Dynamic Backdoors with Global Average Pooling

Conference Paper (2022)
Author(s)

S. Koffas (TU Delft - Cyber Security)

S. Picek (Radboud Universiteit Nijmegen, TU Delft - Cyber Security)

M. Conti (TU Delft - Cyber Security, University of Padua)

Research Group
Cyber Security
Copyright
© 2022 S. Koffas, S. Picek, M. Conti
DOI related publication
https://doi.org/10.1109/AICAS54282.2022.9869920
More Info
expand_more
Publication Year
2022
Language
English
Copyright
© 2022 S. Koffas, S. Picek, M. Conti
Research Group
Cyber Security
Bibliographical Note
Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.@en
Pages (from-to)
320-323
ISBN (print)
978-1-6654-0997-1
ISBN (electronic)
978-1-6654-0996-4
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Outsourced training and machine learning as a service have resulted in novel attack vectors like backdoor attacks. Such attacks embed a secret functionality in a neural network activated when the trigger is added to its input. In most works in the literature, the trigger is static, both in terms of location and pattern. The effectiveness of various detection mechanisms depends on this property. It was recently shown that countermeasures in image classification, like Neural Cleanse and ABS, could be bypassed with dynamic triggers that are effective regardless of their pattern and location. Still, such backdoors are demanding as they require a large percentage of poisoned training data. In this work, we are the first to show that dynamic backdoor attacks could happen due to a global average pooling layer without increasing the percentage of the poisoned training data. Nevertheless, our experiments in sound classification, text sentiment analysis, and image classification show this to be very difficult in practice.

Files

Dynamic_Backdoors_with_Global_... (pdf)
(pdf | 1.86 Mb)
- Embargo expired in 05-03-2023
License info not available