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

At present, backdoor attacks attract attention as they do great harm to deeplearning models. The adversary poisons the training data making the model beinginjected with a backdoor after being trained unconsciously by victims using thepoisoned dataset. In the field of text, however, existing works do not providesufficient defense against backdoor attacks. In this paper, we propose aNoise-augmented Contrastive Learning (NCL) framework to defend against textualbackdoor attacks when training models with untrustworthy data. With the aim ofmitigating the mapping between triggers and the target label, we addappropriate noise perturbing possible backdoor triggers, augment the trainingdataset, and then pull homology samples in the feature space utilizingcontrastive learning objective. Experiments demonstrate the effectiveness ofour method in defending three types of textual backdoor attacks, outperformingthe prior works.

NCL: Textual Backdoor Defense Using Noise-augmented Contrastive Learning