This repository contains the python implementation for the paper "Perfect Recovery for Random Geometric Graph Matching with Shallow Graph Neural Networks" published in AISTATS 2025.

pip install -r requirements.txt

The code for generating the results in the paper is in two IPython notebooks:

• gnn_sim.ipynb contains code for synthetic experiments;
• gnn_real.ipynb contains code for real data experiments.

@inproceedings{liu2025perfect,
  title={Perfect Recovery for Random Geometric Graph Matching with Shallow Graph Neural Networks},
  author={Liu, Suqi and Austern, Morgane},
  booktitle={International Conference on Artificial Intelligence and Statistics},
  year={2025},
  organization={PMLR}
}