Research map / Graph learning / GNN architectures
Going Deep: Graph Convolutional Ladder-Shape Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase GCN/GAT depth from 2 to 8 layers | GCN and GAT · Cora, Citeseer, Pubmed node classification | accuracy: degraded by more than 60% | Hurt |
| use GCLN (ladder shape with contextual feature channels) instead of 8-layer GCN/GAT | 8-layer GCN and GAT · Cora, Citeseer, Pubmed node classification | accuracy: 26.3% to 173.1% margin | Helped |
| add residual connections after every layer | 8-layer GCN · Cora, Citeseer, Pubmed node classification | accuracy | Helped |
| add residual connections after every layer | 8-layer GAT · Cora, Citeseer, Pubmed node classification | accuracy | Hurt |
| use GCLN instead of other clustering baselines | GCLN · Cora and Citeseer clustering | accuracy, F1, NMI, precision, ARI | Helped |
| use GCLN instead of the latest clustering algorithm | GCLN · Pubmed clustering | clustering metrics | No clear effect |
| use 8-layer GCLN with DIFFPOOL instead of 2-layer GraphSAGE with DIFFPOOL | DIFFPOOL · Graph classification | accuracy | Helped |
Closest papers by meaning
- Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning (AAAI 2018 · similarity 0.91)
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.91)
- Simplifying Graph Convolutional Networks (ICML 2019 · similarity 0.90)
- Geom-GCN: Geometric Graph Convolutional Networks (ICLR 2020 · similarity 0.90)
- Simple Spectral Graph Convolution (ICLR 2021 · similarity 0.90)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.