Analyzing the Gaussian ML classifier for limited training samples

Chulhee Lee, Euisun Choi, Byungjoon Baek, Changrak Yoon

Research output: Contribution to conferencePaper

Abstract

The Gaussian ML classifier is one of the most widely used classifiers for remotely sensed data since it is easy to implement and relatively fast. However, as the dimension of hyperspectral images significantly increases, the performance of the Gaussian ML classifier suffers when training samples are not enough, mainly due to inaccurate estimation of covariance matrices. In this paper, we provide thorough performance analyses of the Gaussian ML classifier in terms of the number of training samples. In particular, we analyze how decision boundaries which the Gaussian ML classifier defines vary when limited training samples are available. In order to quantify variations of decision boundaries, we introduce two distance measures. Experimental results show that there is a significant variation in covariance and mean estimation, which subsequently produces noticeably different decision boundaries.

Original languageEnglish
Pages3229-3232
Number of pages4
Publication statusPublished - 2004 Dec 1
Event2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004 - Anchorage, AK, United States
Duration: 2004 Sep 202004 Sep 24

Other

Other2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004
CountryUnited States
CityAnchorage, AK
Period04/9/2004/9/24

Fingerprint

Classifiers
Covariance matrix
matrix
decision

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Earth and Planetary Sciences(all)

Cite this

Lee, C., Choi, E., Baek, B., & Yoon, C. (2004). Analyzing the Gaussian ML classifier for limited training samples. 3229-3232. Paper presented at 2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004, Anchorage, AK, United States.
Lee, Chulhee ; Choi, Euisun ; Baek, Byungjoon ; Yoon, Changrak. / Analyzing the Gaussian ML classifier for limited training samples. Paper presented at 2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004, Anchorage, AK, United States.4 p.
@conference{484416e4def440bdaa74cbe75395c46a,
title = "Analyzing the Gaussian ML classifier for limited training samples",
abstract = "The Gaussian ML classifier is one of the most widely used classifiers for remotely sensed data since it is easy to implement and relatively fast. However, as the dimension of hyperspectral images significantly increases, the performance of the Gaussian ML classifier suffers when training samples are not enough, mainly due to inaccurate estimation of covariance matrices. In this paper, we provide thorough performance analyses of the Gaussian ML classifier in terms of the number of training samples. In particular, we analyze how decision boundaries which the Gaussian ML classifier defines vary when limited training samples are available. In order to quantify variations of decision boundaries, we introduce two distance measures. Experimental results show that there is a significant variation in covariance and mean estimation, which subsequently produces noticeably different decision boundaries.",
author = "Chulhee Lee and Euisun Choi and Byungjoon Baek and Changrak Yoon",
year = "2004",
month = "12",
day = "1",
language = "English",
pages = "3229--3232",
note = "2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004 ; Conference date: 20-09-2004 Through 24-09-2004",

}

Lee, C, Choi, E, Baek, B & Yoon, C 2004, 'Analyzing the Gaussian ML classifier for limited training samples' Paper presented at 2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004, Anchorage, AK, United States, 04/9/20 - 04/9/24, pp. 3229-3232.

Analyzing the Gaussian ML classifier for limited training samples. / Lee, Chulhee; Choi, Euisun; Baek, Byungjoon; Yoon, Changrak.

2004. 3229-3232 Paper presented at 2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004, Anchorage, AK, United States.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Analyzing the Gaussian ML classifier for limited training samples

AU - Lee, Chulhee

AU - Choi, Euisun

AU - Baek, Byungjoon

AU - Yoon, Changrak

PY - 2004/12/1

Y1 - 2004/12/1

N2 - The Gaussian ML classifier is one of the most widely used classifiers for remotely sensed data since it is easy to implement and relatively fast. However, as the dimension of hyperspectral images significantly increases, the performance of the Gaussian ML classifier suffers when training samples are not enough, mainly due to inaccurate estimation of covariance matrices. In this paper, we provide thorough performance analyses of the Gaussian ML classifier in terms of the number of training samples. In particular, we analyze how decision boundaries which the Gaussian ML classifier defines vary when limited training samples are available. In order to quantify variations of decision boundaries, we introduce two distance measures. Experimental results show that there is a significant variation in covariance and mean estimation, which subsequently produces noticeably different decision boundaries.

AB - The Gaussian ML classifier is one of the most widely used classifiers for remotely sensed data since it is easy to implement and relatively fast. However, as the dimension of hyperspectral images significantly increases, the performance of the Gaussian ML classifier suffers when training samples are not enough, mainly due to inaccurate estimation of covariance matrices. In this paper, we provide thorough performance analyses of the Gaussian ML classifier in terms of the number of training samples. In particular, we analyze how decision boundaries which the Gaussian ML classifier defines vary when limited training samples are available. In order to quantify variations of decision boundaries, we introduce two distance measures. Experimental results show that there is a significant variation in covariance and mean estimation, which subsequently produces noticeably different decision boundaries.

UR - http://www.scopus.com/inward/record.url?scp=15944422848&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=15944422848&partnerID=8YFLogxK

M3 - Paper

SP - 3229

EP - 3232

ER -

Lee C, Choi E, Baek B, Yoon C. Analyzing the Gaussian ML classifier for limited training samples. 2004. Paper presented at 2004 IEEE International Geoscience and Remote Sensing Symposium Proceedings: Science for Society: Exploring and Managing a Changing Planet. IGARSS 2004, Anchorage, AK, United States.