Sparse HDLSS discrimination with constrained data piling

Jeongyoun Ahn; Yongho Jeon

doi:10.1016/j.csda.2015.04.006

Sparse HDLSS discrimination with constrained data piling

Jeongyoun Ahn, Yongho Jeon

Department of Applied Statistics

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

3 Citations (Scopus)

Abstract

Abstract Regularization is a key component in high dimensional data analyses. In high dimensional discrimination with binary classes, the phenomenon of data piling occurs when the projection of data onto a discriminant vector is dichotomous, one for each class. Regularizing the degree of data piling yields a new class of discrimination rules for high dimension-low sample size data. A discrimination method that regularizes the degree of data piling while achieving sparsity is proposed and solved via a linear programming. Computational efficiency is further improved by a sign-preserving regularization that forces the signs of the estimator to be the same as the mean difference. The proposed classifier shows competitive performances for simulated and real data examples including speech recognition and gene expressions.

Original language	English
Article number	6079
Pages (from-to)	74-83
Number of pages	10
Journal	Computational Statistics and Data Analysis
Volume	90
DOIs	https://doi.org/10.1016/j.csda.2015.04.006
Publication status	Published - 2015 Apr 28

Bibliographical note

Funding Information:
The authors are grateful to the reviewers and the associate editor for many useful comments that improved the paper. Jeon’s research was supported by Basic Science Research Program of the National Research Foundation of Korea ( NRF-2012R1A1A1012043 ) funded by the Korean government.

Publisher Copyright:
© 2015 Elsevier B.V.

All Science Journal Classification (ASJC) codes

Statistics and Probability
Computational Mathematics
Computational Theory and Mathematics
Applied Mathematics

Access to Document

10.1016/j.csda.2015.04.006

Cite this

@article{51471b97b07f45fda4152e2928e6dcb7,

title = "Sparse HDLSS discrimination with constrained data piling",

abstract = "Abstract Regularization is a key component in high dimensional data analyses. In high dimensional discrimination with binary classes, the phenomenon of data piling occurs when the projection of data onto a discriminant vector is dichotomous, one for each class. Regularizing the degree of data piling yields a new class of discrimination rules for high dimension-low sample size data. A discrimination method that regularizes the degree of data piling while achieving sparsity is proposed and solved via a linear programming. Computational efficiency is further improved by a sign-preserving regularization that forces the signs of the estimator to be the same as the mean difference. The proposed classifier shows competitive performances for simulated and real data examples including speech recognition and gene expressions.",

author = "Jeongyoun Ahn and Yongho Jeon",

note = "Funding Information: The authors are grateful to the reviewers and the associate editor for many useful comments that improved the paper. Jeon{\textquoteright}s research was supported by Basic Science Research Program of the National Research Foundation of Korea ( NRF-2012R1A1A1012043 ) funded by the Korean government. Publisher Copyright: {\textcopyright} 2015 Elsevier B.V.",

year = "2015",

month = apr,

day = "28",

doi = "10.1016/j.csda.2015.04.006",

language = "English",

volume = "90",

pages = "74--83",

journal = "Computational Statistics and Data Analysis",

issn = "0167-9473",

publisher = "Elsevier",

}

TY - JOUR

T1 - Sparse HDLSS discrimination with constrained data piling

AU - Ahn, Jeongyoun

AU - Jeon, Yongho

N1 - Funding Information: The authors are grateful to the reviewers and the associate editor for many useful comments that improved the paper. Jeon’s research was supported by Basic Science Research Program of the National Research Foundation of Korea ( NRF-2012R1A1A1012043 ) funded by the Korean government. Publisher Copyright: © 2015 Elsevier B.V.

PY - 2015/4/28

Y1 - 2015/4/28

N2 - Abstract Regularization is a key component in high dimensional data analyses. In high dimensional discrimination with binary classes, the phenomenon of data piling occurs when the projection of data onto a discriminant vector is dichotomous, one for each class. Regularizing the degree of data piling yields a new class of discrimination rules for high dimension-low sample size data. A discrimination method that regularizes the degree of data piling while achieving sparsity is proposed and solved via a linear programming. Computational efficiency is further improved by a sign-preserving regularization that forces the signs of the estimator to be the same as the mean difference. The proposed classifier shows competitive performances for simulated and real data examples including speech recognition and gene expressions.

AB - Abstract Regularization is a key component in high dimensional data analyses. In high dimensional discrimination with binary classes, the phenomenon of data piling occurs when the projection of data onto a discriminant vector is dichotomous, one for each class. Regularizing the degree of data piling yields a new class of discrimination rules for high dimension-low sample size data. A discrimination method that regularizes the degree of data piling while achieving sparsity is proposed and solved via a linear programming. Computational efficiency is further improved by a sign-preserving regularization that forces the signs of the estimator to be the same as the mean difference. The proposed classifier shows competitive performances for simulated and real data examples including speech recognition and gene expressions.

UR - http://www.scopus.com/inward/record.url?scp=84929193188&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84929193188&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2015.04.006

DO - 10.1016/j.csda.2015.04.006

M3 - Article

AN - SCOPUS:84929193188

SN - 0167-9473

VL - 90

SP - 74

EP - 83

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

M1 - 6079

ER -

Sparse HDLSS discrimination with constrained data piling

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this