A Fresh Look at Sanity Checks for Saliency Maps

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Hedström, Anna, Weber, Leander, Lapuschkin, Sebastian, Höhne, Marina
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914783662440448
author Hedström, Anna
Weber, Leander
Lapuschkin, Sebastian
Höhne, Marina
author_facet Hedström, Anna
Weber, Leander
Lapuschkin, Sebastian
Höhne, Marina
contents The Model Parameter Randomisation Test (MPRT) is highly recognised in the eXplainable Artificial Intelligence (XAI) community due to its fundamental evaluative criterion: explanations should be sensitive to the parameters of the model they seek to explain. However, recent studies have raised several methodological concerns for the empirical interpretation of MPRT. In response, we propose two modifications to the original test: Smooth MPRT and Efficient MPRT. The former reduces the impact of noise on evaluation outcomes via sampling, while the latter avoids the need for biased similarity measurements by re-interpreting the test through the increase in explanation complexity after full model randomisation. Our experiments show that these modifications enhance the metric reliability, facilitating a more trustworthy deployment of explanation methods.
format Preprint
id arxiv_https___arxiv_org_abs_2405_02383
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Fresh Look at Sanity Checks for Saliency Maps
Hedström, Anna
Weber, Leander
Lapuschkin, Sebastian
Höhne, Marina
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
The Model Parameter Randomisation Test (MPRT) is highly recognised in the eXplainable Artificial Intelligence (XAI) community due to its fundamental evaluative criterion: explanations should be sensitive to the parameters of the model they seek to explain. However, recent studies have raised several methodological concerns for the empirical interpretation of MPRT. In response, we propose two modifications to the original test: Smooth MPRT and Efficient MPRT. The former reduces the impact of noise on evaluation outcomes via sampling, while the latter avoids the need for biased similarity measurements by re-interpreting the test through the increase in explanation complexity after full model randomisation. Our experiments show that these modifications enhance the metric reliability, facilitating a more trustworthy deployment of explanation methods.
title A Fresh Look at Sanity Checks for Saliency Maps
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2405.02383