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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Deep Learning is able to solve a plethora of once impossible problems.However, they are vulnerable to input adversarial attacks preventing them frombeing autonomously deployed in critical applications. Severalalgorithm-centered works have discussed methods to cause adversarial attacksand improve adversarial robustness of a Deep Neural Network (DNN). In thiswork, we elicit the advantages and vulnerabilities of hybrid 6T-8T memories toimprove the adversarial robustness and cause adversarial attacks on DNNs. Weshow that bit-error noise in hybrid memories due to erroneous 6T-SRAM cellshave deterministic behaviour based on the hybrid memory configurations (V_DD,8T-6T ratio). This controlled noise (surgical noise) can be strategicallyintroduced into specific DNN layers to improve the adversarial accuracy ofDNNs. At the same time, surgical noise can be carefully injected into the DNNparameters stored in hybrid memory to cause adversarial attacks. To improve theadversarial robustness of DNNs using surgical noise, we propose a methodologyto select appropriate DNN layers and their corresponding hybrid memoryconfigurations to introduce the required surgical noise. Using this, we achieve2-8% higher adversarial accuracy without re-training against white-box attackslike FGSM, than the baseline models (with no surgical noise introduced). Todemonstrate adversarial attacks using surgical noise, we design a novel,white-box attack on DNN parameters stored in hybrid memory banks that causesthe DNN inference accuracy to drop by more than 60% with over 90% confidencevalue. We support our claims with experiments, performed using benchmarkdatasets-CIFAR10 and CIFAR100 on VGG19 and ResNet18 networks.

Exposing the Robustness and Vulnerability of Hybrid 8T-6T SRAM Memory Architectures to Adversarial Attacks in Deep Neural Networks