Awesome

DAGCN

Code of Beyond Homophily: Reconstructing Structure for Graph-agnostic Clustering in ICML 2023.

Environments

torch
numpy 
scipy
torch_geometric
scikit-learn
random
numba
tqdm
scanpy
networkx

Run

All core codes have been included, i.e., the construction of homophilic and heterophilic graphs, the mixed graph filtering, the backbone of DGCN. We apply the graph construction for clustering. In fact, the graphs can be applied on other tasks, like topologival graph in GNNs...

Citation

@misc{pan2023homophily,
      title={Beyond Homophily: Reconstructing Structure for Graph-agnostic Clustering}, 
      author={Erlin Pan and Zhao Kang},
      year={2023},
      eprint={2305.02931},
      archivePrefix={arXiv},
      primaryClass={cs.SI}
}