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

Adversarial attacks against machine learning models have threatened variousreal-world applications such as spam filtering and sentiment analysis. In thispaper, we propose a novel framework, learning to DIScriminate Perturbations(DISP), to identify and adjust malicious perturbations, thereby blockingadversarial attacks for text classification models. To identify adversarialattacks, a perturbation discriminator validates how likely a token in the textis perturbed and provides a set of potential perturbations. For each potentialperturbation, an embedding estimator learns to restore the embedding of theoriginal word based on the context and a replacement token is chosen based onapproximate kNN search. DISP can block adversarial attacks for any NLP modelwithout modifying the model structure or training procedure. Extensiveexperiments on two benchmark datasets demonstrate that DISP significantlyoutperforms baseline methods in blocking adversarial attacks for textclassification. In addition, in-depth analysis shows the robustness of DISPacross different situations.

Learning to Discriminate Perturbations for Blocking Adversarial Attacks in Text Classification