Private Algorithms Can be Always Extended
arXiv: Statistics Theory, 2018.
Abstract:
We consider the following fundamental question on $e$-differential privacy. Consider an arbitrary $e$-differentially private algorithm defined on a subset of the input space. Is it possible to extend it to an $eu0027$-differentially private algorithm on the whole input space for some $eu0027$ comparable with $e$? In this note we answer af...More
Code:
Data:
Tags
Comments