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/

Zendy Plus

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

Zendy Tools

Zendy Open

The adversarial attacks against deep neural networks on computer vision taskshave spawned many new technologies that help protect models from avoiding falsepredictions. Recently, word-level adversarial attacks on deep models of NaturalLanguage Processing (NLP) tasks have also demonstrated strong power, e.g.,fooling a sentiment classification neural network to make wrong decisions.Unfortunately, few previous literatures have discussed the defense of suchword-level synonym substitution based attacks since they are hard to beperceived and detected. In this paper, we shed light on this problem andpropose a novel defense framework called Random Substitution Encoding (RSE),which introduces a random substitution encoder into the training process oforiginal neural networks. Extensive experiments on text classification tasksdemonstrate the effectiveness of our framework on defense of word-leveladversarial attacks, under various base and attack models.

Defense of Word-level Adversarial Attacks via Random Substitution Encoding