NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:5819
Title:Private Testing of Distributions via Sample Permutations

This is a sound technical paper on the question of testing of distribution with privacy constraints, which shows how tests based on splitting techniques making use of a "flattening sample" (with limited l_2 norm of their corresponding distribution) can preserve privacy. The contribution is solid, timely (cf. privacy) and original.