High-dimensional mean estimation via ℓ1 penalized normal likelihood

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

A new method is proposed for estimating the difference between the high-dimensional mean vectors of two multivariate normal populations with equal covariance matrix based on an ℓ1 penalized normal likelihood. It is well known that the normal likelihood involves the covariance matrix which is usually unknown. We substitute the adaptive thresholding estimator given by Cai and Liu (2011) of the covariance matrix, and then estimate the difference between the mean vectors by maximizing the ℓ1 penalized normal likelihood. Under the high-dimensional framework where both the sample size and the dimension tend to infinity, we show that the proposed estimator has sign recovery and also derive its mean squared error. We also compare the proposed estimator with the soft-thresholding and the adaptive soft-thresholding estimators which give simple thresholdings for the sample mean vector.

Original languageEnglish
Pages (from-to)90-106
Number of pages17
JournalJournal of Multivariate Analysis
Volume130
DOIs
Publication statusPublished - 2014 Jan 1
Externally publishedYes

Fingerprint

Covariance matrix
Likelihood
High-dimensional
Thresholding
Estimator
Adaptive Thresholding
Normal Population
Multivariate Normal
Sample mean
Substitute
Mean Squared Error
Recovery
Sample Size
Infinity
Tend
Unknown
Estimate

Keywords

  • Consistency
  • High-dimensional data
  • Mean squared error
  • Penalized normal likelihood
  • Sign recovery

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Numerical Analysis
  • Statistics and Probability

Cite this

High-dimensional mean estimation via ℓ1 penalized normal likelihood. / Katayama, Shota.

In: Journal of Multivariate Analysis, Vol. 130, 01.01.2014, p. 90-106.

Research output: Contribution to journalArticle

@article{b3f8eb11d16446748dc09e4ae7442b10,
title = "High-dimensional mean estimation via ℓ1 penalized normal likelihood",
abstract = "A new method is proposed for estimating the difference between the high-dimensional mean vectors of two multivariate normal populations with equal covariance matrix based on an ℓ1 penalized normal likelihood. It is well known that the normal likelihood involves the covariance matrix which is usually unknown. We substitute the adaptive thresholding estimator given by Cai and Liu (2011) of the covariance matrix, and then estimate the difference between the mean vectors by maximizing the ℓ1 penalized normal likelihood. Under the high-dimensional framework where both the sample size and the dimension tend to infinity, we show that the proposed estimator has sign recovery and also derive its mean squared error. We also compare the proposed estimator with the soft-thresholding and the adaptive soft-thresholding estimators which give simple thresholdings for the sample mean vector.",
keywords = "Consistency, High-dimensional data, Mean squared error, Penalized normal likelihood, Sign recovery",
author = "Shota Katayama",
year = "2014",
month = "1",
day = "1",
doi = "10.1016/j.jmva.2014.05.005",
language = "English",
volume = "130",
pages = "90--106",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - High-dimensional mean estimation via ℓ1 penalized normal likelihood

AU - Katayama, Shota

PY - 2014/1/1

Y1 - 2014/1/1

N2 - A new method is proposed for estimating the difference between the high-dimensional mean vectors of two multivariate normal populations with equal covariance matrix based on an ℓ1 penalized normal likelihood. It is well known that the normal likelihood involves the covariance matrix which is usually unknown. We substitute the adaptive thresholding estimator given by Cai and Liu (2011) of the covariance matrix, and then estimate the difference between the mean vectors by maximizing the ℓ1 penalized normal likelihood. Under the high-dimensional framework where both the sample size and the dimension tend to infinity, we show that the proposed estimator has sign recovery and also derive its mean squared error. We also compare the proposed estimator with the soft-thresholding and the adaptive soft-thresholding estimators which give simple thresholdings for the sample mean vector.

AB - A new method is proposed for estimating the difference between the high-dimensional mean vectors of two multivariate normal populations with equal covariance matrix based on an ℓ1 penalized normal likelihood. It is well known that the normal likelihood involves the covariance matrix which is usually unknown. We substitute the adaptive thresholding estimator given by Cai and Liu (2011) of the covariance matrix, and then estimate the difference between the mean vectors by maximizing the ℓ1 penalized normal likelihood. Under the high-dimensional framework where both the sample size and the dimension tend to infinity, we show that the proposed estimator has sign recovery and also derive its mean squared error. We also compare the proposed estimator with the soft-thresholding and the adaptive soft-thresholding estimators which give simple thresholdings for the sample mean vector.

KW - Consistency

KW - High-dimensional data

KW - Mean squared error

KW - Penalized normal likelihood

KW - Sign recovery

UR - http://www.scopus.com/inward/record.url?scp=84901612796&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901612796&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2014.05.005

DO - 10.1016/j.jmva.2014.05.005

M3 - Article

VL - 130

SP - 90

EP - 106

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

ER -