A0354
Title: Wasserstein Gaussianization and efficient variational Bayes for robust Bayesian synthetic likelihood
Authors: Nhat Minh Nguyen - The University of Sydney (Australia) [presenting]
Minh-Ngoc Tran - University of Sydney (Australia)
David Nott - National University of Singapore (Singapore)
Christopher Drovandi - Queensland University of Technology (Australia)
Abstract: The efficiency of the Bayesian Synthetic Likelihood (BSL) method relies on the normality assumption of the summary statistics. A Wasserstein gradient flow is proposed to be used to approximately transform the distribution of the summary statistics into a Gaussian distribution. BSL also implicitly requires compatibility between simulated summary statistics under the working model and the observed summary statistics. This requirement has been facilitated by the robust BSL method developed recently in the literature. The Wasserstein Gaussianing transformation is combined with robust BSL, together with an efficient Variational Bayes procedure for posterior approximation, to develop a highly efficient and reliable approximate Bayesian inference method for likelihood-free problems.
