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

It is known that Deep Neural networks (DNNs) are vulnerable to adversarialattacks, and the adversarial robustness of DNNs could be improved by addingadversarial noises to training data (e.g., the standard adversarial training(SAT)). However, inappropriate noises added to training data may reduce amodel's performance, which is termed the trade-off between accuracy androbustness. This problem has been sufficiently studied for the classificationof whole images but has rarely been explored for image analysis tasks in themedical application domain, including image segmentation, landmark detection,and object detection tasks. In this study, we show that, for those medicalimage analysis tasks, the SAT method has a severe issue that limits itspractical use: it generates a fixed and unified level of noise for all trainingsamples for robust DNN training. A high noise level may lead to a largereduction in model performance and a low noise level may not be effective inimproving robustness. To resolve this issue, we design an adaptive-marginadversarial training (AMAT) method that generates sample-wise adaptiveadversarial noises for robust DNN training. In contrast to the existing,classification-oriented adversarial training methods, our AMAT method uses aloss-defined-margin strategy so that it can be applied to different tasks aslong as the loss functions are well-defined. We successfully apply our AMATmethod to state-of-the-art DNNs, using five publicly available datasets. Theexperimental results demonstrate that: (1) our AMAT method can be applied tothe three seemingly different tasks in the medical image application domain;(2) AMAT outperforms the SAT method in adversarial robustness; (3) AMAT has aminimal reduction in prediction accuracy on clean data, compared with the SATmethod; and (4) AMAT has almost the same training time cost as SAT.

Adaptive Adversarial Training to Improve Adversarial Robustness of DNNs for Medical Image Segmentation and Detection