lianqing11/cluster-gcn

Comp 5331 course project.

---

Cluster-GCN: An Efficient Algorithm for Training Deep and Large Graph Convolutional Networks Dependencies

• Python 3.7+(for string formatting features)
• PyTorch 1.1.0+
• metis
• sklearn

• install clustering toolkit: metis and its Python interface.

  download and install metis: http://glaros.dtc.umn.edu/gkhome/metis/metis/download

  METIS - Serial Graph Partitioning and Fill-reducing Matrix Ordering (official website)

1) Download metis-5.1.0.tar.gz from http://glaros.dtc.umn.edu/gkhome/metis/metis/download and unpack it
2) cd metis-5.1.0
3) make config shared=1 prefix=~/.local/
4) make install
5) export METIS_DLL=~/.local/lib/libmetis.so
6) `pip install metis`

quick test to see whether you install metis correctly:

>>> import networkx as nx
>>> import metis
>>> G = metis.example_networkx()
>>> (edgecuts, parts) = metis.part_graph(G, 3)

Run experiments

For ppi dataset with gcn, mean, pooling aggregator:

sh auto_gcn_mean_pool_run_ppi.sh

For ppi dataset with lstm aggregator:

sh auto_lstm_run_ppi.sh

For ppi dataset with attention aggregator:

sh auto_attn_run_ppi.sh 

---

For reddit dataset with gcn, mean, pooling aggregator:

sh auto_gcn_mean_pool_run_reddit.sh

For reddit dataset with attention aggregator:

sh auto_attn_run_reddit.sh 

You can get tune the hyper-param in the script and get the result from the "log" folder.

Acknowledgements

This code is heavily borrowed from DGL's implementation