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/

Zendy Plus

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

Zendy Tools

Zendy Open

The deep learning algorithm has achieved great success in the field ofcomputer vision, but some studies have pointed out that the deep learning modelis vulnerable to attacks adversarial examples and makes false decisions. Thischallenges the further development of deep learning, and urges researchers topay more attention to the relationship between adversarial examples attacks anddeep learning security. This work focuses on adversarial examples, optimizesthe generation of adversarial examples from the view of adversarial robustness,takes the perturbations added in adversarial examples as the optimizationparameter. We propose RWR-NM-PGD attack algorithm based on random warm restartmechanism and improved Nesterov momentum from the view of gradientoptimization. The algorithm introduces improved Nesterov momentum, using itscharacteristics of accelerating convergence and improving gradient updatedirection in optimization algorithm to accelerate the generation of adversarialexamples. In addition, the random warm restart mechanism is used foroptimization, and the projected gradient descent algorithm is used to limit therange of the generated perturbations in each warm restart, which can obtainbetter attack effect. Experiments on two public datasets show that thealgorithm proposed in this work can improve the success rate of attacking deeplearning models without extra time cost. Compared with the benchmark attackmethod, the algorithm proposed in this work can achieve better attack successrate for both normal training model and defense model. Our method has averageattack success rate of 46.3077%, which is 27.19% higher than I-FGSM and 9.27%higher than PGD. The attack results in 13 defense models show that the attackalgorithm proposed in this work is superior to the benchmark algorithm inattack universality and transferability.

Adversarial examples attack based on random warm restart mechanism and improved Nesterov momentum