Criteria of efficiency for conformal prediction
Machine Learning
2016-09-15 v2
Abstract
We study optimal conformity measures for various criteria of efficiency of classification in an idealised setting. This leads to an important class of criteria of efficiency that we call probabilistic; it turns out that the most standard criteria of efficiency used in literature on conformal prediction are not probabilistic unless the problem of classification is binary. We consider both unconditional and label-conditional conformal prediction.
Keywords
Cite
@article{arxiv.1603.04416,
title = {Criteria of efficiency for conformal prediction},
author = {Vladimir Vovk and Ilia Nouretdinov and Valentina Fedorova and Ivan Petej and Alex Gammerman},
journal= {arXiv preprint arXiv:1603.04416},
year = {2016}
}
Comments
31 pages