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

In the field of natural language processing, the prevalent approach involvesfine-tuning pretrained language models (PLMs) using local samples. Recentresearch has exposed the susceptibility of PLMs to backdoor attacks, whereinthe adversaries can embed malicious prediction behaviors by manipulating a fewtraining samples. In this study, our objective is to develop abackdoor-resistant tuning procedure that yields a backdoor-free model, nomatter whether the fine-tuning dataset contains poisoned samples. To this end,we propose and integrate a honeypot module into the original PLM, specificallydesigned to absorb backdoor information exclusively. Our design is motivated bythe observation that lower-layer representations in PLMs carry sufficientbackdoor features while carrying minimal information about the original tasks.Consequently, we can impose penalties on the information acquired by thehoneypot module to inhibit backdoor creation during the fine-tuning process ofthe stem network. Comprehensive experiments conducted on benchmark datasetssubstantiate the effectiveness and robustness of our defensive strategy.Notably, these results indicate a substantial reduction in the attack successrate ranging from 10\% to 40\% when compared to prior state-of-the-art methods.

Setting the Trap: Capturing and Defeating Backdoors in Pretrained Language Models through Honeypots