Interpretable Clustering

This repository contains the code for Interpretable Clustering on Dynamic Graphs with Recurrent Graph Neural Networks, AAAI 2021.

Data

6 datasets were used in the paper:

• Simulate
• DBLP-E
• DBLP-3
• DBLP-5
• Brain
• Reddit

Update: Simulate and DBLP-E datasets work well. DBLP-3, DBLP-5, Brain, and Reddit datasets are from another paper and hard to verify the meaning of node features. Will replace it to new datasets soon.

Requirements

• Python 3,
• PyTorch 1.2.0 or higher,
• TensorFlow==1.11.0,
• dgl-cu101,
• dynamicgem,
• networkx,
• sklearn,
• scipy,
• keras==2.2.4,
• numpy>=1.15.3

Main file

• main.ipynb