EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Auditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning

Mohammad Mansour Al-Khasawneh, Mamoon Obidat, Ayoub Alsarhan, Tamam Alsarhan, Malek Barhoush, Osama Harfoushi, Sami Aziz Alshammari

Future Internet · 16.11.2026 · 18 (11) · 632

DOI: 10.3390/fi18110632 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Future Internet, online erschienen am 16.11.2026. Verfasst von Mohammad Mansour Al-Khasawneh und 6 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Privacy-sensitive machine learning programmes need deployable models without repeatedly circulating original records among developers, analysts and downstream services. This paper evaluates a class-conditional variational autoencoder (CVAE) placed as a synthetic-data firewall between a protected corpus and downstream learners that never see the original records, and contributes an audit and release protocol for such a pipeline. The protocol separates four adversary tiers (model, release, sampling and white-box generator access), calibrates membership attacks with 32 shadow pipelines, adds deliberately vulnerable positive controls, tests copying in the lower tail of the distance distribution, and gates the release on utility, exposure, copying and worst-class retention with an exposure threshold anchored to chance. On the digit corpus, over ten independent partitions, synthetic-only training retains % (CNN) and % (GRU) of real-data accuracy. On digits and a clinical table, likelihood-ratio attacks detect leakage in real-data models but not in release-trained models, and memorising learners are protected on all three corpora (including a face-image set). The firewall relocates rather than removes risk: adversaries holding the release or the generator retain a measurable membership signal, and with very small cohorts, the generator passes it through. Conditional GAN, Gaussian-mixture and differentially private baselines, alternative likelihoods, and an independent selection protocol (% for the selected configuration) complete the evaluation; empirical resistance is not a formal (ε,δ) guarantee.

Zitieren

APA

Al-Khasawneh, M. M., Obidat, M., Alsarhan, A., Alsarhan, T., Barhoush, M., Harfoushi, O., & Alshammari, S. A. (2026). Auditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning. Future Internet, 18(11), 632. https://doi.org/10.3390/fi18110632

Vancouver

Al-Khasawneh MM, Obidat M, Alsarhan A, Alsarhan T, Barhoush M, Harfoushi O, et al. Auditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning. Future Internet. 2026;18(11):632. doi:10.3390/fi18110632

BibTeX
@article{AlKhasawneh2026,
  author = {Al-Khasawneh, Mohammad Mansour and Obidat, Mamoon and Alsarhan, Ayoub and Alsarhan, Tamam and Barhoush, Malek and Harfoushi, Osama and Alshammari, Sami Aziz},
  title = {Auditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning},
  journal = {Future Internet},
  year = {2026},
  volume = {18},
  number = {11},
  pages = {632},
  doi = {10.3390/fi18110632}
}
Kostenlos starten →