Research map / Graph learning / GNN architectures
A Simple Spectral Failure Mode for Graph Convolutional Networks
Carey E. Priebe, Cencheng Shen, Ningyuan Teresa Huang, Tianyi Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| embed with ASE into d=2 instead of unsupervised GCN | node classification · Latent position graph, orthogonal first eigenvector | classification error | Improved |
| embed with ASE into d=1 instead of d=2 | adjacency spectral embedding · Latent position graph, orthogonal first eigenvector | classification error | Hurt |
| increase angle difference between first eigenvector and signal support from 0 to pi/14 | unsupervised GCN · Latent position graph, m=100, n=1000 | classification error | Improved |
| use ReLU instead of linear activation (ablation) | unsupervised GCN · Latent position graph | classification error | No clear effect |
| increase labeled ratio m/n from 0.02 to 0.7 | unsupervised GCN and ASE · Latent position graph, n=500 | classification error | Improved |
| tune hidden size and learning rate by cross validation (ablation) | unsupervised GCN · Latent position graph | classification error | No clear effect |
| use semisupervised instead of unsupervised GCN | GCN · Latent position graph | classification error | Improved |
Closest papers by meaning
- HoloNets: Spectral Convolutions do extend to Directed Graphs (ICLR 2024 · similarity 0.89)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.87)
- Spectral Graph Neural Networks are Incomplete on Graphs with a Simple Spectrum (NeurIPS 2025 · similarity 0.87)
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective (NeurIPS 2025 · similarity 0.87)
- Asymmetric Learning for Spectral Graph Neural Networks (AAAI 2025 · similarity 0.86)
- Position: Spectral GNNs Rely Less on Graph Fourier Basis than Conceived (ICML 2025 · similarity 0.86)
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.
