Research map / Graph learning / GNN architectures
Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning
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 |
|---|---|---|---|
| Co-train a GCN with partially absorbing random walks to expand the label set | GCN · Cora and CiteSeer, 1% label rate | accuracy: 23% | Helped |
| Co-train a GCN with partially absorbing random walks to expand the label set | GCN · PubMed, 0.05% label rate | accuracy: 37% | Helped |
| Self-train by adding the GCN's most confident predictions | GCN · CiteSeer | accuracy | Helped |
| Use Union to add confident labels from both random walks and the GCN | GCN · Cora, CiteSeer, and PubMed | accuracy | Mixed |
| Use Intersection to retain only confident labels found by both random walks and the GCN | GCN · Cora, CiteSeer, and PubMed | accuracy | Mixed |
| Increase the number of added labels from 2η to 3η or 4η (ablation) | Co-Training, Self-Training, Union, and Intersection · CiteSeer, Cora, and PubMed | accuracy: 2η, 3η and 4η | No clear effect |
| Increase the number of GCN convolutional layers beyond 2 (ablation) | GCN · CiteSeer, Cora, and PubMed | accuracy | Hurt |
| Apply graph convolution (symmetric Laplacian smoothing) instead of using an FCN | GCN · Cora, 20 labels per class | accuracy: 0.530860 -> 0.707940 | Helped |
Closest papers by meaning
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.95)
- Semi-Supervised Learning With Graph Learning-Convolutional Networks (CVPR 2019 · similarity 0.94)
- Simple and Deep Graph Convolutional Networks (ICML 2020 · similarity 0.93)
- Simplifying Graph Convolutional Networks (ICML 2019 · similarity 0.93)
- Generalization Guarantee of Training Graph Convolutional Networks with Graph Topology Sampling (ICML 2022 · similarity 0.93)
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective (NeurIPS 2025 · similarity 0.91)
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.