Diffusion geometry is a framework for geometric and topological data analysis.

Diffusion geometry methods give new state-of-the-art results on data from manifolds.

We give a categorical perspective on the class of dynamical system models in machine learning.