B1538
Title: Bayesian inference for functional parameters
Authors: David Stephens - McGill University (Canada) [presenting]
Vivian Meng - McGill University (Canada)
Abstract: Bayesian semiparametric inference targets finite-dimensional parameters that are functionals of the (unknown) data-generating distribution, which, to avoid inconsistency due to misspecification, should be modelled in a non-parametric fashion. We will introduce a general framework for performing inference in this setting that ensures compatibility between the inherent finite and infinite-dimensional specifications that comprise the inference. We will illustrate the methodology with examples from causal inference, including regression-based adjustment and inverse probability weighting.
