A Review of Nonconformity Measures for Conformal Prediction in Regression

Yuko Kato, David M.J. Tax, Marco Loog
Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications, PMLR 204:369-383, 2023.

Abstract

Conformal prediction provides distribution-free uncertainty quantification under minimal assumptions. An important ingredient in conformal prediction is the so-called nonconformity measure, which quantifies how the test sample differs from the rest of the data. In this paper, existing nonconformity measures from the current literature are collected and their underlying ideas are analyzed. Furthermore, the influence of different factors on the performance of conformal prediction are pointed out by focusing on the relation between the influencing factors and the choice of nonconformity measures. Lastly, we provide suggestions for future work with regard to currently existing knowledge gaps and development of new nonconformity measures.

Cite this Paper

BibTeX
@InProceedings{pmlr-v204-kato23a, title = {A Review of Nonconformity Measures for Conformal Prediction in Regression}, author = {Kato, Yuko and Tax, David M.J. and Loog, Marco}, booktitle = {Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications}, pages = {369--383}, year = {2023}, editor = {Papadopoulos, Harris and Nguyen, Khuong An and Boström, Henrik and Carlsson, Lars}, volume = {204}, series = {Proceedings of Machine Learning Research}, month = {13--15 Sep}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v204/kato23a/kato23a.pdf}, url = {https://proceedings.mlr.press/v204/kato23a.html}, abstract = {Conformal prediction provides distribution-free uncertainty quantification under minimal assumptions. An important ingredient in conformal prediction is the so-called nonconformity measure, which quantifies how the test sample differs from the rest of the data. In this paper, existing nonconformity measures from the current literature are collected and their underlying ideas are analyzed. Furthermore, the influence of different factors on the performance of conformal prediction are pointed out by focusing on the relation between the influencing factors and the choice of nonconformity measures. Lastly, we provide suggestions for future work with regard to currently existing knowledge gaps and development of new nonconformity measures.} }
Endnote
%0 Conference Paper %T A Review of Nonconformity Measures for Conformal Prediction in Regression %A Yuko Kato %A David M.J. Tax %A Marco Loog %B Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications %C Proceedings of Machine Learning Research %D 2023 %E Harris Papadopoulos %E Khuong An Nguyen %E Henrik Boström %E Lars Carlsson %F pmlr-v204-kato23a %I PMLR %P 369--383 %U https://proceedings.mlr.press/v204/kato23a.html %V 204 %X Conformal prediction provides distribution-free uncertainty quantification under minimal assumptions. An important ingredient in conformal prediction is the so-called nonconformity measure, which quantifies how the test sample differs from the rest of the data. In this paper, existing nonconformity measures from the current literature are collected and their underlying ideas are analyzed. Furthermore, the influence of different factors on the performance of conformal prediction are pointed out by focusing on the relation between the influencing factors and the choice of nonconformity measures. Lastly, we provide suggestions for future work with regard to currently existing knowledge gaps and development of new nonconformity measures.
APA
Kato, Y., Tax, D.M. & Loog, M.. (2023). A Review of Nonconformity Measures for Conformal Prediction in Regression. Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications, in Proceedings of Machine Learning Research 204:369-383 Available from https://proceedings.mlr.press/v204/kato23a.html.

Related Material