Paper link: https://www.ijcai.org/proceedings/2022/498
The performance of GAEs degrades with the increase of layers:
Our main model (i.e. Model (5)) is as follows:
The loss function is as follows:
Where
The performance of our proposed models compared to the orignal models:
The main codes for our proposed models, including DGAE${\alpha}$, DGAE${\alpha}^{\beta}$, and DVGA$_{\alpha}^{\beta}$ in our paper " Codes of Stabilizing and Enhancing Link Prediction through Deepened Graph Auto-Encoders". The files in the subfolder "Python", are the Python source codes, which have been implemented in JupyterLab. For readability, we also provide the html files in the subfolder "Html".
If you find this code useful in your research, please consider citing our work:
Xinxing Wu, Qiang Cheng. Deepened Graph Auto-Encoders Help Stabilize and Enhance Link Prediction. The 31st International Joint Conference on Artificial Intelligence and the 25th European Conference on Artificial Intelligence (IJCAI-ECAI 2022). 2022. https://www.ijcai.org/proceedings/2022/498
- 0.0.1
- Codes published
Distributed under the MIT license. See LICENSE for more information.
Xinxing Wu (xinxingwu@gmail.com) and Qiang Cheng (qiang.cheng@uky.edu)