Variational bridge constructs for approximate Gaussian process regression

TL;DR

This paper presents a novel variational inference method that approximates Gaussian process regression through stochastic differential equations, enabling scalable and flexible modeling with results comparable to full GP regression.

Contribution

The paper introduces a new variational bridge approach for Gaussian process regression using stochastic differential equations, extending applicability to non-linear dynamics.

Findings

Generated paths are indistinguishable from GP samples

Approach extends easily to non-linear dynamics

Comparable accuracy to full GP regression

Abstract

This paper introduces a method to approximate Gaussian process regression by representing the problem as a stochastic differential equation and using variational inference to approximate solutions. The approximations are compared with full GP regression and generated paths are demonstrated to be indistinguishable from GP samples. We show that the approach extends easily to non-linear dynamics and discuss extensions to which the approach can be easily applied.