Classification with reject option

The authors study binary classification that allows for a reject option in which case no decision is made. This reject option is to be used for those observations for which the conditional class probabilities are close and as such are hard to classify. The authors generalize existing theory for both...

Full description

Saved in:
Bibliographic Details
Published in:Canadian journal of statistics Vol. 34; no. 4; pp. 709 - 721
Main Authors: Herbei, Radu, Wegkamp, Marten H.
Format: Journal Article
Language:English
Published: Hoboken Wiley-Blackwell 01.12.2006
Statistical Society of Canada
Wiley‐Blackwell
Wiley Subscription Services, Inc
Subjects:
Bayes classifiers
Bayes rule
Bayesian analysis
Classification
Computer pattern recognition
Conditional probabilities
Cost functions
Datasets
Decision making
Decision theory
Diabetes
empirical risk minimization
Error rates
Machine learning
margin condition
Mathematical functions
Mathematical models
Observational learning
plug-in rules
reject option
Statistical methods
Statistical models
Studies
ISSN:0319-5724, 1708-945X
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The authors study binary classification that allows for a reject option in which case no decision is made. This reject option is to be used for those observations for which the conditional class probabilities are close and as such are hard to classify. The authors generalize existing theory for both plug-in rules and empirical risk minimizers to this setting. /// Les auteurs s'intéressent à une méthode de classification dichotomique permettant de laisser en suspens le classement de certaines observations dont les probabilités conditionnelles d'appartenance à l'une ou l'autre classe sont si proches qu'un choix est difficile à faire. Les auteurs étudient dans ce contexte les propriétés des règles de substitution et des règles de classement minimisant le risque empirique.
Bibliography:ArticleID:CJS5550340410
istex:D2DC216F681BEC75753640CF10CFEA8FD52F52A3
ark:/67375/WNG-FDFT6069-L
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 14
ObjectType-Article-2
content type line 23
ISSN:0319-5724
1708-945X
DOI:10.1002/cjs.5550340410