Maryam Aliakbarpour - Private Testing of Distributions via Sample Permutations | MLxMIT June 23 2020 скачать в хорошем качестве

Maryam Aliakbarpour - Private Testing of Distributions via Sample Permutations | MLxMIT June 23 2020

Collaborators: Ilias Diakonikolas, Daniel Kane, Ronitt Rubinfeld Paper: https://papers.nips.cc/paper/9270-pri... Abstract: Statistical tests are at the heart of many scientific tasks. To validate their hypotheses, researchers in medical and social sciences use individuals’ data. The sensitivity of participants’ data requires the design of statistical tests that ensure the privacy of the individuals in the most efficient way. Our goal is to enable testing algorithms to make global inferences about a collective dataset while protecting the privacy of individuals. In this talk, I present several algorithms in the framework of property testing to test the properties of a distribution with respect to differential privacy. In particular, we study the following fundamental problems: (1) Is the distribution uniform? (2) Is it equal to another distribution that we have access to? (3) Are two random variables independent from each other? In all of these cases, we show that our testers achieve near-optimal sample complexity (up to logarithmic factors). Moreover, our dependence on the privacy parameter is an additive term, which indicates that differential privacy can be obtained in most regimes of parameters for free.