Home

Awesome

Hyperbolic Graph Convolutional Auto-Encoders

Accepted to CVPR2021 :tada:

Official PyTorch code of Unsupervised Hyperbolic Representation Learning via Message Passing Auto-Encoders

Jiwoong Park*, Junho Cho*, Hyung Jin Chang, Jin Young Choi <sub>(* indicates equal contribution)</sub>

vis_cora <sub>Embeddings of cora dataset. GAE is Graph Auto-Encoders in Euclidean space, HGCAE is our method. P is Poincare ball, H is Hyperboloid.</sub>

Overview

This repository provides HGCAE code in PyTorch for reproducibility with

Installation Guide

We use docker to reproduce performance. Please refer guide.md

Usage

1. Run docker

Before training, run our docker image:

docker run --gpus all -it --rm --shm-size 100G -v $PWD:/workspace junhocho/hyperbolicgraphnn:8 bash

If you want to cache edge splits for train/val dataset and load faster afterwards, mkdir ~/tmp and run:

docker run --gpus all -it --rm --shm-size 100G -v $PWD:/workspace -v ~/tmp:/root/tmp junhocho/hyperbolicgraphnn:8 bash

2. train_<dataset>.sh

In the docker session, run each train shell script for each dataset to reproduce performance:

Graph data link prediction

Run following commands to reproduce results:

ROCAP
Cora0.948907030.94726805
Citeseer0.960594070.96305937
Wiki0.955108050.96200790
Pubmed0.962072120.96083080
Blog Catalog0.896839390.88651569
Amazon Photo0.982406730.97655753

Graph data node clustering

ACCNMIARI
Cora0.746676510.572529400.55212928
Citeseer0.693116920.422492940.44101404
Wiki0.459459460.467778810.21517031
Pubmed0.748491150.377592620.40770875
Blog Catalog0.550615860.325573880.25227964
Amazon Photo0.781307190.696236510.60342107

Image clustering

ACCNMIARI
ImageNet100.855923080.790191310.74181220
ImageNetDog0.387384620.360596500.22696503

Parameter description

Acknowledgments

This repo is inspired by hgcn.

And some of the code was forked from the following repositories:

License

This work is licensed under the MIT License

Citation

@inproceedings{park2021unsupervised,
  title={Unsupervised Hyperbolic Representation Learning via Message Passing Auto-Encoders},
  author={Jiwoong Park and Junho Cho and Hyung Jin Chang and Jin Young Choi},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  year={2021}
}