Lateral interaction by Laplacian-based graph smoothing for deep neural networks

doi:10.1049/cit2.12265

CORC > 自动化研究所 > 中国科学院自动化研究所 > 多模态人工智能系统全国重点实验室

	Lateral interaction by Laplacian-based graph smoothing for deep neural networks
	Chen, Jianhui 1,2,4; Wang, Zuoren 3,4,5; Liu, Cheng-Lin 1,2
刊名	CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY
	2023-08-29
页码	18
关键词	artificial neural networks biologically plausible Laplacian-based graph smoothing lateral interaction machine learning
ISSN号	2468-6557
DOI	10.1049/cit2.12265
通讯作者	Wang, Zuoren(zuorenwang@ion.ac.cn) ; Liu, Cheng-Lin(liucl@nlpr.ia.ac.cn)
英文摘要	Lateral interaction in the biological brain is a key mechanism that underlies higher cognitive functions. Linear self-organising map (SOM) introduces lateral interaction in a general form in which signals of any modality can be used. Some approaches directly incorporate SOM learning rules into neural networks, but incur complex operations and poor extendibility. The efficient way to implement lateral interaction in deep neural networks is not well established. The use of Laplacian Matrix-based Smoothing (LS) regularisation is proposed for implementing lateral interaction in a concise form. The authors' derivation and experiments show that lateral interaction implemented by SOM model is a special case of LS-regulated k-means, and they both show the topology-preserving capability. The authors also verify that LS-regularisation can be used in conjunction with the end-to-end training paradigm in deep auto-encoders. Additionally, the benefits of LS-regularisation in relaxing the requirement of parameter initialisation in various models and improving the classification performance of prototype classifiers are evaluated. Furthermore, the topologically ordered structure introduced by LS-regularisation in feature extractor can improve the generalisation performance on classification tasks. Overall, LS-regularisation is an effective and efficient way to implement lateral interaction and can be easily extended to different models.
资助项目	National Natural Science Foundation of China (NSFC)[61836014] ; STI2030-Major Projects[2022ZD0205100] ; Strategic Priority Research Program of Chinese Academy of Science[XDB32010300] ; Shanghai Municipal Science and Technology Major Project[2018SHZDZX05] ; Innovation Academy of Artificial Intelligence, Chinese Academy of Sciences
WOS关键词	VISUAL-CORTEX ; CLASSIFICATION ; MODEL
WOS研究方向	Computer Science
语种	英语
出版者	WILEY
WOS记录号	WOS:001060157300001
资助机构	National Natural Science Foundation of China (NSFC) ; STI2030-Major Projects ; Strategic Priority Research Program of Chinese Academy of Science ; Shanghai Municipal Science and Technology Major Project ; Innovation Academy of Artificial Intelligence, Chinese Academy of Sciences
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/53176]
专题	多模态人工智能系统全国重点实验室
通讯作者	Wang, Zuoren; Liu, Cheng-Lin
作者单位	1.Univ Chinese Acad Sci, Beijing, Peoples R China 2.Chinese Acad Sci, State Key Lab Multimodal Artificial Intelligence S, Inst Automat, Beijing, Peoples R China 3.ShanghaiTech Univ, Sch Life Sci & Technol, Shanghai, Peoples R China 4.Chinese Acad Sci, Inst Neurosci, Ctr Excellence Brain Sci & Intelligence Technol, State Key Lab Neurosci, Shanghai, Peoples R China 5.Univ Chinese Acad Sci, Sch Future Technol, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Chen, Jianhui,Wang, Zuoren,Liu, Cheng-Lin. Lateral interaction by Laplacian-based graph smoothing for deep neural networks[J]. CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY,2023:18.
APA	Chen, Jianhui,Wang, Zuoren,&Liu, Cheng-Lin.(2023).Lateral interaction by Laplacian-based graph smoothing for deep neural networks.CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY,18.
MLA	Chen, Jianhui,et al."Lateral interaction by Laplacian-based graph smoothing for deep neural networks".CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY (2023):18.