Persistence Homology for Link Prediction

Code for paper Link Prediction with Persistent Homology: An Interactive View (ICML2021). In this work, we propose a novel topological approach based on the extended persistent homology to characterize interactions between two nodes. We propose a graph neural network method combining with this topological feature and it outperforms state-of-the-arts on different benchmarks. As another contribution, we propose a novel algorithm to more efficiently compute the extended persistence diagrams for graphs. This algorithm can be generally applied to accelerate many other topological methods for graph learning tasks.

Zuoyu Yan, Tengfei Ma