To compress or not to compress: Understanding the Interactions between Adversarial Attacks and Neural Network Compression
Published version
Peer-reviewed
Repository URI
Repository DOI
Change log
Authors
Zhao, Yiren
Shumailov, Ilia
Mullins, Robert https://orcid.org/0000-0002-8393-2748
Anderson, Ross https://orcid.org/0000-0001-8697-5682
Abstract
As deep neural networks (DNNs) become widely used, pruned and quantised models are becoming ubiquitous on edge devices; such compressed DNNs are popular for lowering computational requirements.Meanwhile, recent studies show that adversarial samples can be effective at making DNNs misclassify. We, therefore, investigate the extent to which adversarial samples are transferable between uncompressed and compressed DNNs. We find that adversarial samples remain transferable for both pruned and quantised models.For pruning, the adversarial samples generated from heavily pruned models remain effective on uncompressed models. For quantisation, we find the transferability of adversarial samples is highly sensitive to integer precision.
Description
Keywords
Journal Title
Conference Name
The Conference on Systems and Machine Learning (SysML)
Journal ISSN
Volume Title
Publisher
Publisher DOI
Rights
All rights reserved
Sponsorship
Partially supported with funds from Bosch-Forschungsstiftung im Stifterverband