Dense message passing for sparse principal component analysis

We describe a novel inference algorithm for sparse Bayesian PCA with a zero-norm prior on the model parameters. Bayesian inference is very challenging in probabilistic models of this type. MCMC proced