Leveraging generative models to characterize the failure conditions of image classifiers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: LeCoz, Adrien, Herbin, Stéphane, Adjed, Faouzi
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929576084504576
author LeCoz, Adrien
Herbin, Stéphane
Adjed, Faouzi
author_facet LeCoz, Adrien
Herbin, Stéphane
Adjed, Faouzi
contents We address in this work the question of identifying the failure conditions of a given image classifier. To do so, we exploit the capacity of producing controllable distributions of high quality image data made available by recent Generative Adversarial Networks (StyleGAN2): the failure conditions are expressed as directions of strong performance degradation in the generative model latent space. This strategy of analysis is used to discover corner cases that combine multiple sources of corruption, and to compare in more details the behavior of different classifiers. The directions of degradation can also be rendered visually by generating data for better interpretability. Some degradations such as image quality can affect all classes, whereas other ones such as shape are more class-specific. The approach is demonstrated on the MNIST dataset that has been completed by two sources of corruption: noise and blur, and shows a promising way to better understand and control the risks of exploiting Artificial Intelligence components for safety-critical applications.
format Preprint
id arxiv_https___arxiv_org_abs_2410_12814
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Leveraging generative models to characterize the failure conditions of image classifiers
LeCoz, Adrien
Herbin, Stéphane
Adjed, Faouzi
Computer Vision and Pattern Recognition
We address in this work the question of identifying the failure conditions of a given image classifier. To do so, we exploit the capacity of producing controllable distributions of high quality image data made available by recent Generative Adversarial Networks (StyleGAN2): the failure conditions are expressed as directions of strong performance degradation in the generative model latent space. This strategy of analysis is used to discover corner cases that combine multiple sources of corruption, and to compare in more details the behavior of different classifiers. The directions of degradation can also be rendered visually by generating data for better interpretability. Some degradations such as image quality can affect all classes, whereas other ones such as shape are more class-specific. The approach is demonstrated on the MNIST dataset that has been completed by two sources of corruption: noise and blur, and shows a promising way to better understand and control the risks of exploiting Artificial Intelligence components for safety-critical applications.
title Leveraging generative models to characterize the failure conditions of image classifiers
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.12814