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

Unrestricted adversarial attacks present a serious threat to deep learningmodels and adversarial defense techniques. They pose severe security problemsfor deep learning applications because they can effectively bypass defensemechanisms. However, previous attack methods often utilize GenerativeAdversarial Networks (GANs), which are not theoretically provable and thusgenerate unrealistic examples by incorporating adversarial objectives,especially for large-scale datasets like ImageNet. In this paper, we propose anew method, called AdvDiff, to generate unrestricted adversarial examples withdiffusion models. We design two novel adversarial guidance techniques toconduct adversarial sampling in the reverse generation process of diffusionmodels. These two techniques are effective and stable to generate high-quality,realistic adversarial examples by integrating gradients of the targetclassifier interpretably. Experimental results on MNIST and ImageNet datasetsdemonstrate that AdvDiff is effective to generate unrestricted adversarialexamples, which outperforms GAN-based methods in terms of attack performanceand generation quality.

AdvDiff: Generating Unrestricted Adversarial Examples using Diffusion Models