Research map / Graph learning / GNN architectures
Effects of Graph Convolutions in Multi-layer 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 |
|---|---|---|---|
| Add one graph convolution beyond the first layer | Multi-layer network on XOR-CSBM · XOR-CSBM with p, q = Ω(log² n / n) | mean-distance threshold for perfect classification: 1/√[4]{n(p+q)} | Helped |
| Add a second graph convolution | Multi-layer network on XOR-CSBM · XOR-CSBM with p, q = Ω(log n / √n) | mean-distance threshold for perfect classification: 1/√[4]{n} | Helped |
| Place a graph convolution in the first layer (ablation) | Network on XOR-CSBM · XOR-CSBM | classification accuracy | Hurt |
| Move one graph convolution between the second and third layers (ablation) | Three-layer network on XOR-CSBM · XOR-CSBM | classification performance | No clear effect |
| Redistribute two graph convolutions among the second and third layers (ablation) | Three-layer network on XOR-CSBM · XOR-CSBM | classification performance | No clear effect |
| Add a second graph convolution (ablation) | Network on a dense XOR-CSBM graph · XOR-CSBM with p, q = Ω_n(1) | classification threshold | No clear effect |
| Use graph convolutions rather than an MLP | Node-classification networks · CORA | accuracy | Helped |
| Use graph convolutions rather than an MLP | Node-classification networks · CiteSeer | accuracy | Helped |
| Use graph convolutions rather than an MLP | Node-classification networks · Pubmed | accuracy | Helped |
| Add a second graph convolution instead of using one | Node-classification networks · CORA, CiteSeer, and Pubmed | accuracy | Helped |
| Move graph convolutions among layers while keeping their number fixed (ablation) | Node-classification networks · CORA, CiteSeer, and Pubmed | accuracy | No clear effect |
Closest papers by meaning
- Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning (AAAI 2018 · similarity 0.91)
- Simplifying Graph Convolutional Networks (ICML 2019 · similarity 0.90)
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective (NeurIPS 2025 · similarity 0.90)
- Feature Distribution on Graph Topology Mediates the Effect of Graph Convolution: Homophily Perspective (ICML 2024 · similarity 0.90)
- Analysis of Corrected Graph Convolutions (NeurIPS 2024 · similarity 0.89)
- Graph Attention is Not Always Beneficial: A Theoretical Analysis of Graph Attention Mechanisms via Contextual Stochastic Block Models (ICML 2025 · 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.