mishne-lab / ignr

Implementation of Implicit Graphon Neural Representation

IGNR

Implementation for our work: Implicit Graphon Neural Representation, Xia Mishne and Wang, AISTATS 2023

To run the single graphon learning task, go to folder IGNR, and run:
python train_IGNR.py --use-pg 1

To run the parameterized graphon learning task, go to folder cIGNR, and run:
python train_cIGNR.py --dataset '2ratio_rand'
python train_cIGNR.py --dataset 'gCircle'

Requirements:
pytorch >= 1.7
torch_geometric >= 2.1.0
networkx >=2.8.4
pot >=0.8.2
cython >= 0.29.32
joblib==0.15.1
einops >=0.6.0

Citation

@inproceedings{xia2023implicit,
  title={Implicit graphon neural representation},
  author={Xia, Xinyue and Mishne, Gal and Wang, Yusu},
  booktitle={International Conference on Artificial Intelligence and Statistics},
  pages={10619--10634},
  year={2023},
  organization={PMLR}
}