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

Adversarial generative models, such as Generative Adversarial Networks(GANs), are widely applied for generating various types of data, i.e., images,text, and audio. Accordingly, its promising performance has led to theGAN-based adversarial attack methods in the white-box and black-box attackscenarios. The importance of transferable black-box attacks lies in theirability to be effective across different models and settings, more closelyaligning with real-world applications. However, it remains challenging toretain the performance in terms of transferable adversarial examples for suchmethods. Meanwhile, we observe that some enhanced gradient-based transferableadversarial attack algorithms require prolonged time for adversarial samplegeneration. Thus, in this work, we propose a novel algorithm named GE-AdvGAN toenhance the transferability of adversarial samples whilst improving thealgorithm's efficiency. The main approach is via optimising the trainingprocess of the generator parameters. With the functional and characteristicsimilarity analysis, we introduce a novel gradient editing (GE) mechanism andverify its feasibility in generating transferable samples on various models.Moreover, by exploring the frequency domain information to determine thegradient editing direction, GE-AdvGAN can generate highly transferableadversarial samples while minimizing the execution time in comparison to thestate-of-the-art transferable adversarial attack algorithms. The performance ofGE-AdvGAN is comprehensively evaluated by large-scale experiments on differentdatasets, which results demonstrate the superiority of our algorithm. The codefor our algorithm is available at: https://github.com/LMBTough/GE-advGAN

GE-AdvGAN: Improving the transferability of adversarial samples by gradient editing-based adversarial generative model