EcoSta 2021: Start Registration
View Submission - EcoSta2021
A0387
Title: Post-processed posteriors for structured covariances Authors:  Jaeyong Lee - Seoul National University (Korea, South) [presenting]
Abstract: Bayesian inference of structured covariance matrices are considered, and a post-processed posterior is proposed. The post-processing of the posterior consists of two steps. In the first step, posterior samples are obtained from the conjugate inverse-Wishart posterior, which does not satisfy any structural restrictions. In the second step, the posterior samples are transformed to satisfy the structural restriction through a post-processing function. The conceptually straightforward procedure of the post-processed posterior makes its computation efficient and can render interval estimators of any functional of covariance matrices. We also show that it has nearly optimal minimax rates for banded and bandable covariances among all possible pairs of priors and post-processing functions. The advantages of the post-processed posterior are demonstrated by a simulation study and a real data analysis.