A method to estimate the true mahalanobis distance from eigenvectors of sample covariance matrix

Masakazu Iwamura, Shinichiro Omachi, Hirotomo Aso

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Citations (Scopus)

Abstract

In statistical pattern recognition, the parameters of distributions are usually estimated from training sample vectors. However, estimated parameters contain estimation errors, and the errors cause bad influence on recognition performance when the sample size is not sufficient. Some methods can obtain better estimates of the eigenvalues of the true covariance matrix and can avoid bad influences caused by estimation errors of eigenvalues. However, estimation errors of eigenvectors of covariance matrix have not been considered enough. In this paper, we consider estimation errors of eigenvectors and show the errors can be regarded as estimation errors of eigenvalues. Then, we present a method to estimate the true Mahalanobis distance from eigenvectors of the sample covariance matrix. Recognition experiments show that by applying the proposed method, the true Mahalanobis distance can be estimated even if the sample size is small, and better recognition accuracy is achieved. The proposed method is useful for the practical applications of pattern recognition since the proposed method is effective without any hyper-parameters.

Original languageEnglish
Title of host publicationStructural, Syntactic, and Statistical Pattern Recognition - Joint IAPR International Workshops SSPR 2002 and SPR 2002, Proceedings
EditorsTerry Caelli, Adnan Amin, Robert P.W. Duin, Dick de Ridder, Mohamed Kamel
PublisherSpringer Verlag
Pages498-507
Number of pages10
ISBN (Print)3540440119, 9783540440116
DOIs
Publication statusPublished - 2002
EventJoint IAPR 9th International Workshop on Structural and Syntactic Pattern Recognition, SSPR 2002 and 4th International Workshop on Statistical Techniques in Pattern Recognition, SPR 2002 - Windsor, Canada
Duration: 2002 Aug 62002 Aug 9

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2396
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceJoint IAPR 9th International Workshop on Structural and Syntactic Pattern Recognition, SSPR 2002 and 4th International Workshop on Statistical Techniques in Pattern Recognition, SPR 2002
Country/TerritoryCanada
CityWindsor
Period02/8/602/8/9

Fingerprint

Dive into the research topics of 'A method to estimate the true mahalanobis distance from eigenvectors of sample covariance matrix'. Together they form a unique fingerprint.

Cite this