Multimodal deep generative adversarial models for scalable doubly semi-supervised learning

doi:10.1016/j.inffus.2020.11.003

CORC > 自动化研究所 > 中国科学院自动化研究所 > 类脑智能研究中心 > 神经计算及脑机交互

	Multimodal deep generative adversarial models for scalable doubly semi-supervised learning
	Du, Changde 1,3,5,6; Du, Changying 4; He, Huiguang 1,2,3,6
刊名	INFORMATION FUSION
	2021-04-01
卷号	68 页码:118-130
关键词	Multiview learning Multimodal fusion Generative adversarial networks Deep generative models Semi-supervised learning
ISSN号	1566-2535
DOI	10.1016/j.inffus.2020.11.003
通讯作者	He, Huiguang(huiguang.he@ia.ac.cn)
英文摘要	The comprehensive utilization of incomplete multi-modality data is a difficult problem with strong practical value. Most of the previous multimodal learning algorithms require massive training data with complete modalities and annotated labels, which greatly limits their practicality. Although some existing algorithms can be used to complete the data imputation task, they still have two disadvantages: (1) they cannot control the semantics of the imputed modalities accurately; and (2) they need to establish multiple independent converters between any two modalities when extended to multimodal cases. To overcome these limitations, we propose a novel doubly semi-supervised multimodal learning (DSML) framework. Specifically, DSML uses a modality-shared latent space and multiple modality-specific generators to associate multiple modalities together. Here we divided the shared latent space into two independent parts, the semantic labels and the semantic-free styles, which allows us to easily control the semantics of generated samples. In addition, each modality has its own separate encoder and classifier to infer the corresponding semantic and semantic-free latent variables. The above DSML framework can be adversarially trained by using our specially designed softmax-based discriminators. Large amounts of experimental results show that the DSML obtains better performance than the baselines on three tasks, including semi-supervised classification, missing modality imputation and cross-modality retrieval.
资助项目	National Natural Science Foundation of China[61976209] ; National Natural Science Foundation of China[62020106015] ; National Natural Science Foundation of China[61906188] ; Chinese Academy of Sciences (CAS) International Collaboration Key, China[173211KYSB20190024] ; Strategic Priority Research Program of CAS, China[XDB32040000]
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER
WOS记录号	WOS:000616409600009
资助机构	National Natural Science Foundation of China ; Chinese Academy of Sciences (CAS) International Collaboration Key, China ; Strategic Priority Research Program of CAS, China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/43265]
专题	类脑智能研究中心_神经计算及脑机交互
通讯作者	He, Huiguang
作者单位	1.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 2.Chinese Acad Sci, Ctr Excellence Brain Sci & Intelligence Technol, Beijing 100190, Peoples R China 3.Chinese Acad Sci, Inst Automat, Res Ctr Brain Inspired Intelligence, Beijing 100190, Peoples R China 4.Huawei Noahs Ark Lab, Beijing 100085, Peoples R China 5.Huawei Cloud BU EI Innovat Lab, Beijing 100085, Peoples R China 6.Univ Chinese Acad Sci, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Du, Changde,Du, Changying,He, Huiguang. Multimodal deep generative adversarial models for scalable doubly semi-supervised learning[J]. INFORMATION FUSION,2021,68:118-130.
APA	Du, Changde,Du, Changying,&He, Huiguang.(2021).Multimodal deep generative adversarial models for scalable doubly semi-supervised learning.INFORMATION FUSION,68,118-130.
MLA	Du, Changde,et al."Multimodal deep generative adversarial models for scalable doubly semi-supervised learning".INFORMATION FUSION 68(2021):118-130.