analiza aiqbench
This paper is a theoretical treatment of membership inference, a privacy attack that asks whether a specific person's data was in a dataset, using only noisy statistics or a trained model. It simplifies the analysis for mean estimation over Gaussian and binary product distributions and surveys the literature. For AI readers, the practical takeaway is that privacy leakage can persist even when the released statistics are heavily noised, which bears on how differential privacy and other protections are evaluated for trained models.
A Simple Presentation of Robust Membership Inference Attacks
arXiv:2610.36458v1 Announce Type: cross Abstract: In a membership inference attack, an attacker is given the data of a particular individual, and tries to determine if that individual was included in a particular dataset, given some kind of statistics derived from the dataset or a model trained on the dataset. Roughly, a membership inference attack is robust if it succeeds even when the statistics are highly noisy, regardless of how that noise is introduced. We give a simple presentation of robust membership inference attacks for mean estimation, covering both multivariate Gaussian distributions and binary product distributions. We also give a summary of the literature on robust membership inference, covering history, extensions, and connections to other research areas.
Abstract:In a membership inference attack, an attacker is given the data of a particular individual, and tries to determine if that individual was included in a particular dataset, given some kind of statistics derived from the dataset or a model trained on the dataset. Roughly, a membership inference attack is robust if it succeeds even when the statistics are highly noisy, regardless of how that noise is introduced. We give a simple presentation of robust membership inference attacks for mean estimation, covering both multivariate Gaussian distributions and binary product distributions. We also give a summary of the literature on robust membership inference, covering history, extensions, and connections to other research areas.
Cryptography and Security (cs.CR); Information Theory (cs.IT); Machine Learning (cs.LG); Statistics Theory (math.ST)
arXiv:2610.36458 [cs.CR]
(or arXiv:2610.36458v1 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2610.36458 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 215 cuvinte
Subiecte conexe
Is My Vision-Language Data in Your AI? Membership Inference Test (MINT) Demo 2
arXiv · 16 iun. 2026
Membership Inference Attacks against Large Audio Language Models
arXiv · 17 iun. 2026
Quantum Estimation of Non-Hermitian Pseudospectra
arXiv · 30 sept. 2026
From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models
arXiv · 2 nov. 2026