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

DNNs are known to be vulnerable to so-called adversarial attacks thatmanipulate inputs to cause incorrect results that can be beneficial to anattacker or damaging to the victim. Recent works have proposed approximatecomputation as a defense mechanism against machine learning attacks. We showthat these approaches, while successful for a range of inputs, are insufficientto address stronger, high-confidence adversarial attacks. To address this, wepropose DNNSHIELD, a hardware-accelerated defense that adapts the strength ofthe response to the confidence of the adversarial input. Our approach relies ondynamic and random sparsification of the DNN model to achieve inferenceapproximation efficiently and with fine-grain control over the approximationerror. DNNSHIELD uses the output distribution characteristics of sparsifiedinference compared to a dense reference to detect adversarial inputs. We showan adversarial detection rate of 86% when applied to VGG16 and 88% when appliedto ResNet50, which exceeds the detection rate of the state of the artapproaches, with a much lower overhead. We demonstrate asoftware/hardware-accelerated FPGA prototype, which reduces the performanceimpact of DNNSHIELD relative to software-only CPU and GPU implementations.

DNNShield: Dynamic Randomized Model Sparsification, A Defense Against Adversarial Machine Learning