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/

Zendy Plus

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

Zendy Tools

Zendy Open

Membership inference (MI) attacks affect user privacy by inferring whethergiven data samples have been used to train a target learning model, e.g., adeep neural network. There are two types of MI attacks in the literature, i.e.,these with and without shadow models. The success of the former heavily dependson the quality of the shadow model, i.e., the transferability between theshadow and the target; the latter, given only blackbox probing access to thetarget model, cannot make an effective inference of unknowns, compared with MIattacks using shadow models, due to the insufficient number of qualifiedsamples labeled with ground truth membership information.  In this paper, we propose an MI attack, called BlindMI, which probes thetarget model and extracts membership semantics via a novel approach, calleddifferential comparison. The high-level idea is that BlindMI first generates adataset with nonmembers via transforming existing samples into new samples, andthen differentially moves samples from a target dataset to the generated,non-member set in an iterative manner. If the differential move of a sampleincreases the set distance, BlindMI considers the sample as non-member and viceversa.  BlindMI was evaluated by comparing it with state-of-the-art MI attackalgorithms. Our evaluation shows that BlindMI improves F1-score by nearly 20%when compared to state-of-the-art on some datasets, such as Purchase-50 andBirds-200, in the blind setting where the adversary does not know the targetmodel's architecture and the target dataset's ground truth labels. We also showthat BlindMI can defeat state-of-the-art defenses.

Practical Blind Membership Inference Attack via Differential Comparisons