Guardado en:
Detalles Bibliográficos
Autores principales: Rathore, Kunal, Tadepalli, Prasad
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2509.15393
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918144148242432
author Rathore, Kunal
Tadepalli, Prasad
author_facet Rathore, Kunal
Tadepalli, Prasad
contents Deep learning models are notoriously opaque. Existing explanation methods often focus on localized visual explanations for individual images. Concept-based explanations, while offering global insights, require extensive annotations, incurring significant labeling cost. We propose an approach that leverages user-defined part labels from a limited set of images and efficiently transfers them to a larger dataset. This enables the generation of global symbolic explanations by aggregating part-based local explanations, ultimately providing human-understandable explanations for model decisions on a large scale.
format Preprint
id arxiv_https___arxiv_org_abs_2509_15393
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Generating Part-Based Global Explanations Via Correspondence
Rathore, Kunal
Tadepalli, Prasad
Computer Vision and Pattern Recognition
Artificial Intelligence
Deep learning models are notoriously opaque. Existing explanation methods often focus on localized visual explanations for individual images. Concept-based explanations, while offering global insights, require extensive annotations, incurring significant labeling cost. We propose an approach that leverages user-defined part labels from a limited set of images and efficiently transfers them to a larger dataset. This enables the generation of global symbolic explanations by aggregating part-based local explanations, ultimately providing human-understandable explanations for model decisions on a large scale.
title Generating Part-Based Global Explanations Via Correspondence
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2509.15393