JAX implementation of Variational Graph Auto-Encoders (Kipf and Welling, 2016).
Run either GAE (Graph Autoencoder) or VGAE (Variational GAE):
python3 train.py --is_vgae=True- Make a runnable notebook with loss plots, etc.
- Try on different datasets
- Compare ROC-AUC with results in Kipf-Welling paper
- Add documentation comments for encoder, decoder, etc.
Attribution: A good chunk of the code, especially those related to dataset preprocessing, is attributed to this amazing Colab tutorial by Lisa Wang and Nikola Jovanović.