Cross-Space Adaptive Filter: Integrating Graph Topology and Node Attributes for Alleviating the Over-smoothing Problem

Contact: Chen Huang

A cross-space filter is proposed to produce the adaptive-frequency information extracted from both the topology and attribute spaces. It alleviates the over-smoothing problem while simultaneously promoting the effectiveness of downstream tasks.

Paper

arvix version

Camera Ready: TODO

Codes

This code is based on Graph Convolutional Networks in PyTorch.

Reference

If you make advantage of us in your research, please cite the following in your manuscript:

@inproceedings{huang2024csf,
  title={Cross-Space Adaptive Filter: Integrating Graph Topology and Node Attributes for Alleviating the Over-smoothing Problem},
  author={Chen Huang, Haoyang Li, Yifan Zhang, Wenqiang Lei, Jiancheng Lv},
  booktitle={Proceedings of the ACM Web Conference 2024},
  year={2024}
}