Locally differentially private estimation of nonlinear functionals of discrete distributions

However, due to the privacy constraint the rates we attain are slower and similar to those obtained in the Gaussian model by Collier et al. [