Nuclear norm regularized convolutional Max Pos@Top machine

Handle URI:
http://hdl.handle.net/10754/622226
Title:
Nuclear norm regularized convolutional Max Pos@Top machine
Authors:
Li, Qinfeng; Zhou, Xiaofeng; Gu, Aihua; Li, Zonghua; Liang, Ru-Ze
Abstract:
In this paper, we propose a novel classification model for the multiple instance data, which aims to maximize the number of positive instances ranked before the top-ranked negative instances. This method belongs to a recently emerged performance, named as Pos@Top. Our proposed classification model has a convolutional structure that is composed by four layers, i.e., the convolutional layer, the activation layer, the max-pooling layer and the full connection layer. In this paper, we propose an algorithm to learn the convolutional filters and the full connection weights to maximize the Pos@Top measure over the training set. Also, we try to minimize the rank of the filter matrix to explore the low-dimensional space of the instances in conjunction with the classification results. The rank minimization is conducted by the nuclear norm minimization of the filter matrix. In addition, we develop an iterative algorithm to solve the corresponding problem. We test our method on several benchmark datasets. The experimental results show the superiority of our method compared with other state-of-the-art Pos@Top maximization methods.
KAUST Department:
Physical Sciences and Engineering (PSE) Division; Materials Science and Engineering Program
Citation:
Li Q, Zhou X, Gu A, Li Z, Liang R-Z (2016) Nuclear norm regularized convolutional Max Pos@Top machine. Neural Computing and Applications. Available: http://dx.doi.org/10.1007/s00521-016-2680-2.
Publisher:
Springer Nature
Journal:
Neural Computing and Applications
Issue Date:
18-Nov-2016
DOI:
10.1007/s00521-016-2680-2
Type:
Article
ISSN:
0941-0643; 1433-3058
Sponsors:
The Foundation of modern educational technology research of Jiangsu Province (No. 2015-R-42631) and The University Natural Science Foundation of Jiangsu Province (14KJD520003).
Additional Links:
http://dx.doi.org/10.1007/s00521-016-2680-2
Appears in Collections:
Articles; Physical Sciences and Engineering (PSE) Division; Materials Science and Engineering Program

Full metadata record

DC FieldValue Language
dc.contributor.authorLi, Qinfengen
dc.contributor.authorZhou, Xiaofengen
dc.contributor.authorGu, Aihuaen
dc.contributor.authorLi, Zonghuaen
dc.contributor.authorLiang, Ru-Zeen
dc.date.accessioned2017-01-02T08:42:38Z-
dc.date.available2017-01-02T08:42:38Z-
dc.date.issued2016-11-18en
dc.identifier.citationLi Q, Zhou X, Gu A, Li Z, Liang R-Z (2016) Nuclear norm regularized convolutional Max Pos@Top machine. Neural Computing and Applications. Available: http://dx.doi.org/10.1007/s00521-016-2680-2.en
dc.identifier.issn0941-0643en
dc.identifier.issn1433-3058en
dc.identifier.doi10.1007/s00521-016-2680-2en
dc.identifier.urihttp://hdl.handle.net/10754/622226-
dc.description.abstractIn this paper, we propose a novel classification model for the multiple instance data, which aims to maximize the number of positive instances ranked before the top-ranked negative instances. This method belongs to a recently emerged performance, named as Pos@Top. Our proposed classification model has a convolutional structure that is composed by four layers, i.e., the convolutional layer, the activation layer, the max-pooling layer and the full connection layer. In this paper, we propose an algorithm to learn the convolutional filters and the full connection weights to maximize the Pos@Top measure over the training set. Also, we try to minimize the rank of the filter matrix to explore the low-dimensional space of the instances in conjunction with the classification results. The rank minimization is conducted by the nuclear norm minimization of the filter matrix. In addition, we develop an iterative algorithm to solve the corresponding problem. We test our method on several benchmark datasets. The experimental results show the superiority of our method compared with other state-of-the-art Pos@Top maximization methods.en
dc.description.sponsorshipThe Foundation of modern educational technology research of Jiangsu Province (No. 2015-R-42631) and The University Natural Science Foundation of Jiangsu Province (14KJD520003).en
dc.publisherSpringer Natureen
dc.relation.urlhttp://dx.doi.org/10.1007/s00521-016-2680-2en
dc.subjectMultivariate performance measureen
dc.subjectPositives at Topen
dc.subjectConvolutional networken
dc.subjectMultiple instancesen
dc.subjectNuclear norm regularizationen
dc.titleNuclear norm regularized convolutional Max Pos@Top machineen
dc.typeArticleen
dc.contributor.departmentPhysical Sciences and Engineering (PSE) Divisionen
dc.contributor.departmentMaterials Science and Engineering Programen
dc.identifier.journalNeural Computing and Applicationsen
dc.contributor.institutionCollege of Computer and Information, Hohai University, Nanjing, Chinaen
dc.contributor.institutionDepartment of Basic, Jinling Institute of Technology, Nanjing, Chinaen
kaust.authorLiang, Ru-Zeen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.