Volltext-Downloads (blau) und Frontdoor-Views (grau)

SpectralDefense: Detecting Adversarial Attacks on CNNs in the Fourier Domain

  • Despite the success of convolutional neural networks (CNNs) in many computer vision and image analysis tasks, they remain vulnerable against so-called adversarial attacks: Small, crafted perturbations in the input images can lead to false predictions. A possible defense is to detect adversarial examples. In this work, we show how analysis in the Fourier domain of input images and feature maps canDespite the success of convolutional neural networks (CNNs) in many computer vision and image analysis tasks, they remain vulnerable against so-called adversarial attacks: Small, crafted perturbations in the input images can lead to false predictions. A possible defense is to detect adversarial examples. In this work, we show how analysis in the Fourier domain of input images and feature maps can be used to distinguish benign test samples from adversarial images. We propose two novel detection methods: Our first method employs the magnitude spectrum of the input images to detect an adversarial attack. This simple and robust classifier can successfully detect adversarial perturbations of three commonly used attack methods. The second method builds upon the first and additionally extracts the phase of Fourier coefficients of feature-maps at different layers of the network. With this extension, we are able to improve adversarial detection rates compared to state-of-the-art detectors on five different attack methods. The code for the methods proposed in the paper is available at github.com/paulaharder/SpectralAdversarialDefense…zeige mehrzeige weniger

Metadaten exportieren

Weitere Dienste

Suche bei Google Scholar

Statistik

frontdoor_oas
Metadaten
Dokumentart:Konferenzveröffentlichung
Art der Konferenzveröffentlichung:Konferenzartikel
Zitierlink: https://opus.hs-offenburg.de/5284
Bibliografische Angaben
Titel (Englisch):SpectralDefense: Detecting Adversarial Attacks on CNNs in the Fourier Domain
Konferenzangaben:2021 International Joint Conference on Neural Networks (IJCNN), 18-22 July 2021, Shenzhen, China
Verfasserangaben:Paula HarderORCiD, Franz-Josef Pfreundt, Margret KeuperORCiD, Janis KeuperStaff MemberORCiDGND
Erscheinungsjahr:2021
Verlag:IEEE
Seitenanzahl:8
Erste Seite:1
Letzte Seite:8
Titel des übergeordneten Werkes (Englisch):IJCNN 2021. The International Joint Conference on Neural Networks : 2021 Conference Proceedings
ISBN:978-1-6654-3900-8 (elektronisch)
ISBN:978-1-6654-4597-9 (Print on Demand)
ISSN:2161-4407 (elektronisch)
ISSN:2161-4393 (Print on Demand)
DOI:https://doi.org/10.1109/IJCNN52387.2021.9533442
URL:https://ieeexplore.ieee.org/document/9533442
Sprache:Englisch
Inhaltliche Informationen
Fakultäten / Einrichtungen:Fakultät Elektrotechnik, Medizintechnik und Informatik (EMI) (ab 04/2019)
Forschung:IMLA - Institute for Machine Learning and Analytics
Sammlungen der Hochschule Offenburg:Bibliografie
Freies Schlagwort / Tag:adversarial attacks; adversarial detection; convolutional neural networks; image classification
Formale Angaben
Open-Access-Status: Closed Access 
Lizenz (Deutsch):License LogoUrheberrechtlich geschützt