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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Artificial intelligence, and specifically deep neural networks (DNNs), hasrapidly emerged in the past decade as the standard for several tasks fromspecific advertising to object detection. The performance offered has led DNNalgorithms to become a part of critical embedded systems, requiring bothefficiency and reliability. In particular, DNNs are subject to maliciousexamples designed in a way to fool the network while being undetectable to thehuman observer: the adversarial examples. While previous studies proposeframeworks to implement such attacks in black box settings, those often rely onthe hypothesis that the attacker has access to the logits of the neuralnetwork, breaking the assumption of the traditional black box. In this paper,we investigate a real black box scenario where the attacker has no access tothe logits. In particular, we propose an architecture-agnostic attack whichsolve this constraint by extracting the logits. Our method combines hardwareand software attacks, by performing a side-channel attack that exploitselectromagnetic leakages to extract the logits for a given input, allowing anattacker to estimate the gradients and produce state-of-the-art adversarialexamples to fool the targeted neural network. Through this example ofadversarial attack, we demonstrate the effectiveness of logits extraction usingside-channel as a first step for more general attack frameworks requiringeither the logits or the confidence scores.

When Side-Channel Attacks Break the Black-Box Property of Embedded Artificial Intelligence