Treffer: Variability and bias in experimentally measured classifer error rates
Title:
Variability and bias in experimentally measured classifer error rates
Authors:
Source:
Pattern recognition letters. 13(10):685-692
Publisher Information:
Amsterdam: Elsevier, 1992.
Publication Year:
1992
Physical Description:
print, 13 ref
Original Material:
INIST-CNRS
Subject Terms:
Computer science, Informatique, Telecommunications, Télécommunications, Sciences exactes et technologie, Exact sciences and technology, Sciences appliquees, Applied sciences, Informatique; automatique theorique; systemes, Computer science; control theory; systems, Intelligence artificielle, Artificial intelligence, Reconnaissance des formes. Traitement numérique des images. Géométrie algorithmique, Pattern recognition. Digital image processing. Computational geometry, Chromosome, Cromosoma, Classification automatique, Automatic classification, Clasificación automática, Complexité calcul, Computing complexity, Complejidad cálculo, Loi normale, Gaussian distribution, Curva Gauss, Reconnaissance forme, Pattern recognition, Reconocimiento patrón, Simulation, Simulación, Système informatique, Computer system, Sistema informático, Taux erreur, Error rate, Indice error, Ensemble apprentissage
Document Type:
Fachzeitschrift
Article
File Description:
text
Language:
English
Author Affiliations:
MRC, Western gen. hosp., human genetic unit, Edinburgh EH4 2XU, United Kingdom
ISSN:
0167-8655
Rights:
Copyright 1993 INIST-CNRS
CC BY 4.0
Sauf mention contraire ci-dessus, le contenu de cette notice bibliographique peut être utilisé dans le cadre d’une licence CC BY 4.0 Inist-CNRS / Unless otherwise stated above, the content of this bibliographic record may be used under a CC BY 4.0 licence by Inist-CNRS / A menos que se haya señalado antes, el contenido de este registro bibliográfico puede ser utilizado al amparo de una licencia CC BY 4.0 Inist-CNRS
CC BY 4.0
Sauf mention contraire ci-dessus, le contenu de cette notice bibliographique peut être utilisé dans le cadre d’une licence CC BY 4.0 Inist-CNRS / Unless otherwise stated above, the content of this bibliographic record may be used under a CC BY 4.0 licence by Inist-CNRS / A menos que se haya señalado antes, el contenido de este registro bibliográfico puede ser utilizado al amparo de una licencia CC BY 4.0 Inist-CNRS
Notes:
Computer science; theoretical automation; systems
Accession Number:
edscal.4373870
Database:
PASCAL Archive
Weitere Informationen
The effect of training and test set sizes on variability and bias in classifier error rates was investigated empirically by subdivision of a data set of 122,735 human chromosomes. Both «independent sample» and «apparent» error estimates from small data sets were found to be approximately equally biased with respect to the underlying error rate that would be obtained from infinite samples, and the variability of the error estimates was higher than would be predicted by a simple model. The results suggest that the minimum adequate number of training samples is at least ten times the number of classifier parameters.