Robustness and uncertainty: two complementary aspects of the reliability of the predictions of a classifier

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Detavernier, Adrián, De Bock, Jasper
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915681996374016
author Detavernier, Adrián
De Bock, Jasper
author_facet Detavernier, Adrián
De Bock, Jasper
contents We consider two conceptually different approaches for assessing the reliability of the individual predictions of a classifier: Robustness Quantification (RQ) and Uncertainty Quantification (UQ). We compare both approaches on a number of benchmark datasets and show that there is no clear winner between the two, but that they are complementary and can be combined to obtain a hybrid approach that outperforms both RQ and UQ. As a byproduct of our approach, for each dataset, we also obtain an assessment of the relative importance of uncertainty and robustness as sources of unreliability.
format Preprint
id arxiv_https___arxiv_org_abs_2512_15492
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Robustness and uncertainty: two complementary aspects of the reliability of the predictions of a classifier
Detavernier, Adrián
De Bock, Jasper
Machine Learning
We consider two conceptually different approaches for assessing the reliability of the individual predictions of a classifier: Robustness Quantification (RQ) and Uncertainty Quantification (UQ). We compare both approaches on a number of benchmark datasets and show that there is no clear winner between the two, but that they are complementary and can be combined to obtain a hybrid approach that outperforms both RQ and UQ. As a byproduct of our approach, for each dataset, we also obtain an assessment of the relative importance of uncertainty and robustness as sources of unreliability.
title Robustness and uncertainty: two complementary aspects of the reliability of the predictions of a classifier
topic Machine Learning
url https://arxiv.org/abs/2512.15492