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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Adversarial attacks on deep neural networks (DNNs) have been found forseveral years. However, the existing adversarial attacks have high successrates only when the information of the victim DNN is well-known or could beestimated by the structure similarity or massive queries. In this paper, wepropose to Attack on Attention (AoA), a semantic property commonly shared byDNNs. AoA enjoys a significant increase in transferability when the traditionalcross entropy loss is replaced with the attention loss. Since AoA alters theloss function only, it could be easily combined with othertransferability-enhancement techniques and then achieve SOTA performance. Weapply AoA to generate 50000 adversarial samples from ImageNet validation set todefeat many neural networks, and thus name the dataset as DAmageNet. 13well-trained DNNs are tested on DAmageNet, and all of them have an error rateover 85%. Even with defenses or adversarial training, most models stillmaintain an error rate over 70% on DAmageNet. DAmageNet is the first universaladversarial dataset. It could be downloaded freely and serve as a benchmark forrobustness testing and adversarial training.

Universal Adversarial Attack on Attention and the Resulting Dataset DAmageNet