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

Despite that deep neural networks (DNNs) have achieved enormous success inmany domains like natural language processing (NLP), they have also been provento be vulnerable to maliciously generated adversarial examples. Such inherentvulnerability has threatened various real-world deployed DNNs-basedapplications. To strength the model robustness, several countermeasures havebeen proposed in the English NLP domain and obtained satisfactory performance.However, due to the unique language properties of Chinese, it is not trivial toextend existing defenses to the Chinese domain. Therefore, we propose AdvGraph,a novel defense which enhances the robustness of Chinese-based NLP models byincorporating adversarial knowledge into the semantic representation of theinput. Extensive experiments on two real-world tasks show that AdvGraphexhibits better performance compared with previous work: (i) effective - itsignificantly strengthens the model robustness even under the adaptive attackssetting without negative impact on model performance over legitimate input;(ii) generic - its key component, i.e., the representation of connotativeadversarial knowledge is task-agnostic, which can be reused in anyChinese-based NLP models without retraining; and (iii) efficient - it is alight-weight defense with sub-linear computational complexity, which canguarantee the efficiency required in practical scenarios.

Enhancing Model Robustness By Incorporating Adversarial Knowledge Into Semantic Representation