An Introduction to Gaussian Process Models

TL;DR

This paper introduces Gaussian process regression, highlighting its advantages for modeling dynamical systems with uncertainty quantification, and provides an accessible overview of its principles and applications.

Contribution

It offers an accessible introduction to Gaussian process models and their application in regression for dynamical systems, emphasizing their benefits over other methods.

Findings

Gaussian processes provide both mean predictions and uncertainty measures.

They are effective for nonlinear regression without extensive prior knowledge.

The paper demonstrates practical usage through an online tool.

Abstract

Within the past two decades, Gaussian process regression has been increasingly used for modeling dynamical systems due to some beneficial properties such as the bias variance trade-off and the strong connection to Bayesian mathematics. As data-driven method, a Gaussian process is a powerful tool for nonlinear function regression without the need of much prior knowledge. In contrast to most of the other techniques, Gaussian Process modeling provides not only a mean prediction but also a measure for the model fidelity. In this article, we give an introduction to Gaussian processes and its usage in regression tasks of dynamical systems. Try Gaussian process regression yourself: https://gpr.tbeckers.com