Privacy-preserving Class Ratio Estimation

In this paper we present learning models for the class ratio estimation problem, which takes as input an unlabeled set of instances and predicts the proportions of instances in the set belonging to th