Cup and Cap Topological Neural Network
The Cup and Cap Topological Neural Network is formulated, a topological deep learning architecture designed to learn node-based variables (or 0-cochains) by taking into account their many-body interactions (e.g. triangles) present in the data.