Ultrahigh-Dimensional Multiclass Linear Discriminant Analysis by Pairwise Sure Independence Screening

Rui Pan, Hansheng Wang, Runze Li

Research output: Contribution to journalArticle

19 Citations (Scopus)

Abstract

This article is concerned with the problem of feature screening for multiclass linear discriminant analysis under ultrahigh-dimensional setting. We allow the number of classes to be relatively large. As a result, the total number of relevant features is larger than usual. This makes the related classification problem much more challenging than the conventional one, where the number of classes is small (very often two). To solve the problem, we propose a novel pairwise sure independence screening method for linear discriminant analysis with an ultrahigh-dimensional predictor. The proposed procedure is directly applicable to the situation with many classes. We further prove that the proposed method is screening consistent. Simulation studies are conducted to assess the finite sample performance of the new procedure. We also demonstrate the proposed methodology via an empirical analysis of a real life example on handwritten Chinese character recognition.

Original languageEnglish (US)
Pages (from-to)169-179
Number of pages11
JournalJournal of the American Statistical Association
Volume111
Issue number513
DOIs
StatePublished - Jan 2 2016

Fingerprint

Multi-class
Discriminant Analysis
Screening
Pairwise
Character Recognition
Empirical Analysis
Classification Problems
Predictors
Simulation Study
Methodology
Demonstrate
Independence
Class
Discriminant analysis

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

@article{2cd3e186cd374a759ca5ef9fcf44e853,
title = "Ultrahigh-Dimensional Multiclass Linear Discriminant Analysis by Pairwise Sure Independence Screening",
abstract = "This article is concerned with the problem of feature screening for multiclass linear discriminant analysis under ultrahigh-dimensional setting. We allow the number of classes to be relatively large. As a result, the total number of relevant features is larger than usual. This makes the related classification problem much more challenging than the conventional one, where the number of classes is small (very often two). To solve the problem, we propose a novel pairwise sure independence screening method for linear discriminant analysis with an ultrahigh-dimensional predictor. The proposed procedure is directly applicable to the situation with many classes. We further prove that the proposed method is screening consistent. Simulation studies are conducted to assess the finite sample performance of the new procedure. We also demonstrate the proposed methodology via an empirical analysis of a real life example on handwritten Chinese character recognition.",
author = "Rui Pan and Hansheng Wang and Runze Li",
year = "2016",
month = "1",
day = "2",
doi = "10.1080/01621459.2014.998760",
language = "English (US)",
volume = "111",
pages = "169--179",
journal = "Journal of the American Statistical Association",
issn = "0162-1459",
publisher = "Taylor and Francis Ltd.",
number = "513",

}

Ultrahigh-Dimensional Multiclass Linear Discriminant Analysis by Pairwise Sure Independence Screening. / Pan, Rui; Wang, Hansheng; Li, Runze.

In: Journal of the American Statistical Association, Vol. 111, No. 513, 02.01.2016, p. 169-179.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Ultrahigh-Dimensional Multiclass Linear Discriminant Analysis by Pairwise Sure Independence Screening

AU - Pan, Rui

AU - Wang, Hansheng

AU - Li, Runze

PY - 2016/1/2

Y1 - 2016/1/2

N2 - This article is concerned with the problem of feature screening for multiclass linear discriminant analysis under ultrahigh-dimensional setting. We allow the number of classes to be relatively large. As a result, the total number of relevant features is larger than usual. This makes the related classification problem much more challenging than the conventional one, where the number of classes is small (very often two). To solve the problem, we propose a novel pairwise sure independence screening method for linear discriminant analysis with an ultrahigh-dimensional predictor. The proposed procedure is directly applicable to the situation with many classes. We further prove that the proposed method is screening consistent. Simulation studies are conducted to assess the finite sample performance of the new procedure. We also demonstrate the proposed methodology via an empirical analysis of a real life example on handwritten Chinese character recognition.

AB - This article is concerned with the problem of feature screening for multiclass linear discriminant analysis under ultrahigh-dimensional setting. We allow the number of classes to be relatively large. As a result, the total number of relevant features is larger than usual. This makes the related classification problem much more challenging than the conventional one, where the number of classes is small (very often two). To solve the problem, we propose a novel pairwise sure independence screening method for linear discriminant analysis with an ultrahigh-dimensional predictor. The proposed procedure is directly applicable to the situation with many classes. We further prove that the proposed method is screening consistent. Simulation studies are conducted to assess the finite sample performance of the new procedure. We also demonstrate the proposed methodology via an empirical analysis of a real life example on handwritten Chinese character recognition.

UR - http://www.scopus.com/inward/record.url?scp=84969872782&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84969872782&partnerID=8YFLogxK

U2 - 10.1080/01621459.2014.998760

DO - 10.1080/01621459.2014.998760

M3 - Article

AN - SCOPUS:84969872782

VL - 111

SP - 169

EP - 179

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

SN - 0162-1459

IS - 513

ER -