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

Deep models are highly susceptible to adversarial attacks. Such attacks arecarefully crafted imperceptible noises that can fool the network and can causesevere consequences when deployed. To encounter them, the model requirestraining data for adversarial training or explicit regularization-basedtechniques. However, privacy has become an important concern, restrictingaccess to only trained models but not the training data (e.g. biometric data).Also, data curation is expensive and companies may have proprietary rights overit. To handle such situations, we propose a completely novel problem of'test-time adversarial defense in absence of training data and even theirstatistics'. We solve it in two stages: a) detection and b) correction ofadversarial samples. Our adversarial sample detection framework is initiallytrained on arbitrary data and is subsequently adapted to the unlabelled testdata through unsupervised domain adaptation. We further correct the predictionson detected adversarial samples by transforming them in Fourier domain andobtaining their low frequency component at our proposed suitable radius formodel prediction. We demonstrate the efficacy of our proposed technique viaextensive experiments against several adversarial attacks and for differentmodel architectures and datasets. For a non-robust Resnet-18 model pre-trainedon CIFAR-10, our detection method correctly identifies 91.42% adversaries.Also, we significantly improve the adversarial accuracy from 0% to 37.37% witha minimal drop of 0.02% in clean accuracy on state-of-the-art 'Auto Attack'without having to retrain the model.

DAD: Data-free Adversarial Defense at Test Time