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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

With the increasing attention to deep neural network (DNN) models, attacksare also upcoming for such models. For example, an attacker may carefullyconstruct images in specific ways (also referred to as adversarial examples)aiming to mislead the DNN models to output incorrect classification results.Similarly, many efforts are proposed to detect and mitigate adversarialexamples, usually for certain dedicated attacks. In this paper, we propose anovel digital watermark-based method to generate image adversarial examples tofool DNN models. Specifically, partial main features of the watermark image areembedded into the host image almost invisibly, aiming to tamper with and damagethe recognition capabilities of the DNN models. We devise an efficientmechanism to select host images and watermark images and utilize the improveddiscrete wavelet transform (DWT) based Patchwork watermarking algorithm with aset of valid hyperparameters to embed digital watermarks from the watermarkimage dataset into original images for generating image adversarial examples.The experimental results illustrate that the attack success rate on common DNNmodels can reach an average of 95.47% on the CIFAR-10 dataset and the highestat 98.71%. Besides, our scheme is able to generate a large number ofadversarial examples efficiently, concretely, an average of 1.17 seconds forcompleting the attacks on each image on the CIFAR-10 dataset. In addition, wedesign a baseline experiment using the watermark images generated by Gaussiannoise as the watermark image dataset that also displays the effectiveness ofour scheme. Similarly, we also propose the modified discrete cosine transform(DCT) based Patchwork watermarking algorithm. To ensure repeatability andreproducibility, the source code is available on GitHub.

Generating Image Adversarial Examples by Embedding Digital Watermarks