Rethinking class orders and transferability in class incremental learning

doi:10.1016/j.patrec.2022.07.014

	Rethinking class orders and transferability in class incremental learning
	He, Chen 1,2; Wang, Ruiping 1,2; Chen, Xilin 1,2
	2022-09-01
发表期刊	PATTERN RECOGNITION LETTERS
ISSN	0167-8655
卷号	161 页码:67-73
摘要	Class Incremental Learning (CIL), an indispensable ability for open-world applications such as service robots, has received increasing attention in recent years. Although many CIL methods sprouted out, re-searchers usually adopt default class orders, leaving the characteristics of different class orders less vis-ited. In this paper, we rethink class orders in CIL from the following aspects: first, we show from prelimi-nary studies that class orders do have an impact on the performance, and mainstream episodic memory -based CIL methods generally favor an interleaved way of arranging class orders; then, we interpret the phenomena above with transferability and propose transferability measures of class orders, which are in line with the method performance under different class orders; based on that, we propose a Class Order Search Algorithm (COSA) to obtain an optimal class order by finding which one has almost the high-est transferability. Experiments on Group ImageNet and iNaturalist verify the importance of class orders in CIL methods, and demonstrate the effectiveness of our proposed transferability measures and COSA. These findings may help raise more attention to the hardly visited class orders in CIL. (c) 2022 Elsevier B.V. All rights reserved.
关键词	Transferability Class incremental learning Class order
DOI	10.1016/j.patrec.2022.07.014
收录类别	SCI
语种	英语
资助项目	Natural Science Foundation of China[U21B2025] ; Natural Science Foundation of China[U19B2036] ; Natural Science Foundation of China[61922080] ; National Key R&D Program of China[2021ZD0111901]
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:000861037800010
出版者	ELSEVIER
引用统计	被引频次：4[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/2XEOYT63/19809
专题	中国科学院计算技术研究所期刊论文
通讯作者	Wang, Ruiping
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, CAS, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	He, Chen,Wang, Ruiping,Chen, Xilin. Rethinking class orders and transferability in class incremental learning[J]. PATTERN RECOGNITION LETTERS,2022,161:67-73.
APA	He, Chen,Wang, Ruiping,&Chen, Xilin.(2022).Rethinking class orders and transferability in class incremental learning.PATTERN RECOGNITION LETTERS,161,67-73.
MLA	He, Chen,et al."Rethinking class orders and transferability in class incremental learning".PATTERN RECOGNITION LETTERS 161(2022):67-73.