graph-filters

This paper proposes a graph dictionary learning framework that represents graphs as Gaussian distributions using filtered Laplacians and optimal transport distances, achieving competitive performance in graph clustering and classification tasks.