Hippocluster: an efficient, brain-inspired adaptation of K-means for graph clustering.
The corresponding research paper by Chalmers et al is here
install using pip:
pip install git+https://github.com/echalmers/hippocluster.git
see demo.py for basic usage examples
and see demo.py for a visual illustration of Hippocluster working.
Hippocluster's fit and update methods accept a RandomWalkGraph object. There are two ways to make one of these from a NetworkX graph object. The first is to extend RandomWalkGraph, implementing communities and n_communities. Or if the graph has no ground truth communities, you can directly instantiate a RandomWalkGraph object with a NetworkX graph. For example:
myGraph = RandomWalkGraph(nx.grid_2d_graph(20, 20))
