xdweixia / scagc

The Pytorch implementation demo for our TMM article: Self-consistent Contrastive Attributed Graph Clustering with Pseudo-label Prompt.

SCAGC

This repo provides a demo for the TMM-2022 article: Self-consistent Contrastive Attributed Graph Clustering with Pseudo-label Prompt on the ACM dataset.

Requirements

• Python 3.6
• PyTorch 1.6.0
• PyTorch Geometric 1.6.1

Training

-Step 1: Warm-Up:

python \ACM-Pretrain-1\Pre_train_1.py

Once the training is finished, then remember copy the generated Pre_train_1.pkl to the folder ACM-Pretrain-2:

python \ACM-Pretrain-2\Pretrain_2.py

Once the training is finished, then remember copy the generated ACM_pretrain.pkl to the folder ACM-Final:

-Step 2: Clustering:

python \ACM-Final\run_cluster.py

Acknowledgements

Some codes are adapted from GCA and SupContrast. We thank them for their excellent projects.

Contact

If you have any problem about our code, feel free to contact [email protected] or describe your problem in Issues.

scagc's Issues

关于代码

您好：
读了您的文章，但是您没有释放代码，能要一下代码吗？想将您的算法做一个应用试试

谢谢！

模型复现问题

您好，我对您的工作很感兴趣，于是复现了一下您的代码，但是遇到了个问题，想和你讨论一下原因。
我Pretrain1训练之后的结果是acc = 0.88 ,nmi = 0.63，但是使用pretrain2加了对比学习训练之后，反而最终结果只有acc = 0.84 nmi = 0.54，这也导致了最后的Final train准确率也只有0.85，和论文结果有些差距，我也在找原因，但是目前没看出来，想和您交流请教谢谢