Towards a general framework for improving the performance of classifiers using XAI methods

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Apicella, Andrea, Giugliano, Salvatore, Isgrò, Francesco, Prevete, Roberto
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914715573157888
author Apicella, Andrea
Giugliano, Salvatore
Isgrò, Francesco
Prevete, Roberto
author_facet Apicella, Andrea
Giugliano, Salvatore
Isgrò, Francesco
Prevete, Roberto
contents Modern Artificial Intelligence (AI) systems, especially Deep Learning (DL) models, poses challenges in understanding their inner workings by AI researchers. eXplainable Artificial Intelligence (XAI) inspects internal mechanisms of AI models providing explanations about their decisions. While current XAI research predominantly concentrates on explaining AI systems, there is a growing interest in using XAI techniques to automatically improve the performance of AI systems themselves. This paper proposes a general framework for automatically improving the performance of pre-trained DL classifiers using XAI methods, avoiding the computational overhead associated with retraining complex models from scratch. In particular, we outline the possibility of two different learning strategies for implementing this architecture, which we will call auto-encoder-based and encoder-decoder-based, and discuss their key aspects.
format Preprint
id arxiv_https___arxiv_org_abs_2403_10373
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Towards a general framework for improving the performance of classifiers using XAI methods
Apicella, Andrea
Giugliano, Salvatore
Isgrò, Francesco
Prevete, Roberto
Machine Learning
Modern Artificial Intelligence (AI) systems, especially Deep Learning (DL) models, poses challenges in understanding their inner workings by AI researchers. eXplainable Artificial Intelligence (XAI) inspects internal mechanisms of AI models providing explanations about their decisions. While current XAI research predominantly concentrates on explaining AI systems, there is a growing interest in using XAI techniques to automatically improve the performance of AI systems themselves. This paper proposes a general framework for automatically improving the performance of pre-trained DL classifiers using XAI methods, avoiding the computational overhead associated with retraining complex models from scratch. In particular, we outline the possibility of two different learning strategies for implementing this architecture, which we will call auto-encoder-based and encoder-decoder-based, and discuss their key aspects.
title Towards a general framework for improving the performance of classifiers using XAI methods
topic Machine Learning
url https://arxiv.org/abs/2403.10373