Dynamic Persistent Homology for Brain Networks via Wasserstein Graph Clustering

01/01/2022

∙

by   Moo K. Chung, et al.

∙

0

∙

share

We present the novel Wasserstein graph clustering for dynamically changing graphs. The Wasserstein clustering penalizes the topological discrepancy between graphs. The Wasserstein clustering is shown to outperform the widely used k-means clustering. The method applied in more accurate determination of the state spaces of dynamically changing functional brain networks.