Error estimation of the Gaussian ML classifier

Research output: Contribution to conferencePaper

1 Citation (Scopus)

Abstract

The relationship between the classification error and the Bhattacharyya distance of two normally distributed classes is investigated and a new formula which provides an error estimation of the Gaussian maximum likelihood (ML) classifier is proposed. A significant improvement over the previous theoretical bounds is the possibility to predict the classification error within 1-2% margin from the Bhattacharyya distance. The new formula can be successfully used for feature selection and feature evaluation.

Original languageEnglish
Number of pages1
Publication statusPublished - 1997 Jan 1
EventProceedings of the 1997 IEEE International Symposium on Information Theory - Ulm, Ger
Duration: 1997 Jun 291997 Jul 4

Other

OtherProceedings of the 1997 IEEE International Symposium on Information Theory
CityUlm, Ger
Period97/6/2997/7/4

Fingerprint

Error Estimation
Error analysis
Maximum likelihood
Maximum Likelihood
Classifiers
Classifier
Margin
Feature Selection
Feature extraction
Predict
Evaluation
Relationships
Class

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering
  • Applied Mathematics
  • Modelling and Simulation
  • Theoretical Computer Science
  • Information Systems

Cite this

Lee, C. (1997). Error estimation of the Gaussian ML classifier. Paper presented at Proceedings of the 1997 IEEE International Symposium on Information Theory, Ulm, Ger, .
Lee, Chulhee. / Error estimation of the Gaussian ML classifier. Paper presented at Proceedings of the 1997 IEEE International Symposium on Information Theory, Ulm, Ger, .1 p.
@conference{8466bcfe5dd24ba68bd6fdfe733800db,
title = "Error estimation of the Gaussian ML classifier",
abstract = "The relationship between the classification error and the Bhattacharyya distance of two normally distributed classes is investigated and a new formula which provides an error estimation of the Gaussian maximum likelihood (ML) classifier is proposed. A significant improvement over the previous theoretical bounds is the possibility to predict the classification error within 1-2{\%} margin from the Bhattacharyya distance. The new formula can be successfully used for feature selection and feature evaluation.",
author = "Chulhee Lee",
year = "1997",
month = "1",
day = "1",
language = "English",
note = "Proceedings of the 1997 IEEE International Symposium on Information Theory ; Conference date: 29-06-1997 Through 04-07-1997",

}

Lee, C 1997, 'Error estimation of the Gaussian ML classifier' Paper presented at Proceedings of the 1997 IEEE International Symposium on Information Theory, Ulm, Ger, 97/6/29 - 97/7/4, .

Error estimation of the Gaussian ML classifier. / Lee, Chulhee.

1997. Paper presented at Proceedings of the 1997 IEEE International Symposium on Information Theory, Ulm, Ger, .

Research output: Contribution to conferencePaper

TY - CONF

T1 - Error estimation of the Gaussian ML classifier

AU - Lee, Chulhee

PY - 1997/1/1

Y1 - 1997/1/1

N2 - The relationship between the classification error and the Bhattacharyya distance of two normally distributed classes is investigated and a new formula which provides an error estimation of the Gaussian maximum likelihood (ML) classifier is proposed. A significant improvement over the previous theoretical bounds is the possibility to predict the classification error within 1-2% margin from the Bhattacharyya distance. The new formula can be successfully used for feature selection and feature evaluation.

AB - The relationship between the classification error and the Bhattacharyya distance of two normally distributed classes is investigated and a new formula which provides an error estimation of the Gaussian maximum likelihood (ML) classifier is proposed. A significant improvement over the previous theoretical bounds is the possibility to predict the classification error within 1-2% margin from the Bhattacharyya distance. The new formula can be successfully used for feature selection and feature evaluation.

UR - http://www.scopus.com/inward/record.url?scp=0030677390&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030677390&partnerID=8YFLogxK

M3 - Paper

ER -

Lee C. Error estimation of the Gaussian ML classifier. 1997. Paper presented at Proceedings of the 1997 IEEE International Symposium on Information Theory, Ulm, Ger, .