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

Spiking neural networks (SNNs) have attracted much attention for their highenergy efficiency and for recent advances in their classification performance.However, unlike traditional deep learning approaches, the analysis and study ofthe robustness of SNNs to adversarial examples remain relativelyunderdeveloped. In this work, we focus on advancing the adversarial attack sideof SNNs and make three major contributions. First, we show that successfulwhite-box adversarial attacks on SNNs are highly dependent on the underlyingsurrogate gradient technique, even in the case of adversarially trained SNNs.Second, using the best surrogate gradient technique, we analyze thetransferability of adversarial attacks on SNNs and other state-of-the-artarchitectures like Vision Transformers (ViTs) and Big Transfer ConvolutionalNeural Networks (CNNs). We demonstrate that the adversarial examples created bynon-SNN architectures are not misclassified often by SNNs. Third, due to thelack of an ubiquitous white-box attack that is effective across both the SNNand CNN/ViT domains, we develop a new white-box attack, the Auto Self-AttentionGradient Attack (Auto-SAGA). Our novel attack generates adversarial examplescapable of fooling both SNN and non-SNN models simultaneously. Auto-SAGA is asmuch as $91.1\%$ more effective on SNN/ViT model ensembles and provides a$3\times$ boost in attack effectiveness on adversarially trained SNN ensemblescompared to conventional white-box attacks like Auto-PGD. Our experiments andanalyses are broad and rigorous covering three datasets (CIFAR-10, CIFAR-100and ImageNet), five different white-box attacks and nineteen classifier models(seven for each CIFAR dataset and five models for ImageNet).

Attacking the Spike: On the Transferability and Security of Spiking Neural Networks to Adversarial Examples