msga.py is the architecture of our model implemented by tensorflow.
msga_main.py is the main function and also used to load the real-world dataset .
trainer_msga.py records the way the model trained.
Since the dataset is too large, we give the following access.
链接：https://pan.baidu.com/s/1i_3WJaCtSYmet-prceTqFw
提取码：qq6h
Our paper is available at https://doi.org/10.1016/j.neucom.2021.06.058.
If you make advantage of our proposed model in your research, please cite our work.
@article{wang2021multi,
title={Multi-scale graph attention subspace clustering network},
author={Wang, Tong and Wu, Junhua and Zhang, Zhenquan and Zhou, Wen and Chen, Guang and Liu, Shasha},
journal={Neurocomputing},
volume={459},
pages={302--314},
year={2021},
publisher={Elsevier}
}