Covariance estimation with limited training samples

被引：126

作者：

Tadjudin, S ^{[1
]}

Landgrebe, DA ^{[1
]}

机构：

[1] Purdue Univ, Sch Elect & Comp Engn, W Lafayette, IN 47907 USA

来源：

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING | 1999年 / 37卷 / 04期

关键词：

covariance estimation; Gaussian maximum likelihood; leave-one-out log likelihood; regularization;

D O I：

10.1109/36.774728

中图分类号：

P3 [地球物理学]; P59 [地球化学];

学科分类号：

0708 ; 070902 ;

摘要：

,This paper describes a covariance estimator formulated under an empirical Bayesian setting to mitigate the problem of limited training samples in the Gaussian maximum likelihood (ML) classification for remote sensing. The most suitable covariance mixture is selected by maximizing the average leave-one-out log likelihood. Experimental results using AVIRIS data are presented.

引用

页码：2113 / 2118

页数：6

共 8 条

[1]

FRIEDMAN JF, 1989, J R STAT SOC, V84, P17

[2]

Fukunaga F, 1990, INTRO STAT PATTERN R

[3] Covariance matrix estimation and classification with limited training data [J].

Hoffbeck, JP ;

Landgrebe, DA .

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1996, 18 (07) :763-767

[4] CLASSIFICATION OF MULTISPECTRAL IMAGE DATA BY EXTRACTION AND CLASSIFICATION OF HOMOGENEOUS OBJECTS [J].

KETTIG, RL ;

LANDGREBE, DA .

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 1976, 14 (01) :19-26

[5] FEATURE-EXTRACTION BASED ON DECISION BOUNDARIES [J].

LEE, CH ;

LANDGREBE, DA .

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1993, 15 (04) :388-400

[6]

LIN S.P., 1985, MULTIVARIATE ANAL, V6, P411

[7] COVARIANCE POOLING AND STABILIZATION FOR CLASSIFICATION [J].

RAYENS, W ;

GREENE, T .

COMPUTATIONAL STATISTICS & DATA ANALYSIS, 1991, 11 (01) :17-42

[8] DISCRIMINANT FUNCTIONS WHEN COVARIANCES ARE UNEQUAL AND SAMPLE SIZES ARE MODERATE [J].

WAHL, PW ;

KRONMAL, RA .

BIOMETRICS, 1977, 33 (03) :479-484

← 1 →