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

Backdoor attacks have been shown to be a serious security threat against deeplearning models, and detecting whether a given model has been backdooredbecomes a crucial task. Existing defenses are mainly built upon the observationthat the backdoor trigger is usually of small size or affects the activation ofonly a few neurons. However, the above observations are violated in many casesespecially for advanced backdoor attacks, hindering the performance andapplicability of the existing defenses. In this paper, we propose a backdoordefense DTInspector built upon a new observation. That is, an effectivebackdoor attack usually requires high prediction confidence on the poisonedtraining samples, so as to ensure that the trained model exhibits the targetedbehavior with a high probability. Based on this observation, DTInspector firstlearns a patch that could change the predictions of most high-confidence data,and then decides the existence of backdoor by checking the ratio of predictionchanges after applying the learned patch on the low-confidence data. Extensiveevaluations on five backdoor attacks, four datasets, and three advancedattacking types demonstrate the effectiveness of the proposed defense.

Confidence Matters: Inspecting Backdoors in Deep Neural Networks via Distribution Transfer