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

Deep Neural Networks (DNNs) have been shown to be susceptible to Trojanattacks. Neural Trojan is a type of targeted poisoning attack that embeds thebackdoor into the victim and is activated by the trigger in the input space.The increasing deployment of DNNs in critical systems and the surge ofoutsourcing DNN training (which makes Trojan attack easier) makes the detectionof Trojan attacks necessary. While Neural Trojan detection has been studied inthe image domain, there is a lack of solutions in the NLP domain. In thispaper, we propose a model-level Trojan detection framework by analyzing thedeviation of the model output when we introduce a specially craftedperturbation to the input. Particularly, we extract the model's responses toperturbed inputs as the `signature' of the model and train a meta-classifier todetermine if a model is Trojaned based on its signature. We demonstrate theeffectiveness of our proposed method on both a dataset of NLP models we createand a public dataset of Trojaned NLP models from TrojAI. Furthermore, wepropose a lightweight variant of our detection method that reduces thedetection time while preserving the detection rates.

PerD: Perturbation Sensitivity-based Neural Trojan Detection Framework on NLP Applications