Differential privacy (DP) quantifies privacy loss by analyzing noise injected into output statistics. For non-trivial statistics, this noise is necessary to ensure finite privacy loss. However, data curators frequently release collections of statistics where some use DP mechanisms and others are released as-is, i.e., without additional randomized noise. Consequently, DP alone cannot characterize the privacy loss attributable to the entire collection of releases. In this paper, we present a privacy formalism, $(\epsilon, \{ \Theta_z\}_{z \in \mathcal{Z}})$-Pufferfish ($\epsilon$-TP for short when $\{ \Theta_z\}_{z \in \mathcal{Z}}$ is implied), a collection of Pufferfish mechanisms indexed by realizations of a random variable $Z$ representing public information not protected with DP noise. First, we prove that this definition has similar properties to DP. Next, we introduce mechanisms for releasing partially private data (PPD) satisfying $\epsilon$-TP and prove their desirable properties. We provide algorithms for sampling from the posterior of a parameter given PPD. We then compare this inference approach to the alternative where noisy statistics are deterministically combined with Z. We derive mild conditions under which using our algorithms offers both theoretical and computational improvements over this more common approach. Finally, we demonstrate all the effects above on a case study on COVID-19 data.
翻译:不同隐私(DP) 通过分析输入产出统计的噪音来量化隐私损失。 对于非三维统计来说,这种噪音对于确保有限隐私损失是必要的。 但是, 数据管理员经常发布统计数据收集, 某些使用DP机制和其他机制的人会自动发布, 也就是说, 没有额外的随机噪音。 因此, 光是DP 无法描述由于整个发布收集而导致的隐私损失。 在本文中, 我们呈现一种隐私形式主义, $ (\ epsilon, \\ Theta_ z ⁇ z ⁇ z \ in\ mathcal ⁇ ) $- Pufferferfish ($\ epslon$- $- TP) 是用来确保有限隐私损失的。 然而, 当 $\\\ Theta_ z ⁇ z z \ \ \ \ \ mathcalcalçal $ Qalphia 时, 数据会短短短时, 数据( $\ $19- sqolioalal comalalal scrial comgraphal grational grational grationalationalationalationalationalation) viewation view viewations view viewationalational viewations viewations viewations as we we we we we weweat view view view viewational view viewmational view) view view) viewmational view view viewmational subal view view viewmental subal subal subal subal subal subal subal subaldal vical vical subal subal vicaldaldaldaldal vical vical vicaldal subaldaldal vicaldaldaldaldaldalticaldaldaldaldal vical subal subal subal sub