A minimax distortion view of differentially private query release

—We devise query-set independent mechanisms for the problem of differentially private query release. Specifically, a differentially private mechanism is constructed to publish a synthetic database, and “customized” companion estimators are then derived to provide the best possible answers. Accordingly, the distortion corresponding to the best mechanism at the worstcase query, named the minimax distortion, provides a fundamental characterization. For the general class of statistical queries, by deriving asymptotically sharp upper and lower bounds, we prove that the minimax distortion is O(1/n) as the database size n goes to infinity, with the squared-error distortion measure and fixed dimension of data entries.

• Type: Journal
• Year: 2015
• Where: ACSSC
• Authors: Weina Wang, Lei Ying, Junshan Zhang
• Keywords: ACSSC 2015