LATGCN

This is the implementation of Latent Adversarial Training of Graph Convolutional Networks LATGCN by Hongwei Jin and Xinhua Zhang.

Motivation

maximize the perturbation on the latent layer
perturbation applied to all the nodes, due to the transductive learning property
improve the accuracy
reduce the success rate from adversarial attacks

Usage

Instal the required packages

pip install -r requirements.txt

Run the demo of nettack with vanilla GCN

python demo.py

run the demo of nettack with LATGCN

python demo.py

demo.py has a set of parameter to specify

--dataset:      choose from 
                'citeseer', 'cora', 'cora_ml', 'pubmed', 'polblogs', 'dblp'
--train_share:  specify the size for training
--reg:          a flag to toggle the LATGCN
--eta:          the norm constraint of noise applied to each node embedding
--gamma:        regularizer factor

Note that without specify the reg flag, it is simply the vanilla GCN.

Primary Result

Performance with perturbation on cora dataset:

Success rate with perturbation on cora dataset:

Reference

The Nettack model is implemented by Daniel Zügner
The original GCN model is implemented by Thomas N Kipf

Contact

Feel free to create issue in the repo, I will try to answer as soon as I can.

Citation

@article{jin2019latent,
  title={Latent Adversarial Training of Graph Convolutional Networks},
  author={Jin, Hongwei and Zhang, Xinhua},
  journal={ICML workshop - Learning and Reasoning with Graph-Structured Representations},
  year={2019}
}

Copyright

The work is under MIT license.

cshjin/LATGCN