README.md

February 5, 2025 ยท View on GitHub

Diffusion-based Graph-agnostic Clustering

Test environment

The test environment are recorded in environment.yml. We recommond install with conda from the environment file.

Datasets

# nodes# edges# attributes# clustersHomo. Ratio
Texas183325170350.108
Wisconsin251515170350.196
Cornell183298170350.305
Squirrel5201108536208950.223
Chameleon227718050232550.235
Flickr75752397381204790.239
Cora27085429143370.81
Citeseer33274732370360.739
Pubmed197174432450030.771
BlogCatalog5196171743818960.40
BAT13110388140.45
UAT11901359923940.698

We have include Texas, Cornell, Cora, Citeseer, BAT and UAT datasets under dataset/ directory for test. Other datasets can be loaded from torch_geometric.

Running code

To reproduce the results in the paper, please run the provided script.

bash run.sh