arxivst stuff from arxiv that you should probably bookmark

Adaptive Gaussian process approximation for Bayesian inference with expensive likelihood functions

Abstract · Mar 29, 2017 08:32 ·

stat-co stat-ml

Arxiv Abstract

  • Hongqiao Wang
  • Jinglai Li

We consider Bayesian inference problems with computationally intensive likelihood functions. We propose a Gaussian process (GP) based method to approximate the joint distribution of the unknown parameters and the data. In particular, we write the joint density approximately as a product of an approximate posterior density and an exponentiated GP surrogate. We then provide an adaptive algorithm to construct such an approximation, where an active learning method is used to choose the design points. With numerical examples, we illustrate that the proposed method has competitive performance against existing approaches for Bayesian computation.

Read the paper (pdf) »