From Class-Specific to Class-Mixture: Cascaded Feature Representations via Restricted Boltzmann Machine Learning

doi:10.1109/ACCESS.2018.2878553

	From Class-Specific to Class-Mixture: Cascaded Feature Representations via Restricted Boltzmann Machine Learning
	Xie, Guo-Sen 1,2; Jin, Xiao-Bo 3; Zhang, Xu-Yao4 ; Zang, Shao-Fei 1,2; Yang, Chunlei 1,2; Wang, Zhiheng 5; Pu, Jiexin 1,2
刊名	IEEE ACCESS
	2018
卷号	6 页码:69393-69406
关键词	Feature extraction class-specific RBM feature learning
ISSN号	2169-3536
DOI	10.1109/ACCESS.2018.2878553
通讯作者	Xie, Guo-Sen(gsxiehm@gmail.com)
英文摘要	In this paper, we propose two kinds of feature extracting frameworks that can extract cascaded class-specific and class-mixture features, respectively, by taking the restricted Boltzmann machine (RBM) as the basic building blocks; we further call them as a CS-RBM and CM-RBM feature extractor. The discriminations of features from both CS-RBM and CM-RBM are verified better than the class-independent (traditional) RBM (CI-RBM) feature extractor. As one mini-batch samples are randomly selected from all classes during the training phase of the traditional RBM, which can make that the above mini-batch data contain easy-confusing samples from different categories. Therefore, the features from CI-RBM are difficult to distinguish these samples from the confused categories. CS-RBM and CM-RBM can overcome the above sample confusing problem efficiently and effectively. To cope with the real-valued input samples, we further extend the binary RBM to Gaussian-Bernoulli RBM (GBRBM), leading to the CS-GBRBM (CM-GBRBM) feature extracting framework. Experiments on binary datasets, i.e., MNIST and USPS, scene image dataset (Scene-15), and object image dataset (Coil-100), well verify the above facts and show the competitive results.
资助项目	National Natural Science Foundation of China[61702163] ; National Natural Science Foundation of China[61103138] ; National Natural Science Foundation of China[U1804326] ; Henan University Scientific and Technological Innovation Team Support Program[19IRTSTHN012] ; Fundamental Research Funds for the Henan Provincial Colleges and Universities in the Henan University of Technology[2016RCJH06]
WOS关键词	CLASSIFICATION ; RECOGNITION ; MODEL
WOS研究方向	Computer Science ; Engineering ; Telecommunications
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000452598800001
资助机构	National Natural Science Foundation of China ; Henan University Scientific and Technological Innovation Team Support Program ; Fundamental Research Funds for the Henan Provincial Colleges and Universities in the Henan University of Technology
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/25671]
专题	中国科学院自动化研究所
通讯作者	Xie, Guo-Sen
作者单位	1.Henan Univ Sci & Technol, Dept Informat, Engn Coll, Luoyang 471023, Peoples R China 2.Henan Univ Sci & Technol, Henan Joint Int Res Lab Image Proc & Intelligent, Luoyang 471023, Peoples R China 3.Henan Univ Technol, Sch Informat Sci & Engn, Zhengzhou 450001, Henan, Peoples R China 4.Chinese Acad Sci, NLPR, Inst Automat, Beijing 100190, Peoples R China 5.Henan Polytech Univ, Dept Comp Sci & Technol, Jiaozuo 454003, Peoples R China
推荐引用方式 GB/T 7714	Xie, Guo-Sen,Jin, Xiao-Bo,Zhang, Xu-Yao,et al. From Class-Specific to Class-Mixture: Cascaded Feature Representations via Restricted Boltzmann Machine Learning[J]. IEEE ACCESS,2018,6:69393-69406.
APA	Xie, Guo-Sen.,Jin, Xiao-Bo.,Zhang, Xu-Yao.,Zang, Shao-Fei.,Yang, Chunlei.,...&Pu, Jiexin.(2018).From Class-Specific to Class-Mixture: Cascaded Feature Representations via Restricted Boltzmann Machine Learning.IEEE ACCESS,6,69393-69406.
MLA	Xie, Guo-Sen,et al."From Class-Specific to Class-Mixture: Cascaded Feature Representations via Restricted Boltzmann Machine Learning".IEEE ACCESS 6(2018):69393-69406.