基于边缘概率的合成数据存在隐私漏洞，可被高效恢复个体信息。
Privacy Vulnerabilities in Marginals-based Synthetic Data
- 提出新型成员推断攻击MAMA-MIA，利用算法类型信息加速破解
- 在3种主流差分隐私合成算法上验证，攻击效率提升数量级
- 揭示边缘概率类合成数据的隐蔽隐私风险，适合隐私安全研究者
作为隐私保护技术，合成数据生成（SDG）旨在保持与真实数据相似性的同时排除个人身份信息。许多SDG算法提供强大的差分隐私（DP）保障。然而，我们发现最强大的一类算法——即保留底层数据边际概率或类似统计量的算法——会泄露可被更高效恢复的个体信息。本文提出新型成员推断攻击MAMA-MIA，针对三种经典DP SDG算法（MST、PrivBayes、Private-GSD）进行评估。MAMA-MIA利用已知的SDG算法类型，能比现有领先攻击更准确、快数个数量级地获取隐藏数据信息。该方法最终赢得首届SNAKE（SaNitization Algorithm under attacK ... ε）竞赛。
原文摘要 · Abstract (English)
When acting as a privacy-enhancing technology, synthetic data generation (SDG) aims to maintain a resemblance to the real data while excluding personally-identifiable information. Many SDG algorithms provide robust differential privacy (DP) guarantees to this end. However, we show that the strongest class of SDG algorithms--those that preserve \textit{marginal probabilities}, or similar statistics, from the underlying data--leak information about individuals that can be recovered more efficiently than previously understood. We demonstrate this by presenting a novel membership inference attack, MAMA-MIA, and evaluate it against three seminal DP SDG algorithms: MST, PrivBayes, and Private-GSD. MAMA-MIA leverages knowledge of which SDG algorithm was used, allowing it to learn information about the hidden data more accurately, and orders-of-magnitude faster, than other leading attacks. We use MAMA-MIA to lend insight into existing SDG vulnerabilities. Our approach went on to win the first SNAKE (SaNitization Algorithm under attacK ... $\varepsilon$) competition.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。