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

Generating adversarial examples for Neural Machine Translation (NMT) withsingle Round-Trip Translation (RTT) has achieved promising results by releasingthe meaning-preserving restriction. However, a potential pitfall for thisapproach is that we cannot decide whether the generated examples areadversarial to the target NMT model or the auxiliary backward one, as thereconstruction error through the RTT can be related to either. To remedy thisproblem, we propose a new criterion for NMT adversarial examples based on theDoubly Round-Trip Translation (DRTT). Specifically, apart from thesource-target-source RTT, we also consider the target-source-target one, whichis utilized to pick out the authentic adversarial examples for the target NMTmodel. Additionally, to enhance the robustness of the NMT model, we introducethe masked language models to construct bilingual adversarial pairs based onDRTT, which are used to train the NMT model directly. Extensive experiments onboth the clean and noisy test sets (including the artificial and natural noise)show that our approach substantially improves the robustness of NMT models.

Generating Authentic Adversarial Examples beyond Meaning-preserving with Doubly Round-trip Translation