Multiple graph regularized nonnegative matrix factorization

Jim Jing Yan Wang, Halima Bensmail, Xin Gao

Research output: Contribution to journalArticle

75 Citations (Scopus)

Abstract

Non-negative matrix factorization (NMF) has been widely used as a data representation method based on components. To overcome the disadvantage of NMF in failing to consider the manifold structure of a data set, graph regularized NMF (GrNMF) has been proposed by Cai et al. by constructing an affinity graph and searching for a matrix factorization that respects graph structure. Selecting a graph model and its corresponding parameters is critical for this strategy. This process is usually carried out by cross-validation or discrete grid search, which are time consuming and prone to overfitting. In this paper, we propose a GrNMF, called MultiGrNMF, in which the intrinsic manifold is approximated by a linear combination of several graphs with different models and parameters inspired by ensemble manifold regularization. Factorization metrics and linear combination coefficients of graphs are determined simultaneously within a unified object function. They are alternately optimized in an iterative algorithm, thus resulting in a novel data representation algorithm. Extensive experiments on a protein subcellular localization task and an Alzheimer's disease diagnosis task demonstrate the effectiveness of the proposed algorithm.

Original languageEnglish
Pages (from-to)2840-2847
Number of pages8
JournalPattern Recognition
Volume46
Issue number10
DOIs
Publication statusPublished - 1 Oct 2013

Fingerprint

Factorization
Proteins
Experiments

Keywords

  • Data representation
  • Ensemble manifold regularization
  • Graph Laplacian
  • Nonnegative matrix factorization

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Multiple graph regularized nonnegative matrix factorization. / Wang, Jim Jing Yan; Bensmail, Halima; Gao, Xin.

In: Pattern Recognition, Vol. 46, No. 10, 01.10.2013, p. 2840-2847.

Research output: Contribution to journalArticle

Wang, Jim Jing Yan ; Bensmail, Halima ; Gao, Xin. / Multiple graph regularized nonnegative matrix factorization. In: Pattern Recognition. 2013 ; Vol. 46, No. 10. pp. 2840-2847.
@article{15d021d01f6341928f80b6038f4952c1,
title = "Multiple graph regularized nonnegative matrix factorization",
abstract = "Non-negative matrix factorization (NMF) has been widely used as a data representation method based on components. To overcome the disadvantage of NMF in failing to consider the manifold structure of a data set, graph regularized NMF (GrNMF) has been proposed by Cai et al. by constructing an affinity graph and searching for a matrix factorization that respects graph structure. Selecting a graph model and its corresponding parameters is critical for this strategy. This process is usually carried out by cross-validation or discrete grid search, which are time consuming and prone to overfitting. In this paper, we propose a GrNMF, called MultiGrNMF, in which the intrinsic manifold is approximated by a linear combination of several graphs with different models and parameters inspired by ensemble manifold regularization. Factorization metrics and linear combination coefficients of graphs are determined simultaneously within a unified object function. They are alternately optimized in an iterative algorithm, thus resulting in a novel data representation algorithm. Extensive experiments on a protein subcellular localization task and an Alzheimer's disease diagnosis task demonstrate the effectiveness of the proposed algorithm.",
keywords = "Data representation, Ensemble manifold regularization, Graph Laplacian, Nonnegative matrix factorization",
author = "Wang, {Jim Jing Yan} and Halima Bensmail and Xin Gao",
year = "2013",
month = "10",
day = "1",
doi = "10.1016/j.patcog.2013.03.007",
language = "English",
volume = "46",
pages = "2840--2847",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "10",

}

TY - JOUR

T1 - Multiple graph regularized nonnegative matrix factorization

AU - Wang, Jim Jing Yan

AU - Bensmail, Halima

AU - Gao, Xin

PY - 2013/10/1

Y1 - 2013/10/1

N2 - Non-negative matrix factorization (NMF) has been widely used as a data representation method based on components. To overcome the disadvantage of NMF in failing to consider the manifold structure of a data set, graph regularized NMF (GrNMF) has been proposed by Cai et al. by constructing an affinity graph and searching for a matrix factorization that respects graph structure. Selecting a graph model and its corresponding parameters is critical for this strategy. This process is usually carried out by cross-validation or discrete grid search, which are time consuming and prone to overfitting. In this paper, we propose a GrNMF, called MultiGrNMF, in which the intrinsic manifold is approximated by a linear combination of several graphs with different models and parameters inspired by ensemble manifold regularization. Factorization metrics and linear combination coefficients of graphs are determined simultaneously within a unified object function. They are alternately optimized in an iterative algorithm, thus resulting in a novel data representation algorithm. Extensive experiments on a protein subcellular localization task and an Alzheimer's disease diagnosis task demonstrate the effectiveness of the proposed algorithm.

AB - Non-negative matrix factorization (NMF) has been widely used as a data representation method based on components. To overcome the disadvantage of NMF in failing to consider the manifold structure of a data set, graph regularized NMF (GrNMF) has been proposed by Cai et al. by constructing an affinity graph and searching for a matrix factorization that respects graph structure. Selecting a graph model and its corresponding parameters is critical for this strategy. This process is usually carried out by cross-validation or discrete grid search, which are time consuming and prone to overfitting. In this paper, we propose a GrNMF, called MultiGrNMF, in which the intrinsic manifold is approximated by a linear combination of several graphs with different models and parameters inspired by ensemble manifold regularization. Factorization metrics and linear combination coefficients of graphs are determined simultaneously within a unified object function. They are alternately optimized in an iterative algorithm, thus resulting in a novel data representation algorithm. Extensive experiments on a protein subcellular localization task and an Alzheimer's disease diagnosis task demonstrate the effectiveness of the proposed algorithm.

KW - Data representation

KW - Ensemble manifold regularization

KW - Graph Laplacian

KW - Nonnegative matrix factorization

UR - http://www.scopus.com/inward/record.url?scp=84878017859&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878017859&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2013.03.007

DO - 10.1016/j.patcog.2013.03.007

M3 - Article

VL - 46

SP - 2840

EP - 2847

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 10

ER -