Maryam Aliakbarpour, Rose Silver, Thomas Steinke, Jonathan Ullman
TL;DR构建差分隐私估计量，可通过低样本复杂度估计在满足非常温和的矩条件下的任意分布的中位数。
Abstract
We construct differentially private estimators with low sample complexity that estimate the median of an arbitrary distribution over $\mathbb{R}$ satisfying very mild →