Not All Attributes are Created Equal: d _X -Private Mechanisms for Linear Queries

Abstract

Differential privacy provides strong privacy guarantees simultaneously enabling useful insights from sensitive datasets. However, it provides the same level of protection for all elements (individuals and attributes) in the data. There are practical scenarios where some data attributes need more/less protection than others. In this paper, we consider dX -privacy, an instantiation of the privacy notion introduced in [6], which allows this flexibility by specifying a separate privacy budget for each pair of elements in the data domain. We describe a systematic procedure to tailor any existing differentially private mechanism that assumes a query set and a sensitivity vector as input into its dX -private variant, specifically focusing on linear queries. Our proposed meta procedure has broad applications as linear queries form the basis of a range of data analysis and machine learning algorithms, and the ability to define a more flexible privacy budget across the data domain results in improved privacy/utility tradeoff in these applications. We propose several dX -private mechanisms, and provide theoretical guarantees on the trade-off between utility and privacy. We also experimentally demonstrate the effectiveness of our procedure, by evaluating our proposed dX -private Laplace mechanism on both synthetic and real datasets using a set of randomly generated linear queries.

CAPE: Context-Aware Private Embeddings for Private Language Learning

“…This is an interesting counterfinding to Welch et al (2020) which found better embedding performance with age-and genderaware representations in a global population. Differing privacy requirements for separate attributes are a feature of multiple variations on differential privacy regimes (Kamalaruban et al, 2020;Alaggan et al, 2017;Jorgensen et al, 2015).…”

Locality Sensitive Hashing with Extended Differential Privacy

“…As explained in Sect. 1, there are a number of existing extended DP mechanisms [4,7,9,25,32,53] designed for other metrics (e.g., the Euclidean metric, the l 1 metric), which cannot be applied to the angular distance. To our knowledge, our mechanisms are the first to provide extended DP with the angular distance.…”

“…In addition, most of the studies on extended DP have studied low-dimensional data such as two-dimensional [4,7,9,32] and six-dimensional [53] data. One exception is the work in [25], which proposed the multivariate Laplace mechanism for 300-dimensional vectors.…”

“…However, the range of actual applications is limited by the particular metrics for which extended DP mechanisms have been designed. For example, the existing works on locations [4,7,9], documents [25], range queries [53], and linear queries [32] are designed for the Euclidean metric, the Earth Mover's metric, the l 1 metric, and the summation of privacy budgets for attributes, respectively. However, there have been no known extended DP mechanisms designed for the angular distance (or cosine distance).…”