https://github.com/vikasverma1077/GraphMix
Code for reproducing results in GraphMix paper
https://github.com/vikasverma1077/GraphMix
gcn gnn graphneuralnetwork regularizer semisupervised-learning
Last synced: 4 months ago
JSON representation
Code for reproducing results in GraphMix paper
- Host: GitHub
- URL: https://github.com/vikasverma1077/GraphMix
- Owner: vikasverma1077
- Created: 2019-09-25T17:07:08.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2022-11-22T03:16:17.000Z (over 2 years ago)
- Last Synced: 2024-08-24T17:26:25.814Z (8 months ago)
- Topics: gcn, gnn, graphneuralnetwork, regularizer, semisupervised-learning
- Language: Python
- Homepage:
- Size: 7.99 MB
- Stars: 71
- Watchers: 5
- Forks: 14
- Open Issues: 5
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- Awesome-Mixup - [Code
README
# GraphMix
This is the Pytorch code for reproducing the results of the Paper GraphMix (https://arxiv.org/pdf/1909.11715.pdf). GraphMix is a simple and efficient regularization method for Graph Neural Networks (GNN) based Semi-Supervised node classification. We show that with this regularizer, even the simpler GNN architecture such as GCN (Kipf et. al.) can achiever state-of-the-art results on benchmark graph datasets such as Cora/Citeseer/Pubmed.
## Requirements
This code is tested with Python 3.6.3 and requires following packages (see requirements.txt list of all the packages):
torch==1.1.0
numpy==1.16.3
pandas==0.24.1
Pillow==5.3.0
scikit-learn==0.21.2
scipy==1.2.1
seaborn==0.9.0
six==1.12.0
tqdm==4.32.2
# How to run
For reproducing results of GraphMix(GCN) of Table1 in the paper, go to directory GraphMix/semisupervised/codes and run the following commands:
`python run_cora.py`
`python run_citeseer.py`
`python run_pubmed.py`
This codebase is based on the github repo : https://github.com/DeepGraphLearning/GMNN
To all the people using this codebase: let us know in the "issues" if you are having some difficulity in reproducing the results.