Skip to content
Pytorch implementation of Graph U-Nets (ICML19)
Branch: master
Clone or download
Latest commit 74c8ec1 May 29, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data remove 12k May 27, 2019
pytorch_structure2vec-master init project May 27, 2019
.gitignore add gitignore May 27, 2019
LICENSE init project May 27, 2019
README.md Update README.md May 27, 2019
main.py init project May 27, 2019
mlp_dropout.py init project May 27, 2019
network.py init project May 27, 2019
ops.py
pytorch_structure2vec-master.zip init project May 27, 2019
run_DGCNN.sh
util.py init project May 27, 2019

README.md

PyTorch Graph U-Nets

Created by Hongyang Gao, and Shuiwang Ji at Texas A&M University.

About

PyTorch implementation of Graph U-Nets. Check https://arxiv.org/abs/1905.05178 for more information.

Installation

The implementation is based on the pytorch version of DGCNN.

unzip pytorch_structure2vec-master.zip

Then, under the "pytorch_structure2vec-master/s2vlib/" directory, type

make -j4

to build the necessary c++ backend.

Type

./run_GUNet.sh DATA FOLD

to run on dataset using fold number (1-10). You can run ./run_GUNet.sh DD 0 to run on DD dataset with 10-fold cross validation.

Code

The detail implementation is in ops.py

Datasets

Check the "data/README.md" for the format.

Reference

If you find the code useful, please cite our paper:

@inproceedings{gao2019graph,
  title={Graph {U-nets}},
  author={Gao, Hongyang and Ji, Shuiwang},
  booktitle={Proceedings of The 36th International Conference on Machine Learning},
  year={2019},
}
You can’t perform that action at this time.