1: \begin{abstract}

2:     Lipschitz learning is a graph-based semi-supervised learning method where one extends labels from a labeled to an unlabeled data set by solving the infinity Laplace equation on a weighted graph.

3:     In this work we prove uniform convergence rates for solutions of the graph infinity Laplace equation as the number of vertices grows to infinity. 

4:     Their continuum limits are absolutely minimizing Lipschitz extensions with respect to the geodesic metric of the domain where the graph vertices are sampled from.

5:     We work under very general assumptions on the graph weights, the set of labeled vertices, and the continuum domain.

6:     Our main contribution is that we obtain quantitative convergence rates even for very sparsely connected graphs, as they typically appear in applications like semi-supervised learning. In particular, our framework allows for graph bandwidths down to the connectivity radius. 

7:     For proving this we first show a quantitative convergence statement for graph distance functions to geodesic distance functions in the continuum. 

8:     Using the ``comparison with distance functions'' principle, we can pass these convergence statements to infinity harmonic functions and absolutely minimizing Lipschitz extensions.

9:     \\

10:     \\

11:     \textbf{Key words:} Lipschitz learning, graph-based semi-supervised learning, continuum limit, ab\-so\-lute\-ly min\-i\-miz\-ing Lipschitz extensions, infinity Laplacian

12:     \\

13:     \textbf{AMS subject classifications:} 35J20, 35R02, 65N12, 68T05

14: \end{abstract}

15: