Privacy Aware Learning
This work addresses privacy concerns in machine learning for data-sensitive applications, providing foundational insights into the privacy-utility tradeoff.
The paper tackles the problem of statistical risk minimization under local privacy, where data is kept confidential from the learner, by establishing sharp upper and lower bounds on convergence rates. It reveals a precise tradeoff between privacy preservation and utility, measured by convergence rates, for statistical estimators.
We study statistical risk minimization problems under a privacy model in which the data is kept confidential even from the learner. In this local privacy framework, we establish 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, as measured by convergence rate, of any statistical estimator or learning procedure.