MMT: Cross Domain Few-Shot Learning via Meta-Memory Transfer

doi:10.1109/TPAMI.2023.3306352

CASIA OpenIR > 模式识别实验室

	MMT: Cross Domain Few-Shot Learning via Meta-Memory Transfer
	Wang, Wenjian 1,2,3; Duan, Lijuan 1,2,3; Wang, Yuxi4 ; Fan, Junsong; Zhang, Zhaoxiang4,5,6
发表期刊	IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
ISSN	0162-8828
	2023-12-01
卷号	45 期号:12 页码:15018-15035
通讯作者	Zhang, Zhaoxiang(zhaoxiang.zhang@ia.ac.cn)
摘要	Few-shot learning aims to recognize novel categories solely relying on a few labeled samples, with existing few-shot methods primarily focusing on the categories sampled from the same distribution. Nevertheless, this assumption cannot always be ensured, and the actual domain shift problem significantly reduces the performance of few-shot learning. To remedy this problem, we investigate an interesting and challenging cross-domain few-shot learning task, where the training and testing tasks employ different domains. Specifically, we propose aMeta-Memory scheme to bridge the domain gap between source and target domains, leveraging style-memory and content-memory components. The former stores intra-domain style information from source domain instances and provides a richer feature distribution. The latter stores semantic information through exploration of knowledge of different categories. Under the contrastive learning strategy, our model effectively alleviates the cross-domain problem in few-shot learning. Extensive experiments demonstrate that our proposed method achieves state-of-the-art performance on cross-domain few-shot semantic segmentation tasks on the COCO-20(i), PASCAL-5(i), FSS-1000, and SUIM datasets and positively affects few-shot classification tasks on Meta-Dataset.
关键词	Memory few-shot learning semantic segmentation cross-domain
DOI	10.1109/TPAMI.2023.3306352
关键词[WOS]	NETWORKS
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[61836014] ; National Natural Science Foundation of China[U21B2042] ; National Natural Science Foundation of China[62072457] ; National Natural Science Foundation of China[62006231] ; National Natural Science Foundation of China[62176009] ; Major Project for New Generation of AI[2018AAA0100400]
项目资助者	National Natural Science Foundation of China ; Major Project for New Generation of AI
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:001130146400060
出版者	IEEE COMPUTER SOC
引用统计
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/55516
专题	模式识别实验室
通讯作者	Zhang, Zhaoxiang
作者单位	1.Beijing Univ Technol, Fac Informat Technol, Beijing 100021, Peoples R China 2.Beijing Key Lab Trusted Comp, Beijing 100124, Peoples R China 3.Natl Engn Lab Key Technol Informat Secur Level Pr, Beijing 100124, Peoples R China 4.Chinese Acad Sci, Hong Kong Inst Sci & Innovat, Ctr Artificial Intelligence & Robot, Hong Kong 999077, Peoples R China 5.Chinese Acad Sci CASIA, Inst Automat, Natl Lab Pattern Recognit NLPR, Beijing 100045, Peoples R China 6.Univ Chinese Acad Sci UCAS, Beijing 101408, Peoples R China
通讯作者单位	模式识别国家重点实验室
推荐引用方式 GB/T 7714	Wang, Wenjian,Duan, Lijuan,Wang, Yuxi,et al. MMT: Cross Domain Few-Shot Learning via Meta-Memory Transfer[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2023,45(12):15018-15035.
APA	Wang, Wenjian,Duan, Lijuan,Wang, Yuxi,Fan, Junsong,&Zhang, Zhaoxiang.(2023).MMT: Cross Domain Few-Shot Learning via Meta-Memory Transfer.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,45(12),15018-15035.
MLA	Wang, Wenjian,et al."MMT: Cross Domain Few-Shot Learning via Meta-Memory Transfer".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 45.12(2023):15018-15035.