Pattern Recognition,133(109012),109012.
MLA Zhang Ming-Liang,et al."Towards Prior Gap and Representation Gap for Long-tailed Recognition, Pattern Recognition".Pattern Recognition 133.109012(2023):109012.
Files in This Item: Download All
File Name/Size DocType Version Access License
1-s2.0-S003132032200(2258KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Zhang Ming-Liang]'s Articles
[Zhang Xu-Yao]'s Articles
[Wang Chang]'s Articles
Baidu academic
Similar articles in Baidu academic
[Zhang Ming-Liang]'s Articles
[Zhang Xu-Yao]'s Articles
[Wang Chang]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Zhang Ming-Liang]'s Articles
[Zhang Xu-Yao]'s Articles
[Wang Chang]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 1-s2.0-S0031320322004927-main.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.