Gaussian Process Quadrature Moment Transform

TL;DR

This paper introduces a Gaussian process quadrature-based method for moment transformation of Gaussian variables, explicitly accounting for numerical integration errors, and demonstrates improved performance over classical methods in engineering applications.

Contribution

It presents a novel moment transformation technique using Gaussian process quadrature that models integration error, enhancing accuracy in nonlinear transformations.

Findings

Outperforms classical moment transforms in experiments

Provides a measure of integration error via Bayesian quadrature

Improves nonlinear filtering accuracy

Abstract

Computation of moments of transformed random variables is a problem appearing in many engineering applications. The current methods for moment transformation are mostly based on the classical quadrature rules which cannot account for the approximation errors. Our aim is to design a method for moment transformation for Gaussian random variables which accounts for the error in the numerically computed mean. We employ an instance of Bayesian quadrature, called Gaussian process quadrature (GPQ), which allows us to treat the integral itself as a random variable, where the integral variance informs about the incurred integration error. Experiments on the coordinate transformation and nonlinear filtering examples show that the proposed GPQ moment transform performs better than the classical transforms.