Investigating the Corruption Robustness of Image Classifiers with Random p-norm Corruptions

Georg Siedel, Georg Siedel, Weijia Shao, Silvia Vock, Andrey Morozov

2024

Abstract

Robustness is a fundamental property of machine learning classifiers required to achieve safety and reliability. In the field of adversarial robustness of image classifiers, robustness is commonly defined as the stability of a model to all input changes within a p-norm distance. However, in the field of random corruption robustness, variations observed in the real world are used, while p-norm corruptions are rarely considered. This study investigates the use of random p-norm corruptions to augment the training and test data of image classifiers. We evaluate the model robustness against imperceptible random p-norm corruptions and propose a novel robustness metric. We empirically investigate whether robustness transfers across different p-norms and derive conclusions on which p-norm corruptions a model should be trained and evaluated. We find that training data augmentation with a combination of p-norm corruptions significantly improves corruption robustness, even on top of state-of-the-art data augmentation schemes.

Paper Citation

in Harvard Style

Siedel G., Shao W., Vock S. and Morozov A. (2024). Investigating the Corruption Robustness of Image Classifiers with Random p-norm Corruptions. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP; ISBN 978-989-758-679-8, SciTePress, pages 171-181. DOI: 10.5220/0012397100003660

in Bibtex Style

@conference{visapp24,
author={Georg Siedel and Weijia Shao and Silvia Vock and Andrey Morozov},
title={Investigating the Corruption Robustness of Image Classifiers with Random p-norm Corruptions},
booktitle={Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP},
year={2024},
pages={171-181},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012397100003660},
isbn={978-989-758-679-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP
TI - Investigating the Corruption Robustness of Image Classifiers with Random p-norm Corruptions
SN - 978-989-758-679-8
AU - Siedel G.
AU - Shao W.
AU - Vock S.
AU - Morozov A.
PY - 2024
SP - 171
EP - 181
DO - 10.5220/0012397100003660
PB - SciTePress