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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Neural networks are susceptible to data inference attacks such as themembership inference attack, the adversarial model inversion attack and theattribute inference attack, where the attacker could infer useful informationsuch as the membership, the reconstruction or the sensitive attributes of adata sample from the confidence scores predicted by the target classifier. Inthis paper, we propose a method, namely PURIFIER, to defend against membershipinference attacks. It transforms the confidence score vectors predicted by thetarget classifier and makes purified confidence scores indistinguishable inindividual shape, statistical distribution and prediction label between membersand non-members. The experimental results show that PURIFIER helps defendmembership inference attacks with high effectiveness and efficiency,outperforming previous defense methods, and also incurs negligible utilityloss. Besides, our further experiments show that PURIFIER is also effective indefending adversarial model inversion attacks and attribute inference attacks.For example, the inversion error is raised about 4+ times on the Facescrub530classifier, and the attribute inference accuracy drops significantly whenPURIFIER is deployed in our experiment.

Purifier: Defending Data Inference Attacks via Transforming Confidence Scores