Automatic Concept Extraction with CRAFT
Project description
👋 CRAFT: Concept Recursive Activation FacTorization for Explainability (CVPR 2023)
This repository contains code for the paper:
CRAFT: Concept Recursive Activation FacTorization for Explainability, Thomas Fel*, Agustin Picard*, Louis Bethune*, Thibaut Boissin*, David Vigouroux, Julien Colin, Rémi Cadène, Thomas Serre. CVPR 2023, [arXiv].
The code is implemented and available for Pytorch & Tensorflow. A notebook for each of them is available: notebook Pytorch, notebook Tensorflow.
@inproceedings{fel2023craft,
title={CRAFT: Concept Recursive Activation FacTorization for Explainability},
author={Thomas, Fel and Agustin, Picard and Louis, Bethune and Thibaut, Boissin and David, Vigouroux and Julien, Colin and Rémi, Cadène and Thomas, Serre},
year={2023},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}
}
To go further
The code for the metrics and the other attribution methods used in the paper come from the Xplique toolbox.
Authors of the code
- Thomas Fel - thomas_fel@brown.edu, PhD Student DEEL (ANITI), Brown University
- Agustin Picard - agustin-martin.picard@irt-saintexupery.com, IRT Saint-exupéry, DEEL
- Louis Béthune - louis.bethune@univ-toulouse.fr, PhD Student DEEL (ANITI)
- Thibaut Boissin - thibaut.boissin@irt-saintexupery.com, IRT Saint-exupéry, DEEL
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Craft-xai-0.0.1.tar.gz
(13.4 kB
view hashes)
Built Distribution
Craft_xai-0.0.1-py3-none-any.whl
(17.7 kB
view hashes)
Close
Hashes for Craft_xai-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 99f9fdc179ad2e6c389f963fb724c8e4cb6e29e9a78ee9a0ddfb45dc3267de91 |
|
MD5 | aa738f3bbf94ee61ec0853ec5102d04e |
|
BLAKE2b-256 | 894b16a9f05ba40ebaa643de5df3989c644e436c1e6bcbd49e10219874e6bada |