WSSGCN: Wide Sub-stage Graph Convolutional Networks

• Paper

Summary

WSSGCN studies a wider, staged alternative to deep graph convolutional networks. Its architecture combines node and edge, subgraph, and global features, with response-, feature-, and relationship-based consistency across stages. A width-expansion strategy supports efficient training. Experiments on graph benchmarks compare accuracy and training speed with conventional GCN architectures.

Citation

@article{Wang24WSSGCN,
  author = {Chao Wang and Zheng Tang and Hailu Xu},
  title = { {WSSGCN}: {W}ide sub-stage graph convolutional networks},
  journal = {Neurocomputing},
  volume = {602},
  pages = {128273},
  year = {2024}
}