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

Convolutional neural networks (CNNs) have achieved beyond human-levelaccuracy in the image classification task and are widely deployed in real-worldenvironments. However, CNNs show vulnerability to adversarial perturbationsthat are well-designed noises aiming to mislead the classification models. Inorder to defend against the adversarial perturbations, adversarially trainedGAN (ATGAN) is proposed to improve the adversarial robustness generalization ofthe state-of-the-art CNNs trained by adversarial training. ATGAN incorporatesadversarial training into standard GAN training procedure to remove obfuscatedgradients which can lead to a false sense in defending against the adversarialperturbations and are commonly observed in existing GANs-based adversarialdefense methods. Moreover, ATGAN adopts the image-to-image generator as dataaugmentation to increase the sample complexity needed for adversarialrobustness generalization in adversarial training. Experimental results inMNIST SVHN and CIFAR-10 datasets show that the proposed method doesn't rely onobfuscated gradients and achieves better global adversarial robustnessgeneralization performance than the adversarially trained state-of-the-artCNNs.

Improving Global Adversarial Robustness Generalization With Adversarially Trained GAN