Mposterior: R package for Robust and Scalable Bayes via a Median
of Subset Posterior Measures
Description
Mposterior package provides a general framework for estimating a
median of subset posterior measures (M-posterior). Each subset posterior measure
is represented as a matrix of posterior samples; rows represent the atoms and
columns index the dimensions. All subset measures are represented as list of
matrices; length of the list equals the number of subsets (or machines). The
distance between subset posterior measures and M-posterior is measured using the RBF kernel.
M-posterior is represented as a weighted combination of empirical measures based on
subsets measures. These weights are estimated using Weiszfeld algorithm implemented
in this package.