Differential privacy is a definition of "privacy" for statistical databases. The definition is simple, yet it implies strong semantics even in the presence of an adversary with arbitrary auxiliary information about the database. In this talk, we discuss recent work on measuring the utility of differentially private analyses via the traditional yardsticks of statistical inference. Specifically, we discuss two differentially private estimators that, given i.i.d. samples from a probability distribution, converge to the correct answer at the same rate as the optimal nonprivate estimator. © 2009 Springer-Verlag.
CITATION STYLE
Smith, A. (2009). Asymptotically optimal and private statistical estimation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 5888 LNCS, pp. 53–57). https://doi.org/10.1007/978-3-642-10433-6_4
Mendeley helps you to discover research relevant for your work.