Proposing principled Gaussian processes for modeling functions defined over the edge set of a simplicial 2-complex, enabling direct and independent learning for different Hodge components of edge functions.
Proposing Hodge-compositional edge Gaussian processes for modeling edge functions on networks.