1: \begin{abstract}

2:   \looseness -1

3:   Gaussian process regression is widely used because of its ability to provide well-calibrated uncertainty estimates and handle small or sparse datasets.

4:   However, it struggles with high-dimensional data.

5:   One possible way to scale this technique to higher dimensions is to leverage the implicit low-dimensional \emph{manifold} upon which the data actually lies, as postulated by the \emph{manifold hypothesis}.

6:   Prior work ordinarily requires the manifold structure to be explicitly provided though, i.e. given by a mesh or be known to be one of the well-known manifolds like the sphere.

7:   In contrast, in this paper we propose a Gaussian process regression technique capable of inferring implicit structure directly from data (labeled and unlabeled) in a fully differentiable way.

8:   For the resulting model, we discuss its convergence to the Matérn Gaussian process on the assumed manifold.

9:   Our technique scales up to hundreds of thousands of data points, and may improve the predictive performance and calibration of the standard Gaussian process regression in high-dimensional~settings.

10: \end{abstract}

11: