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 examples in NLP are receiving increasing research attention. Oneline of investigation is the generation of word-level adversarial examplesagainst fine-tuned Transformer models that preserve naturalness andgrammaticality. Previous work found that human- and machine-generatedadversarial examples are comparable in their naturalness and grammaticalcorrectness. Most notably, humans were able to generate adversarial examplesmuch more effortlessly than automated attacks. In this paper, we provide adetailed analysis of exactly how humans create these adversarial examples. Byexploring the behavioural patterns of human workers during the generationprocess, we identify statistically significant tendencies based on which wordshumans prefer to select for adversarial replacement (e.g., word frequencies,word saliencies, sentiment) as well as where and when words are replaced in aninput sequence. With our findings, we seek to inspire efforts that harnesshuman strategies for more robust NLP models.

Identifying Human Strategies for Generating Word-Level Adversarial Examples