An implementation of GLapGAN: "Performing Group Difference Testing on Graph Structured Data from GANs: Analysis and Applications in Neuroimaging"

This is an PyTorch implementation of GLapGAN as described in the paper [Performing Group Difference Testing on Graph Structured Data from GANs: Analysis and Applications in Neuroimaging]

Training Recipe

1. number of epochs: 500
2. initial learning rates (1e-5, 5*1e-5) for generator and discriminator

Datasets

Processed ADNI data can not be released without special permission in public. Please email us for the whole processed adni data. Here, we provide two samples (one health and one disease) for understanding the input data. The two-sample ADNI data can be downloaded at here. Then put it under src folder.

Trained models

Trained models can be downloaded at here.

Usage

cd src && python demo_lwgan.py --flag_ttest --num_ads 183 --num_cns 293 --lr_g 0.00001 --lr_d 0.00005 --gpu 0 --flag_reg --result_path ./results/lwgan_adni_293_reg

Reference

If you use any part of this code in your research, please cite our paper:

@article{dinhperforming,
  title={Performing Group Difference Testing on Graph Structured Data from GANs: Analysis and Applications in Neuroimaging},
  author={Dinh, Tuan Quang and Xiong, Yunyang and Huang, Zhichun and Vo, Tien and Mishra, Akshay and Kim, Won Hwa and Ravi, Sathya and Singh, Vikas},
  journal={IEEE transactions on pattern analysis and machine intelligence}
}