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

Generative Adversarial Networks (GAN)-synthesized table publishing letspeople privately learn insights without access to the private table. However,existing studies on Membership Inference (MI) Attacks show promising results ondisclosing membership of training datasets of GAN-synthesized tables. Differentfrom those works focusing on discovering membership of a given data point, inthis paper, we propose a novel Membership Collision Attack against GANs(TableGAN-MCA), which allows an adversary given only synthetic entries randomlysampled from a black-box generator to recover partial GAN training data.Namely, a GAN-synthesized table immune to state-of-the-art MI attacks isvulnerable to the TableGAN-MCA. The success of TableGAN-MCA is boosted by anobservation that GAN-synthesized tables potentially collide with the trainingdata of the generator.  Our experimental evaluations on TableGAN-MCA have five main findings. First,TableGAN-MCA has a satisfying training data recovery rate on three commonlyused real-world datasets against four generative models. Second, factors,including the size of GAN training data, GAN training epochs and the number ofsynthetic samples available to the adversary, are positively correlated to thesuccess of TableGAN-MCA. Third, highly frequent data points have high risks ofbeing recovered by TableGAN-MCA. Fourth, some unique data are exposed tounexpected high recovery risks in TableGAN-MCA, which may attribute to GAN'sgeneralization. Fifth, as expected, differential privacy, without theconsideration of the correlations between features, does not show commendablemitigation effect against the TableGAN-MCA. Finally, we propose two mitigationmethods and show promising privacy and utility trade-offs when protectingagainst TableGAN-MCA.

TableGAN-MCA: Evaluating Membership Collisions of GAN-Synthesized Tabular Data Releasing