skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Quantifying Multivariate Classification Performance - the Problem of Overfitting

Abstract

We have been studying the use of spectral imagery to locate targets in spectrally interfering backgrounds. In making performance estimates for various sensors it has become evident that some calculations are unreliable because of overflying. Hence, we began a thorough study of the problem of overfitting in multivariate classification. In this paper we present some model based results describing the problem. From the model we know the ideal covariance matrix, the ideal discriminant vector, and the ideal classification performance. We then investigate how experimental conditions such as noise, number of bands, and number of samples cause discrepancies from the ideal results. We also suggest ways to discover and alleviate overfitting.

Authors:
;
Publication Date:
Research Org.:
Sandia National Labs., Albuquerque, NM (US); Sandia National Labs., Livermore, CA (US)
Sponsoring Org.:
US Department of Energy (US)
OSTI Identifier:
9702
Report Number(s):
SAND99-2070C
TRN: AH200125%%16
DOE Contract Number:  
AC04-94AL85000
Resource Type:
Conference
Resource Relation:
Conference: SPIE Annual Conference, Denver, CO (US), 07/21/1999; Other Information: PBD: 9 Aug 1999
Country of Publication:
United States
Language:
English
Subject:
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; CLASSIFICATION; PERFORMANCE; MULTIVARIATE ANALYSIS; SPECTRA; SIGNAL-TO-NOISE RATIO; IMAGING SPECTROMETRY; MULTIVARIATE CLASSIFICATION; OVERFITTING

Citation Formats

Stallard, Brian R., and Taylor, John G. Quantifying Multivariate Classification Performance - the Problem of Overfitting. United States: N. p., 1999. Web.
Stallard, Brian R., & Taylor, John G. Quantifying Multivariate Classification Performance - the Problem of Overfitting. United States.
Stallard, Brian R., and Taylor, John G. Mon . "Quantifying Multivariate Classification Performance - the Problem of Overfitting". United States. https://www.osti.gov/servlets/purl/9702.
@article{osti_9702,
title = {Quantifying Multivariate Classification Performance - the Problem of Overfitting},
author = {Stallard, Brian R. and Taylor, John G.},
abstractNote = {We have been studying the use of spectral imagery to locate targets in spectrally interfering backgrounds. In making performance estimates for various sensors it has become evident that some calculations are unreliable because of overflying. Hence, we began a thorough study of the problem of overfitting in multivariate classification. In this paper we present some model based results describing the problem. From the model we know the ideal covariance matrix, the ideal discriminant vector, and the ideal classification performance. We then investigate how experimental conditions such as noise, number of bands, and number of samples cause discrepancies from the ideal results. We also suggest ways to discover and alleviate overfitting.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {1999},
month = {8}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share: