HDLSS discrimination with adaptive data piling

Myung Hee Lee, Jeongyoun Ahn, Yongho Jeon

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

We propose new discrimination methods for classification of high dimension, low sample size (HDLSS) data that regularize the degree of data piling. The within-class scatter of the HDLSS data, when projected onto a low-dimensional discriminant subspace, can be selected to be arbitrarily small. Using this fact, we develop two different ways of tuning the amount of within-class scatter, or equivalently, the degree of data piling. In the first approach,we consider a linear path connecting the maximal data piling and the least data piling directions. We also formulate a problem of finding the optimal classifier under a constraint on data piling. The data piling regularization methods are extended to multicategory problems. Simulated and real data examples show competitive performances of the proposed classification methods. Supplementary materials for this article are available online on the journal web site.

Original languageEnglish
Pages (from-to)433-451
Number of pages19
JournalJournal of Computational and Graphical Statistics
Volume22
Issue number2
DOIs
Publication statusPublished - 2013 Dec 17

Fingerprint

Higher Dimensions
Discrimination
Sample Size
Scatter
Sample size
Regularization Method
Discriminant
Tuning
Classifier
Subspace
Path
Regularization
Web sites

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Discrete Mathematics and Combinatorics
  • Statistics, Probability and Uncertainty

Cite this

Lee, Myung Hee ; Ahn, Jeongyoun ; Jeon, Yongho. / HDLSS discrimination with adaptive data piling. In: Journal of Computational and Graphical Statistics. 2013 ; Vol. 22, No. 2. pp. 433-451.
@article{2bd28aeb9f2e4c83960a82cf4d0fbbf7,
title = "HDLSS discrimination with adaptive data piling",
abstract = "We propose new discrimination methods for classification of high dimension, low sample size (HDLSS) data that regularize the degree of data piling. The within-class scatter of the HDLSS data, when projected onto a low-dimensional discriminant subspace, can be selected to be arbitrarily small. Using this fact, we develop two different ways of tuning the amount of within-class scatter, or equivalently, the degree of data piling. In the first approach,we consider a linear path connecting the maximal data piling and the least data piling directions. We also formulate a problem of finding the optimal classifier under a constraint on data piling. The data piling regularization methods are extended to multicategory problems. Simulated and real data examples show competitive performances of the proposed classification methods. Supplementary materials for this article are available online on the journal web site.",
author = "Lee, {Myung Hee} and Jeongyoun Ahn and Yongho Jeon",
year = "2013",
month = "12",
day = "17",
doi = "10.1080/10618600.2012.681235",
language = "English",
volume = "22",
pages = "433--451",
journal = "Journal of Computational and Graphical Statistics",
issn = "1061-8600",
publisher = "American Statistical Association",
number = "2",

}

HDLSS discrimination with adaptive data piling. / Lee, Myung Hee; Ahn, Jeongyoun; Jeon, Yongho.

In: Journal of Computational and Graphical Statistics, Vol. 22, No. 2, 17.12.2013, p. 433-451.

Research output: Contribution to journalArticle

TY - JOUR

T1 - HDLSS discrimination with adaptive data piling

AU - Lee, Myung Hee

AU - Ahn, Jeongyoun

AU - Jeon, Yongho

PY - 2013/12/17

Y1 - 2013/12/17

N2 - We propose new discrimination methods for classification of high dimension, low sample size (HDLSS) data that regularize the degree of data piling. The within-class scatter of the HDLSS data, when projected onto a low-dimensional discriminant subspace, can be selected to be arbitrarily small. Using this fact, we develop two different ways of tuning the amount of within-class scatter, or equivalently, the degree of data piling. In the first approach,we consider a linear path connecting the maximal data piling and the least data piling directions. We also formulate a problem of finding the optimal classifier under a constraint on data piling. The data piling regularization methods are extended to multicategory problems. Simulated and real data examples show competitive performances of the proposed classification methods. Supplementary materials for this article are available online on the journal web site.

AB - We propose new discrimination methods for classification of high dimension, low sample size (HDLSS) data that regularize the degree of data piling. The within-class scatter of the HDLSS data, when projected onto a low-dimensional discriminant subspace, can be selected to be arbitrarily small. Using this fact, we develop two different ways of tuning the amount of within-class scatter, or equivalently, the degree of data piling. In the first approach,we consider a linear path connecting the maximal data piling and the least data piling directions. We also formulate a problem of finding the optimal classifier under a constraint on data piling. The data piling regularization methods are extended to multicategory problems. Simulated and real data examples show competitive performances of the proposed classification methods. Supplementary materials for this article are available online on the journal web site.

UR - http://www.scopus.com/inward/record.url?scp=84884181705&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84884181705&partnerID=8YFLogxK

U2 - 10.1080/10618600.2012.681235

DO - 10.1080/10618600.2012.681235

M3 - Article

VL - 22

SP - 433

EP - 451

JO - Journal of Computational and Graphical Statistics

JF - Journal of Computational and Graphical Statistics

SN - 1061-8600

IS - 2

ER -