Enhancing Conformal Prediction Using E-Test Statistics

Alexander A. Balinsky, Alexander David Balinsky
Proceedings of the Thirteenth Symposium on Conformal and Probabilistic Prediction with Applications, PMLR 230:65-72, 2024.

Abstract

Conformal Prediction (CP) serves as a robust framework that quantifies uncertainty in predictions made by Machine Learning (ML) models. Unlike traditional point predictors, CP generates statistically valid prediction regions, also known as prediction intervals, based on the assumption of data exchangeability. Typically, the construction of conformal predictions hinges on p-values. This paper, however, ventures down an alternative path, harnessing the power of e-test statistics to augment the efficacy of conformal predictions by introducing a BB-predictor (bounded from the below predictor). The BB-predictor can be constructed under even more lenient assumptions than exchangeability.

Cite this Paper

BibTeX
@InProceedings{pmlr-v230-balinsky24a, title = {Enhancing Conformal Prediction Using E-Test Statistics}, author = {Balinsky, Alexander A. and Balinsky, Alexander David}, booktitle = {Proceedings of the Thirteenth Symposium on Conformal and Probabilistic Prediction with Applications}, pages = {65--72}, year = {2024}, editor = {Vantini, Simone and Fontana, Matteo and Solari, Aldo and Boström, Henrik and Carlsson, Lars}, volume = {230}, series = {Proceedings of Machine Learning Research}, month = {09--11 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v230/main/assets/balinsky24a/balinsky24a.pdf}, url = {https://proceedings.mlr.press/v230/balinsky24a.html}, abstract = {Conformal Prediction (CP) serves as a robust framework that quantifies uncertainty in predictions made by Machine Learning (ML) models. Unlike traditional point predictors, CP generates statistically valid prediction regions, also known as prediction intervals, based on the assumption of data exchangeability. Typically, the construction of conformal predictions hinges on p-values. This paper, however, ventures down an alternative path, harnessing the power of e-test statistics to augment the efficacy of conformal predictions by introducing a BB-predictor (bounded from the below predictor). The BB-predictor can be constructed under even more lenient assumptions than exchangeability.} }
Endnote
%0 Conference Paper %T Enhancing Conformal Prediction Using E-Test Statistics %A Alexander A. Balinsky %A Alexander David Balinsky %B Proceedings of the Thirteenth Symposium on Conformal and Probabilistic Prediction with Applications %C Proceedings of Machine Learning Research %D 2024 %E Simone Vantini %E Matteo Fontana %E Aldo Solari %E Henrik Boström %E Lars Carlsson %F pmlr-v230-balinsky24a %I PMLR %P 65--72 %U https://proceedings.mlr.press/v230/balinsky24a.html %V 230 %X Conformal Prediction (CP) serves as a robust framework that quantifies uncertainty in predictions made by Machine Learning (ML) models. Unlike traditional point predictors, CP generates statistically valid prediction regions, also known as prediction intervals, based on the assumption of data exchangeability. Typically, the construction of conformal predictions hinges on p-values. This paper, however, ventures down an alternative path, harnessing the power of e-test statistics to augment the efficacy of conformal predictions by introducing a BB-predictor (bounded from the below predictor). The BB-predictor can be constructed under even more lenient assumptions than exchangeability.
APA
Balinsky, A.A. & Balinsky, A.D.. (2024). Enhancing Conformal Prediction Using E-Test Statistics. Proceedings of the Thirteenth Symposium on Conformal and Probabilistic Prediction with Applications, in Proceedings of Machine Learning Research 230:65-72 Available from https://proceedings.mlr.press/v230/balinsky24a.html.

Related Material