Invariant Priors for Bayesian Quadrature
Machine Learning
2021-12-06 v1 Machine Learning
Abstract
Bayesian quadrature (BQ) is a model-based numerical integration method that is able to increase sample efficiency by encoding and leveraging known structure of the integration task at hand. In this paper, we explore priors that encode invariance of the integrand under a set of bijective transformations in the input domain, in particular some unitary transformations, such as rotations, axis-flips, or point symmetries. We show initial results on superior performance in comparison to standard Bayesian quadrature on several synthetic and one real world application.
Keywords
Cite
@article{arxiv.2112.01578,
title = {Invariant Priors for Bayesian Quadrature},
author = {Masha Naslidnyk and Javier Gonzalez and Maren Mahsereci},
journal= {arXiv preprint arXiv:2112.01578},
year = {2021}
}