kkdey/flashr: Factor Loading Adaptive SHrinkage in R

Fits low-rank matrix and tensor mean models using non-parametric unimodal priors on the components. For fitting a rank-1 matrix or tensor mean model, flashr will run a variational expectation-maximization (VEM) algorithm where the componenets and variances are assumed to be separable. For higher-ranks, a greedy algorithm is available where the VEM algorithm is iteratively run on the residuals of the previous iteration. A backfitting procedure is available for refined estimation.

install.packages("remotes")
remotes::install_github("kkdey/flashr")