Unsupervised Learning of Graph Hierarchical Abstractions

Source codes are for the paper “Unsupervised Learning of Graph Hierarchical Abstractions with Differentiable Coarsening and Optimal Transport” (AAAI 2021). The codes are built on the libarary of Pytorch Geometric. This paper proposed a new unsupervised graph learning method based on optimal transport and graph coarsening.

Tengfei Ma