A low-cost variational-Bayes technique for merging mixtures of probabilistic principal component analyzers

摘要: Mixtures of probabilistic principal component analyzers (MPPCA) have shown effective for modeling high-dimensional data sets living on non-linear manifolds. Briefly stated, they conduct mixture model estimation and dimensionality reduction through a single process. This paper makes two contributions: first, we disclose Bayesian technique estimating such models. Then, assuming several MPPCA models are available, address the problem aggregating them into model, which should be as parsimonious possible. We in detail how this can achieved cost-effective way, without sampling nor access to data, but solely requiring parameters. The proposed approach is based novel variational-Bayes scheme operating over Numerous experimental results discussion provided.