Dataset Bias in Few-Shot Image Recognition

doi:10.1109/TPAMI.2022.3153611

	Dataset Bias in Few-Shot Image Recognition
	Jiang, Shuqiang 1,2; Zhu, Yaohui 1,2; Liu, Chenlong 1,2; Song, Xinhang 1,2; Li, Xiangyang 1,2; Min, Weiqing 1,2
	2023
发表期刊	IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
ISSN	0162-8828
卷号	45 期号:1 页码:229-246
摘要	The goal of few-shot image recognition (FSIR) is to identify novel categories with a small number of annotated samples by exploiting transferable knowledge from training data (base categories). Most current studies assume that the transferable knowledge can be well used to identify novel categories. However, such transferable capability may be impacted by the dataset bias, and this problem has rarely been investigated before. Besides, most of few-shot learning methods are biased to different datasets, which is also an important issue that needs to be investigated deeply. In this paper, we first investigate the impact of transferable capabilities learned from base categories. Specifically, we use the relevance to measure relationships between base categories and novel categories. Distributions of base categories are depicted via the instance density and category diversity. The FSIR model learns better transferable knowledge from relevant training data. In the relevant data, dense instances or diverse categories can further enrich the learned knowledge. Experimental results on different sub-datasets of Imagenet demonstrate category relevance, instance density and category diversity can depict transferable bias from distributions of base categories. Second, we investigate performance differences on different datasets from the aspects of dataset structures and different few-shot learning methods. Specifically, we introduce image complexity, intra-concept visual consistency, and inter-concept visual similarity to quantify characteristics of dataset structures. We use these quantitative characteristics and eight few-shot learning methods to analyze performance differences on multiple datasets. Based on the experimental analysis, some insightful observations are obtained from the perspective of both dataset structures and few-shot learning methods. We hope these observations are useful to guide future few-shot learning research on new datasets or tasks. Our data is available at http://123.57.42.89/dataset-bias/dataset-bias.html.
关键词	Dataset bias few-shot image recognition knowledge transfer meta-learning
DOI	10.1109/TPAMI.2022.3153611
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[62125207] ; National Natural Science Foundation of China[U1936203] ; National Natural Science Foundation of China[62032022] ; National Natural Science Foundation of China[U19B2040] ; Beijing Natural Science Foundation[Z190020]
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:000899419900015
出版者	IEEE COMPUTER SOC
引用统计	被引频次：9[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/2XEOYT63/20140
专题	中国科学院计算技术研究所期刊论文
通讯作者	Jiang, Shuqiang
作者单位	1.Chinese Acad Sci, CAS, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Jiang, Shuqiang,Zhu, Yaohui,Liu, Chenlong,et al. Dataset Bias in Few-Shot Image Recognition[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2023,45(1):229-246.
APA	Jiang, Shuqiang,Zhu, Yaohui,Liu, Chenlong,Song, Xinhang,Li, Xiangyang,&Min, Weiqing.(2023).Dataset Bias in Few-Shot Image Recognition.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,45(1),229-246.
MLA	Jiang, Shuqiang,et al."Dataset Bias in Few-Shot Image Recognition".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 45.1(2023):229-246.