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

Achieving robustness against adversarial input perturbation is an importantand intriguing problem in machine learning. In the area of semantic imagesegmentation, a number of adversarial training approaches have been proposed asa defense against adversarial perturbation, but the methodology of evaluatingthe robustness of the models is still lacking, compared to imageclassification. Here, we demonstrate that, just like in image classification,it is important to evaluate the models over several different and hard attacks.We propose a set of gradient based iterative attacks and show that it isessential to perform a large number of iterations. We include attacks againstthe internal representations of the models as well. We apply two types ofattacks: maximizing the error with a bounded perturbation, and minimizing theperturbation for a given level of error. Using this set of attacks, we show forthe first time that a number of models in previous work that are claimed to berobust are in fact not robust at all. We then evaluate simple adversarialtraining algorithms that produce reasonably robust models even under our set ofstrong attacks. Our results indicate that a key design decision to achieve anyrobustness is to use only adversarial examples during training. However, thisintroduces a trade-off between robustness and accuracy.

On Evaluating the Adversarial Robustness of Semantic Segmentation Models