On the Expected Size of Conformal Prediction Sets

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dhillon, Guneet S., Deligiannidis, George, Rainforth, Tom
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911792053092352
author Dhillon, Guneet S.
Deligiannidis, George
Rainforth, Tom
author_facet Dhillon, Guneet S.
Deligiannidis, George
Rainforth, Tom
contents While conformal predictors reap the benefits of rigorous statistical guarantees on their error frequency, the size of their corresponding prediction sets is critical to their practical utility. Unfortunately, there is currently a lack of finite-sample analysis and guarantees for their prediction set sizes. To address this shortfall, we theoretically quantify the expected size of the prediction sets under the split conformal prediction framework. As this precise formulation cannot usually be calculated directly, we further derive point estimates and high-probability interval bounds that can be empirically computed, providing a practical method for characterizing the expected set size. We corroborate the efficacy of our results with experiments on real-world datasets for both regression and classification problems.
format Preprint
id arxiv_https___arxiv_org_abs_2306_07254
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle On the Expected Size of Conformal Prediction Sets
Dhillon, Guneet S.
Deligiannidis, George
Rainforth, Tom
Machine Learning
While conformal predictors reap the benefits of rigorous statistical guarantees on their error frequency, the size of their corresponding prediction sets is critical to their practical utility. Unfortunately, there is currently a lack of finite-sample analysis and guarantees for their prediction set sizes. To address this shortfall, we theoretically quantify the expected size of the prediction sets under the split conformal prediction framework. As this precise formulation cannot usually be calculated directly, we further derive point estimates and high-probability interval bounds that can be empirically computed, providing a practical method for characterizing the expected set size. We corroborate the efficacy of our results with experiments on real-world datasets for both regression and classification problems.
title On the Expected Size of Conformal Prediction Sets
topic Machine Learning
url https://arxiv.org/abs/2306.07254