Meta-evaluating stability measures: MAX-Senstivity & AVG-Sensitivity

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Miró-Nicolau, Miquel, Jaume-i-Capó, Antoni, Moyà-Alcover, Gabriel
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909428406550528
author Miró-Nicolau, Miquel
Jaume-i-Capó, Antoni
Moyà-Alcover, Gabriel
author_facet Miró-Nicolau, Miquel
Jaume-i-Capó, Antoni
Moyà-Alcover, Gabriel
contents The use of eXplainable Artificial Intelligence (XAI) systems has introduced a set of challenges that need resolution. The XAI robustness, or stability, has been one of the goals of the community from its beginning. Multiple authors have proposed evaluating this feature using objective evaluation measures. Nonetheless, many questions remain. With this work, we propose a novel approach to meta-evaluate these metrics, i.e. analyze the correctness of the evaluators. We propose two new tests that allowed us to evaluate two different stability measures: AVG-Sensitiviy and MAX-Senstivity. We tested their reliability in the presence of perfect and robust explanations, generated with a Decision Tree; as well as completely random explanations and prediction. The metrics results showed their incapacity of identify as erroneous the random explanations, highlighting their overall unreliability.
format Preprint
id arxiv_https___arxiv_org_abs_2412_10942
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Meta-evaluating stability measures: MAX-Senstivity & AVG-Sensitivity
Miró-Nicolau, Miquel
Jaume-i-Capó, Antoni
Moyà-Alcover, Gabriel
Computer Vision and Pattern Recognition
The use of eXplainable Artificial Intelligence (XAI) systems has introduced a set of challenges that need resolution. The XAI robustness, or stability, has been one of the goals of the community from its beginning. Multiple authors have proposed evaluating this feature using objective evaluation measures. Nonetheless, many questions remain. With this work, we propose a novel approach to meta-evaluate these metrics, i.e. analyze the correctness of the evaluators. We propose two new tests that allowed us to evaluate two different stability measures: AVG-Sensitiviy and MAX-Senstivity. We tested their reliability in the presence of perfect and robust explanations, generated with a Decision Tree; as well as completely random explanations and prediction. The metrics results showed their incapacity of identify as erroneous the random explanations, highlighting their overall unreliability.
title Meta-evaluating stability measures: MAX-Senstivity & AVG-Sensitivity
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.10942