Verification Under Increasing Dimensionality

A.J. Hendrikse; Raymond N.J. Veldhuis; Lieuwe Jan Spreeuwers

doi:10.1109/ICPR.2010.149

Verification Under Increasing Dimensionality

A.J. Hendrikse, Raymond N.J. Veldhuis, Lieuwe Jan Spreeuwers

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

1 Citation (Scopus)

166 Downloads (Pure)

Abstract

Verification decisions are often based on second order statistics estimated from a set of samples. Ongoing growth of computational resources allows for considering more and more features, increasing the dimensionality of the samples. If the dimensionality is of the same order as the number of samples used in the estimation or even higher, then the accuracy of the estimate decreases significantly. In particular, the eigenvalues of the covariance matrix are estimated with a bias and the estimate of the eigenvectors differ considerably from the real eigenvectors. We show how a classical approach of verification in high dimensions is severely affected by these problems, and we show how bias correction methods can reduce these problems.

Original language	Undefined
Title of host publication	20th International Conference on Pattern Recognition (ICPR 2010)
Place of Publication	Los Alamitos, CA, USA
Publisher	IEEE
Pages	589-592
Number of pages	4
ISBN (Print)	978-0-7695-4109-9
DOIs	https://doi.org/10.1109/ICPR.2010.149
Publication status	Published - Aug 2010
Event	20th International Conference on Pattern Recognition 2010 - Istanbul Convention & Exhibition Centre, Istanbul, Turkey Duration: 23 Aug 2010 → 26 Aug 2010 Conference number: 20 https://www.ieee.org/conferences_events/conferences/conferencedetails/index.html?Conf_ID=16097

Publication series

Name
Publisher	IEEE Computer Society
ISSN (Print)	1051-4651

Conference

Conference	20th International Conference on Pattern Recognition 2010
Abbreviated title	ICPR 2010
Country/Territory	Turkey
City	Istanbul
Period	23/08/10 → 26/08/10
Internet address	https://www.ieee.org/conferences_events/conferences/conferencedetails/index.html?Conf_ID=16097

Keywords

METIS-271099
Bias correction
IR-74103
EWI-18676
General Statistical Analysis
SCS-Safety
High dimensional verification

Access to Document

10.1109/ICPR.2010.149

Cite this

@inproceedings{1cb5b0f041b04d92a2e3106ae0a3149c,

title = "Verification Under Increasing Dimensionality",

abstract = "Verification decisions are often based on second order statistics estimated from a set of samples. Ongoing growth of computational resources allows for considering more and more features, increasing the dimensionality of the samples. If the dimensionality is of the same order as the number of samples used in the estimation or even higher, then the accuracy of the estimate decreases significantly. In particular, the eigenvalues of the covariance matrix are estimated with a bias and the estimate of the eigenvectors differ considerably from the real eigenvectors. We show how a classical approach of verification in high dimensions is severely affected by these problems, and we show how bias correction methods can reduce these problems.",

keywords = "METIS-271099, Bias correction, IR-74103, EWI-18676, General Statistical Analysis, SCS-Safety, High dimensional verification",

author = "A.J. Hendrikse and Veldhuis, {Raymond N.J.} and Spreeuwers, {Lieuwe Jan}",

note = "10.1109/ICPR.2010.149 ; 20th International Conference on Pattern Recognition 2010 ; Conference date: 23-08-2010 Through 26-08-2010",

year = "2010",

month = aug,

doi = "10.1109/ICPR.2010.149",

language = "Undefined",

isbn = "978-0-7695-4109-9",

publisher = "IEEE",

pages = "589--592",

booktitle = "20th International Conference on Pattern Recognition (ICPR 2010)",

address = "United States",

url = "https://www.ieee.org/conferences_events/conferences/conferencedetails/index.html?Conf_ID=16097",

}

TY - GEN

T1 - Verification Under Increasing Dimensionality

AU - Hendrikse, A.J.

AU - Veldhuis, Raymond N.J.

AU - Spreeuwers, Lieuwe Jan

N1 - Conference code: 20

PY - 2010/8

Y1 - 2010/8

N2 - Verification decisions are often based on second order statistics estimated from a set of samples. Ongoing growth of computational resources allows for considering more and more features, increasing the dimensionality of the samples. If the dimensionality is of the same order as the number of samples used in the estimation or even higher, then the accuracy of the estimate decreases significantly. In particular, the eigenvalues of the covariance matrix are estimated with a bias and the estimate of the eigenvectors differ considerably from the real eigenvectors. We show how a classical approach of verification in high dimensions is severely affected by these problems, and we show how bias correction methods can reduce these problems.

AB - Verification decisions are often based on second order statistics estimated from a set of samples. Ongoing growth of computational resources allows for considering more and more features, increasing the dimensionality of the samples. If the dimensionality is of the same order as the number of samples used in the estimation or even higher, then the accuracy of the estimate decreases significantly. In particular, the eigenvalues of the covariance matrix are estimated with a bias and the estimate of the eigenvectors differ considerably from the real eigenvectors. We show how a classical approach of verification in high dimensions is severely affected by these problems, and we show how bias correction methods can reduce these problems.

KW - METIS-271099

KW - Bias correction

KW - IR-74103

KW - EWI-18676

KW - General Statistical Analysis

KW - SCS-Safety

KW - High dimensional verification

U2 - 10.1109/ICPR.2010.149

DO - 10.1109/ICPR.2010.149

M3 - Conference contribution

SN - 978-0-7695-4109-9

SP - 589

EP - 592

BT - 20th International Conference on Pattern Recognition (ICPR 2010)

PB - IEEE

CY - Los Alamitos, CA, USA

T2 - 20th International Conference on Pattern Recognition 2010

Y2 - 23 August 2010 through 26 August 2010

ER -