Title:Model Inference with Stein Density Ratio Estimation

Abstract:The Kullback-Leilber divergence from model to data is a classic goodness of fit measure but can be intractable in many cases. In this paper, we estimate the ratio function between a data density and a model density with the help of Stein operator. The estimated density ratio allows us to compute the likelihood ratio function which is a surrogate to the actual Kullback-Leibler divergence from model to data. By minimizing this surrogate, we can perform model fitting and inference from either frequentist or Bayesian point of view. This paper discusses methods, theories and algorithms for performing such tasks. Our theoretical claims are verified by experiments and examples are given demonstrating the usefulness of our methods.