Title: Scalable computation with shrinkage priors
Authors: Anirban Bhattacharya - Texas AM University (United States) [presenting]
Abstract: The aim is to discuss some recent developments in scaling Markov chain Monte Carlo methods to Big (n, p) regression models, i.e., where the number of subjects and the dimensionality can both be large. We present an exact algorithm whose computational complexity scales linearly in the dimensionality and a randomized extension to deal with large n problems. Several applications to high-dimensional linear and logistic regression, dictionary learning, reduced-rank regression are illustrated.