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

Generative models have demonstrated revolutionary success in various visualcreation tasks, but in the meantime, they have been exposed to the threat ofleaking private information of their training data. Several membershipinference attacks (MIAs) have been proposed to exhibit the privacyvulnerability of generative models by classifying a query image as a trainingdataset member or nonmember. However, these attacks suffer from majorlimitations, such as requiring shadow models and white-box access, and eitherignoring or only focusing on the unique property of diffusion models, whichblock their generalization to multiple generative models. In contrast, wepropose the first generalized membership inference attack against a variety ofgenerative models such as generative adversarial networks, [variational]autoencoders, implicit functions, and the emerging diffusion models. Weleverage only generated distributions from target generators and auxiliarynon-member datasets, therefore regarding target generators as black boxes andagnostic to their architectures or application scenarios. Experiments validatethat all the generative models are vulnerable to our attack. For instance, ourwork achieves attack AUC $>0.99$ against DDPM, DDIM, and FastDPM trained onCIFAR-10 and CelebA. And the attack against VQGAN, LDM (for thetext-conditional generation), and LIIF achieves AUC $>0.90.$ As a result, weappeal to our community to be aware of such privacy leakage risks whendesigning and publishing generative models.

Generated Distributions Are All You Need for Membership Inference Attacks Against Generative Models