A Comprehensive Framework for Long-Tailed Learning via Pretraining and Normalization

doi:10.1109/TNNLS.2022.3192475

	A Comprehensive Framework for Long-Tailed Learning via Pretraining and Normalization
	Kang, Nan 1,2; Chang, Hong 1,2,3; Ma, Bingpeng 4; Shan, Shiguang 1,2,3
	2022-07-27
发表期刊	IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
ISSN	2162-237X
页码	13
摘要	Data in the visual world often present long-tailed distributions. However, learning high-quality representations and classifiers for imbalanced data is still challenging for data-driven deep learning models. In this work, we aim at improving the feature extractor and classifier for long-tailed recognition via contrastive pretraining and feature normalization, respectively. First, we carefully study the influence of contrastive pretraining under different conditions, showing that current self-supervised pretraining for long-tailed learning is still suboptimal in both performance and speed. We thus propose a new balanced contrastive loss and a fast contrastive initialization scheme to improve previous long-tailed pretraining. Second, based on the motivative analysis on the normalization for classifier, we propose a novel generalized normalization classifier that consists of generalized normalization and grouped learnable scaling. It outperforms traditional inner product classifier as well as cosine classifier. Both the two components proposed can improve recognition ability on tail classes without the expense of head classes. We finally build a unified framework that achieves competitive performance compared with state of the arts on several long-tailed recognition benchmarks and maintains high efficiency.
关键词	Training Tail Task analysis Head Visualization Feature extraction Data models Classifier design contrastive learning long-tailed recognition normalization
DOI	10.1109/TNNLS.2022.3192475
收录类别	SCI
语种	英语
资助项目	Natural Science Foundation of China (NSFC)[U19B2036] ; Natural Science Foundation of China (NSFC)[61976203] ; Natural Science Foundation of China (NSFC)[61876171] ; Fundamental Research Funds for the Central Universities
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Computer Science, Hardware & Architecture ; Computer Science, Theory & Methods ; Engineering, Electrical & Electronic
WOS记录号	WOS:000833050600001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：1[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/19487
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Chang, Hong
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp Sci & Technol, Beijing 100049, Peoples R China 3.Peng Cheng Lab, Shenzhen 518055, Peoples R China 4.Univ Chinese Acad Sci, Sch Comp Sci & Technol, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Kang, Nan,Chang, Hong,Ma, Bingpeng,et al. A Comprehensive Framework for Long-Tailed Learning via Pretraining and Normalization[J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,2022:13.
APA	Kang, Nan,Chang, Hong,Ma, Bingpeng,&Shan, Shiguang.(2022).A Comprehensive Framework for Long-Tailed Learning via Pretraining and Normalization.IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,13.
MLA	Kang, Nan,et al."A Comprehensive Framework for Long-Tailed Learning via Pretraining and Normalization".IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2022):13.