Asymptotic Consistency of Loss-Calibrated Variational Bayes
Machine Learning
2019-11-05 v1 Machine Learning
Abstract
This paper establishes the asymptotic consistency of the {\it loss-calibrated variational Bayes} (LCVB) method. LCVB was proposed in~\cite{LaSiGh2011} as a method for approximately computing Bayesian posteriors in a `loss aware' manner. This methodology is also highly relevant in general data-driven decision-making contexts. Here, we not only establish the asymptotic consistency of the calibrated approximate posterior, but also the asymptotic consistency of decision rules. We also establish the asymptotic consistency of decision rules obtained from a `naive' variational Bayesian procedure.
Keywords
Cite
@article{arxiv.1911.01288,
title = {Asymptotic Consistency of Loss-Calibrated Variational Bayes},
author = {Prateek Jaiswal and Harsha Honnappa and Vinayak A. Rao},
journal= {arXiv preprint arXiv:1911.01288},
year = {2019}
}