Beyond Manifolds: Learning on Graphs and Trees with Geodesic Metric Spaces

Calissano, A., Pereira, L., Lueg, J., Miolane, N.

We present a Python package—built as a Geomstats plug-in—for analyzing datasets of graphs and trees as strongly non-Euclidean data in geodesic metric spaces, leveraging their connections to Euclidean and Riemannian structures, with a modular design, practical examples, and full documentation.