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

We analyze the influence of adversarial training on the loss landscape ofmachine learning models. To this end, we first provide analytical studies ofthe properties of adversarial loss functions under different adversarialbudgets. We then demonstrate that the adversarial loss landscape is lessfavorable to optimization, due to increased curvature and more scatteredgradients. Our conclusions are validated by numerical analyses, which show thattraining under large adversarial budgets impede the escape from suboptimalrandom initialization, cause non-vanishing gradients and make the model findsharper minima. Based on these observations, we show that a periodicadversarial scheduling (PAS) strategy can effectively overcome thesechallenges, yielding better results than vanilla adversarial training whilebeing much less sensitive to the choice of learning rate.

On the Loss Landscape of Adversarial Training: Identifying Challenges and How to Overcome Them