/VGNAE

(CIKM'21) Variational Graph Normalized Auto-Encoders

Primary LanguagePython

VGNAE

An implement of CIKM 2021 paper "Variational Graph Normalized Auto-Encoders" (CIKM 2021).

Variational Graph Normalized Auto-Encoders.
Seong Jin Ahn, Myoung Ho Kim.
CIKM '21: The 30th ACM International Conference on Information and Knowledge Management Proceedings.
Short paper

Thank you for your interest in our works!
You can access our paper in https://arxiv.org/abs/2108.08046

Motivation

We find out that GAEs make embeddings of isolated nodes (nodes with no-observed links) zero vectors regardless of their feature information.
Our works try to distinguish embeddings of isolated nodes by reflecting their feature information better. image
image

Dependencies

Recent versions of the following packages for Python 3 are required:

  • Anaconda3
  • Python 3.8.0
  • Pytorch 1.8.1
  • torch_geometric 1.7.0
  • torch_scatter 2.0.6

Easy Run

python main.py --dataset=Cora --training_rate=0.2 --epochs=300

Citing

If you make advantage of our VGNAE in your research, please cite the following in your manuscript:

@article{ahn2021variational,

title={Variational Graph Normalized Auto-Encoders},
author={Ahn, Seong Jin and Kim, Myoung Ho},
journal={arXiv preprint arXiv:2108.08046},
year={2021}
}