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

Low-latency deep spiking neural networks (SNNs) have become a promisingalternative to conventional artificial neural networks (ANNs) because of theirpotential for increased energy efficiency on event-driven neuromorphichardware. Neural networks, including SNNs, however, are subject to variousadversarial attacks and must be trained to remain resilient against suchattacks for many applications. Nevertheless, due to prohibitively high trainingcosts associated with SNNs, analysis, and optimization of deep SNNs undervarious adversarial attacks have been largely overlooked. In this paper, wefirst present a detailed analysis of the inherent robustness of low-latencySNNs against popular gradient-based attacks, namely fast gradient sign method(FGSM) and projected gradient descent (PGD). Motivated by this analysis, toharness the model robustness against these attacks we present an SNN trainingalgorithm that uses crafted input noise and incurs no additional training time.To evaluate the merits of our algorithm, we conducted extensive experimentswith variants of VGG and ResNet on both CIFAR-10 and CIFAR-100 datasets.Compared to standard trained direct input SNNs, our trained models yieldimproved classification accuracy of up to 13.7% and 10.1% on FGSM and PGDattack-generated images, respectively, with negligible loss in clean imageaccuracy. Our models also outperform inherently robust SNNs trained onrate-coded inputs with improved or similar classification performance onattack-generated images while having up to 25x and 4.6x lower latency andcomputation energy, respectively.

HIRE-SNN: Harnessing the Inherent Robustness of Energy-Efficient Deep Spiking Neural Networks by Training with Crafted Input Noise