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

Artificial neural networks in general and deep learning networks inparticular established themselves as popular and powerful machine learningalgorithms. While the often tremendous sizes of these networks are beneficialwhen solving complex tasks, the tremendous number of parameters also causessuch networks to be vulnerable to malicious behavior such as adversarialperturbations. These perturbations can change a model's classificationdecision. Moreover, while single-step adversaries can easily be transferredfrom network to network, the transfer of more powerful multi-step adversarieshas - usually -- been rather difficult. In this work, we introduce a method forgenerating strong ad-versaries that can easily (and frequently) be transferredbetween different models. This method is then used to generate a large set ofadversaries, based on which the effects of selected defense methods areexperimentally assessed. At last, we introduce a novel, simple, yet effectiveapproach to enhance the resilience of neural networks against adversaries andbenchmark it against established defense methods. In contrast to the alreadyexisting methods, our proposed defense approach is much more efficient as itonly requires a single additional forward-pass to achieve comparableperformance results.

Enhancing Resilience of Deep Learning Networks by Means of Transferable Adversaries