Pointwise spectral density estimation under local differential privacy

Description

Invited talk held at the Session "Time series 2"
Abstract: We propose a new interactive locally differentially private mechanism for estimating Höldersmooth spectral density functions of stationary Gaussian processes. Anonymization is achieved through two-stage truncation and subsequent Laplace perturbation. In particular, we show that our method
achieves a pointwise L2-rate with a dependency of only a^2 on the privacy parameter a. This rate stands in contrast to the results of (Kroll, 2024), who proposed a non-interactive mechanism for spectral density estimation and showed a dependency of a^4 on the privacy parameter for the uniform L2-rate.

Period	26 Jun 2024
Event title	International Symposium on Nonparametric Statistics 2024
Event type	Conference
Location	Braga, PortugalShow on map
Degree of Recognition	International