Part of Advances in Neural Information Processing Systems 25 (NIPS 2012)
Martin J. Wainwright, Michael Jordan, John C. Duchi
We study statistical risk minimization problems under a version of privacy in which the data is kept confidential even from the learner. In this local privacy framework, we show sharp upper and lower bounds on the convergence rates of statistical estimation procedures. As a consequence, we exhibit a precise tradeoff between the amount of privacy the data preserves and the utility, measured by convergence rate, of any statistical estimator.