Polya-gamma augmentations for factor models.

Arto Klami

Abstract: Bayesian inference for latent factor models, such as principal component and canonical correlation analysis, is easy for Gaussian likelihoods. In particular, full conjugacy makes both Gibbs sampler...