English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

State-of-the-art deep neural networks (DNNs) have been proven to bevulnerable to adversarial manipulation and backdoor attacks. Backdoored modelsdeviate from expected behavior on inputs with predefined triggers whileretaining performance on clean data. Recent works focus on software simulationof backdoor injection during the inference phase by modifying network weights,which we find often unrealistic in practice due to restrictions in hardware.  In contrast, in this work for the first time, we present an end-to-endbackdoor injection attack realized on actual hardware on a classifier modelusing Rowhammer as the fault injection method. To this end, we firstinvestigate the viability of backdoor injection attacks in real-lifedeployments of DNNs on hardware and address such practical issues in hardwareimplementation from a novel optimization perspective. We are motivated by thefact that vulnerable memory locations are very rare, device-specific, andsparsely distributed. Consequently, we propose a novel network trainingalgorithm based on constrained optimization to achieve a realistic backdoorinjection attack in hardware. By modifying parameters uniformly across theconvolutional and fully-connected layers as well as optimizing the triggerpattern together, we achieve state-of-the-art attack performance with fewer bitflips. For instance, our method on a hardware-deployed ResNet-20 model trainedon CIFAR-10 achieves over 89% test accuracy and 92% attack success rate byflipping only 10 out of 2.2 million bits.

Don't Knock! Rowhammer at the Backdoor of DNN Models