Gradient-Aligned convolution neural network

doi:10.1016/j.patcog.2021.108354

	Gradient-Aligned convolution neural network
	Hao, You 1,2,5; Hu, Ping 3; Li, Shirui 4; Udupa, Jayaram K.2,4; Tong, Yubing 2; Li, Hua 1
	2022-02-01
发表期刊	PATTERN RECOGNITION
ISSN	0031-3203
卷号	122 页码:10
摘要	Although Convolution Neural Networks (CNN) have achieved great success in many applications of computer vision in recent years, rotation invariance is still a difficult problem for CNN. Especially for some images, the content can appear in the image at any angle of rotation, such as medical images, microscopic images, remote sensing images and astronomical images. In this paper, we propose a novel convolution operation, called Gradient-Aligned Convolution (GAConv), which can help CNN achieve rotation invariance by replacing vanilla convolutions in CNN. GAConv is implemented with a prior pixel-level gradient alignment operation before regular convolution. With GAConv, Gradient-Aligned CNN (GACNN) can achieve rotation invariance without any data augmentation, feature-map augmentation, and filter enrichment. In GACNN, rotation invariance does not learn from the training set, but bases on the network model. Different from the vanilla CNN, GACNN will output invariant results for all rotated versions of an object, no matter whether the network is trained or not. This means that we only need to train the network with one canonical version of the object and all other rotated versions of this object should be recognized with the same accuracy. Classification experiments have been conducted to evaluate GACNN compared with some rotation invariant approaches. GACNN achieved the best results on the 360 degrees rotated test set of MNIST-rotation, Plankton-sub-rotation, and Galaxy Zoo 2. (C) 2021 Elsevier Ltd. All rights reserved.
关键词	Gradient alignment Rotation equivariant convolution Rotation invariant neural network
DOI	10.1016/j.patcog.2021.108354
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2017YFB1002703] ; National Key Basic Research Program of China[2015CB554507] ; National Natural Science Foundation of China[61379082] ; China Scholarship Council (CSC)
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:000704891800007
出版者	ELSEVIER SCI LTD
引用统计	被引频次：3[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/2XEOYT63/16960
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Hao, You
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100190, Peoples R China 2.Univ Penn, Dept Radiol, Med Image Proc Grp, Philadelphia, PA 19104 USA 3.Microsoft Inc, Beijing 100190, Peoples R China 4.Baidu Inc, Beijing 100085, Peoples R China 5.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Hao, You,Hu, Ping,Li, Shirui,et al. Gradient-Aligned convolution neural network[J]. PATTERN RECOGNITION,2022,122:10.
APA	Hao, You,Hu, Ping,Li, Shirui,Udupa, Jayaram K.,Tong, Yubing,&Li, Hua.(2022).Gradient-Aligned convolution neural network.PATTERN RECOGNITION,122,10.
MLA	Hao, You,et al."Gradient-Aligned convolution neural network".PATTERN RECOGNITION 122(2022):10.