Diagonal principal component analysis with non-greedy ℓ1-norm maximization for face recognition

Qiang Yu; Rong Wang; Xiaojun Yang; Bing Nan Li; Minli Yao

doi:10.1016/j.neucom.2015.06.011

Diagonal principal component analysis with non-greedy ℓ₁-norm maximization for face recognition

Qiang Yu, Rong Wang, Xiaojun Yang, Bing Nan Li, Minli Yao

Research output: Contribution to journal › Article › peer-review

18 Scopus citations

Abstract

Diagonal principal component analysis (DiaPCA) is an important method for dimensionality reduction and feature extraction. It usually makes use of the ℓ₂-norm criterion for optimization, and is thus sensitive to outliers. In this paper, we present a DiaPCA with non-greedy ℓ₁-norm maximization (DiaPCA-L1 non-greedy), which is more robust to outliers. Experimental results on two benchmark datasets show the effectiveness and advantages of our proposed method.

Original language	English
Pages (from-to)	57-62
Number of pages	6
Journal	Neurocomputing
Volume	171
DOIs	https://doi.org/10.1016/j.neucom.2015.06.011
State	Published - 1 Jan 2016
Externally published	Yes

Keywords

Diagonal PCA
Face recognition
Non-greedy strategy
Principal component analysis (PCA)
ℓ-norm

Access to Document

10.1016/j.neucom.2015.06.011

Cite this

@article{9f2322f6b96948839b7448292f0c470c,

title = "Diagonal principal component analysis with non-greedy ℓ1-norm maximization for face recognition",

abstract = "Diagonal principal component analysis (DiaPCA) is an important method for dimensionality reduction and feature extraction. It usually makes use of the ℓ2-norm criterion for optimization, and is thus sensitive to outliers. In this paper, we present a DiaPCA with non-greedy ℓ1-norm maximization (DiaPCA-L1 non-greedy), which is more robust to outliers. Experimental results on two benchmark datasets show the effectiveness and advantages of our proposed method.",

keywords = "Diagonal PCA, Face recognition, Non-greedy strategy, Principal component analysis (PCA), ℓ-norm",

author = "Qiang Yu and Rong Wang and Xiaojun Yang and Li, {Bing Nan} and Minli Yao",

note = "Publisher Copyright: {\textcopyright} 2015 Elsevier B.V.",

year = "2016",

month = jan,

day = "1",

doi = "10.1016/j.neucom.2015.06.011",

language = "英语",

volume = "171",

pages = "57--62",

journal = "Neurocomputing",

issn = "0925-2312",

publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - Diagonal principal component analysis with non-greedy ℓ1-norm maximization for face recognition

AU - Yu, Qiang

AU - Wang, Rong

AU - Yang, Xiaojun

AU - Li, Bing Nan

AU - Yao, Minli

PY - 2016/1/1

Y1 - 2016/1/1

N2 - Diagonal principal component analysis (DiaPCA) is an important method for dimensionality reduction and feature extraction. It usually makes use of the ℓ2-norm criterion for optimization, and is thus sensitive to outliers. In this paper, we present a DiaPCA with non-greedy ℓ1-norm maximization (DiaPCA-L1 non-greedy), which is more robust to outliers. Experimental results on two benchmark datasets show the effectiveness and advantages of our proposed method.

AB - Diagonal principal component analysis (DiaPCA) is an important method for dimensionality reduction and feature extraction. It usually makes use of the ℓ2-norm criterion for optimization, and is thus sensitive to outliers. In this paper, we present a DiaPCA with non-greedy ℓ1-norm maximization (DiaPCA-L1 non-greedy), which is more robust to outliers. Experimental results on two benchmark datasets show the effectiveness and advantages of our proposed method.

KW - Diagonal PCA

KW - Face recognition

KW - Non-greedy strategy

KW - Principal component analysis (PCA)

KW - ℓ-norm

UR - http://www.scopus.com/inward/record.url?scp=84944510769&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2015.06.011

DO - 10.1016/j.neucom.2015.06.011

M3 - 文章

AN - SCOPUS:84944510769

SN - 0925-2312

VL - 171

SP - 57

EP - 62

JO - Neurocomputing

JF - Neurocomputing

ER -

Diagonal principal component analysis with non-greedy ℓ₁-norm maximization for face recognition

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this