Unfolding Local Growth Rate Estimates for (Almost) Perfect Adversarial Detection

Peter Lorenz, Margret Keuper, Margret Keuper, Janis Keuper, Janis Keuper

2023

Abstract

Convolutional neural networks (CNN) define the state-of-the-art solution on many perceptual tasks. However, current CNN approaches largely remain vulnerable against adversarial perturbations of the input that have been crafted specifically to fool the system while being quasi-imperceptible to the human eye. In recent years, various approaches have been proposed to defend CNNs against such attacks, for example by model hardening or by adding explicit defence mechanisms. Thereby, a small “detector” is included in the network and trained on the binary classification task of distinguishing genuine data from data containing adversarial perturbations. In this work, we propose a simple and light-weight detector, which leverages recent findings on the relation between networks’ local intrinsic dimensionality (LID) and adversarial attacks. Based on a re-interpretation of the LID measure and several simple adaptations, we surpass the state-of-the-art on adversarial detection by a significant margin and reach almost perfect results in terms of F1-score for several networks and datasets. Sources available at: https://github.com/adverML/multiLID

Paper Citation

in Harvard Style

Lorenz P., Keuper M. and Keuper J. (2023). Unfolding Local Growth Rate Estimates for (Almost) Perfect Adversarial Detection. In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP; ISBN 978-989-758-634-7, SciTePress, pages 27-38. DOI: 10.5220/0011586500003417

in Bibtex Style

@conference{visapp23,
author={Peter Lorenz and Margret Keuper and Janis Keuper},
title={Unfolding Local Growth Rate Estimates for (Almost) Perfect Adversarial Detection},
booktitle={Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP},
year={2023},
pages={27-38},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011586500003417},
isbn={978-989-758-634-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP
TI - Unfolding Local Growth Rate Estimates for (Almost) Perfect Adversarial Detection
SN - 978-989-758-634-7
AU - Lorenz P.
AU - Keuper M.
AU - Keuper J.
PY - 2023
SP - 27
EP - 38
DO - 10.5220/0011586500003417
PB - SciTePress