Coupled-Worlds Privacy: Exploiting Adversarial Uncertainty in Statistical Data Privacy

By Raef Bassily, Adam Groce, Jonathan Katz, Adam D. Smith. 2013

100 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Safety Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety Research

Citations in the corpus

Rényi Differential Privacy (2017, background): “Several other works, most notably the Pufferfish and the coupled-worlds frameworks [15], [16], propose different stability constraints on the output distribution of privacypreserving mechanisms.”
Privacy for the Protected (Only) (2015, background): “[1], or the worst-case collusion assumptions on collections of data analysts as in Kearns et al.”
Privacy-preserving Analysis of Correlated Data (2016, background): “Finally, coupled-worlds privacy (Bassily et al., 2013) is an elegant privacy framework alternative to Pufferfish, which is also capable of accounting for correlation in data.”
Differentially Private Confidence Intervals (2020, background): “Interested readers should see [11] or [2] for more discussion.)”
Privacy and Statistical Risk: Formalisms and Minimax Bounds (2014, background): “We believe that future work on alternate deﬁnitions of privacy, which consider weaker adversaries (see Bassily et al. [2]), should be fruitful.”
Vuvuzela: scalable private messaging resistant to traffic analysis (2015, unknown): “In a system with many users, it may be reasonable to assume that some fraction of users are honest (i.e., the adversary does not know what they are doing), formalized with the help of coupled-worlds privacy [3] or noiseless database privacy [5].”
Stadium: A Distributed Metadata-Private Messaging System (2017, background): “As suggested in [2, 5, 24], Stadium can benefit from using legitimate users’ traffic as noise.”
Cardinality Estimators do not Preserve Privacy (2018, background): “Data-generating probability distributions were considered as a source of uncertainty in [7, 11, 15, 28, 30, 37, 43]; and some deterministic algorithms, which do not add noise, have been shown to preserve privacy under this assumption [7, 11, 28].”
Towards Measuring Membership Privacy (2017, background): “Follow-up works on DP try to relax the background knowledge by building more realistic background knowledge models [6, 21, 23, 27, 41].”