Graph clustering

Deep Learning based Graph embedding for nodes clustering Project for the Machine Learning in Network Science class at CentraleSupélec.

Structure of the repository

Project report

The project report is in the file project_report.pdf.

Non-deep algorithms

Non-deep algorithms including:

• KMeans
• Spectral Clustering
• RMSC

are in the file clustering.py, with rather clean names. Can be launched with Python CLI to get the visualization results (can be a bit long).

VAEs

VAE and IWAE are in the file vae_clustering.py. Can be launched with Python CLI to get the visualization results (can be a bit long).

Visualization functions

In the file visualization.py.

DAEGC

This algorithm is seperated from the rest. You should go to the daegc directory:

cd daegc/

Then, to launch this algorithm, you should first pretrain the auto-encoder. For this you should run this command from the daegc folder:

python pretrain.py

To train the other algorithm, you can launch:

python training.py