Research on Point-wise Gated Deep Networks

doi:10.1016/j.asoc.2016.08.056

CORC > 计算技术研究所 > 中国科学院计算技术研究所 > 中国科学院计算技术研究所期刊论文 > 英文

	Research on Point-wise Gated Deep Networks
	Zhang, Nan 1,2; Ding, Shifei 1,2; Zhang, Jian 1,2; Xue, Yu 3
刊名	APPLIED SOFT COMPUTING
	2017-03-01
卷号	52 页码:1210-1221
关键词	Restricted boltzmann machine Deep Boltzmann machine Deep belief network Dropout Weight uncertainty Feature selection
ISSN号	1568-4946
DOI	10.1016/j.asoc.2016.08.056
英文摘要	Stacking Restricted Boltzmann Machines (RBM) to create deep networks, such as Deep Belief Networks( DBN) and Deep Boltzmann Machines (DBM), has become one of the most important research fields indeep learning. DBM and DBN provide state-of-the-art results in many fields such as image recognition, but they don't show better learning abilities than RBM when dealing with data containing irrelevant patterns. Point-wise Gated Restricted Boltzmann Machines (pgRBM) can effectively find the task-relevant patterns from data containing irrelevant patterns and thus achieve satisfied classification results. For the limitations of the DBN and the DBM in the processing of data containing irrelevant patterns, we introduce the pgRBM into the DBN and the DBM and present Point-wise Gated Deep Belief Networks (pgDBN) and Pointwise Gated Deep Boltzmann Machines (pgDBM). The pgDBN and the pgDBM both utilize the pgRBM instead of the RBM to pre-train the weights connecting the networks' the visible layer and the hidden layer, and apply the pgRBM learning task-relevant data subset for traditional networks. Then, this paper discusses the validity that dropout and weight uncertainty methods are developed to prevent over fitting in pgRBMs, pgDBNs, and pgDBMs networks. Experimental results on MNIST variation datasets show that the pgDBN and the pgDBM are effective deep neural networks learning (C) 2016 Elsevier B.V. All rights reserved.
资助项目	National Natural Science Foundations of China[61379101] ; National Natural Science Foundations of China[61672522] ; National Key Basic Research Program of China[2013CB329502] ; Graduate Student Scientific Research Innovation Project in Jiangsu Province[KYZZ16_0215] ; Priority Academic Program Development of Jiangsu Higer Education Institutions (PAPD) ; Jiangsu Collaborative Innovation Center on Atmospheric Environment and Equipment Technology(CICAEET)
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER SCIENCE BV
WOS记录号	WOS:000395896500089
内容类型	期刊论文
源URL	[http://119.78.100.204/handle/2XEOYT63/7365]
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Ding, Shifei
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100090, Peoples R China 2.China Univ Min & Technol, Sch Comp Sci & Technol, Xuzhou 221116, Peoples R China 3.Nanjing Univ Informat Sci & Technol, Sch Comp & Software, Nanjing 210044, Jiangsu, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Nan,Ding, Shifei,Zhang, Jian,et al. Research on Point-wise Gated Deep Networks[J]. APPLIED SOFT COMPUTING,2017,52:1210-1221.
APA	Zhang, Nan,Ding, Shifei,Zhang, Jian,&Xue, Yu.(2017).Research on Point-wise Gated Deep Networks.APPLIED SOFT COMPUTING,52,1210-1221.
MLA	Zhang, Nan,et al."Research on Point-wise Gated Deep Networks".APPLIED SOFT COMPUTING 52(2017):1210-1221.