Conformal Prediction under Hypergraphical Models

Fedorova, Valentina; Gammerman, Alex; Nouretdinov, Ilia; Vovk, Vladimir

doi:10.4230/OASIcs.ICCSW.2013.27

Document

Conformal Prediction under Hypergraphical Models

Authors Valentina Fedorova, Alex Gammerman, Ilia Nouretdinov, Vladimir Vovk

Part of: Volume: 2013 Imperial College Computing Student Workshop (ICCSW 2013)
Part of: Series: Open Access Series in Informatics (OASIcs)
License: Creative Commons Attribution 3.0 Unported license
Publication Date: 2013-10-14

PDF

File

PDF

OASIcs.ICCSW.2013.27.pdf

Filesize: 0.7 MB
8 pages

Document Identifiers

DOI: 10.4230/OASIcs.ICCSW.2013.27
URN: urn:nbn:de:0030-drops-42685

Subject Classification

Keywords

conformal prediction
hypergraphical models
conformity measure

Metrics

Access Statistics
Total Accesses (updated on a weekly basis)

0

PDF Downloads

0

Metadata Views

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.

Cite As Get BibTex

Valentina Fedorova, Alex Gammerman, Ilia Nouretdinov, and Vladimir Vovk. Conformal Prediction under Hypergraphical Models. In 2013 Imperial College Computing Student Workshop. Open Access Series in Informatics (OASIcs), Volume 35, pp. 27-34, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2013) https://doi.org/10.4230/OASIcs.ICCSW.2013.27

Author Details

Valentina Fedorova

Alex Gammerman

Ilia Nouretdinov

Vladimir Vovk

Questions / Remarks / Feedback

Feedback for Dagstuhl Publishing

Thanks for your feedback!

Feedback submitted

Could not send message

Please try again later or send an E-mail