License: Creative Commons Attribution 3.0 Unported license (CC BY 3.0)
When quoting this document, please refer to the following
DOI: 10.4230/OASIcs.ICCSW.2013.27
URN: urn:nbn:de:0030-drops-42685
URL: http://dagstuhl.sunsite.rwth-aachen.de/volltexte/2013/4268/
Fedorova, Valentina ;
Gammerman, Alex ;
Nouretdinov, Ilia ;
Vovk, Vladimir
Conformal Prediction under Hypergraphical Models
Abstract
Conformal predictors are usually defined and studied under the exchangeability assumption. However, their definition can be extended to a wide class of statistical models, called online compression models, while retaining their property of automatic validity. This paper is devoted to conformal prediction under hypergraphical models that are more specific than the exchangeability model. We define conformity measures for such hypergraphical models and study the corresponding conformal predictors empirically on benchmark LED data sets. Our experiments show that they are more efficient than conformal predictors that use only the exchangeability assumption.
BibTeX - Entry
@InProceedings{fedorova_et_al:OASIcs:2013:4268,
author = {Valentina Fedorova and Alex Gammerman and Ilia Nouretdinov and Vladimir Vovk},
title = {{Conformal Prediction under Hypergraphical Models}},
booktitle = {2013 Imperial College Computing Student Workshop},
pages = {27--34},
series = {OpenAccess Series in Informatics (OASIcs)},
ISBN = {978-3-939897-63-7},
ISSN = {2190-6807},
year = {2013},
volume = {35},
editor = {Andrew V. Jones and Nicholas Ng},
publisher = {Schloss Dagstuhl--Leibniz-Zentrum fuer Informatik},
address = {Dagstuhl, Germany},
URL = {http://drops.dagstuhl.de/opus/volltexte/2013/4268},
URN = {urn:nbn:de:0030-drops-42685},
doi = {10.4230/OASIcs.ICCSW.2013.27},
annote = {Keywords: conformal prediction, hypergraphical models, conformity measure}
}
Keywords: |
|
conformal prediction, hypergraphical models, conformity measure |
Collection: |
|
2013 Imperial College Computing Student Workshop |
Issue Date: |
|
2013 |
Date of publication: |
|
14.10.2013 |