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

Despite the rapid development of adversarial machine learning, mostadversarial attack and defense researches mainly focus on theperturbation-based adversarial examples, which is constrained by the inputimages. In comparison with existing works, we propose non-constrainedadversarial examples, which are generated entirely from scratch without anyconstraint on the input. Unlike perturbation-based attacks, or the so-calledunrestricted adversarial attack which is still constrained by the input noise,we aim to learn the distribution of adversarial examples to generatenon-constrained but semantically meaningful adversarial examples. Followingthis spirit, we propose a novel attack framework called AT-GAN (AdversarialTransfer on Generative Adversarial Net). Specifically, we first develop anormal GAN model to learn the distribution of benign data, and then transferthe pre-trained GAN model to estimate the distribution of adversarial examplesfor the target model. In this way, AT-GAN can learn the distribution ofadversarial examples that is very close to the distribution of real data. Toour knowledge, this is the first work of building an adversarial generatormodel that could produce adversarial examples directly from any input noise.Extensive experiments and visualizations show that the proposed AT-GAN can veryefficiently generate diverse adversarial examples that are more realistic tohuman perception. In addition, AT-GAN yields higher attack success ratesagainst adversarially trained models under white-box attack setting andexhibits moderate transferability against black-box models.

AT-GAN: An Adversarial Generator Model for Non-constrained Adversarial Examples