Research map / Graph learning / GNN architectures
Graph in Graph Neural Network
Jiongshu Wang, Jing Yang, Jiankang Deng, Hatice Güneş, Siyang Song
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 |
|---|---|---|---|
| use GIG network with graph vertices instead of vector vertices (same GatedGCN updating functions) | GatedGCN · ZINC (10K) | MAE: 0.214 -> 0.125 | Improved |
| use GIG network instead of plain GatedGCN | GatedGCN · MNIST | accuracy: 97.34 -> 98.80 | Improved |
| use GIG network instead of plain GatedGCN | GatedGCN · CIFAR10 | accuracy: 67.31 -> 77.11 | Improved |
| use GIG network instead of CTR-GCN | CTR-GCN · NTU RGB+D X-Sub | top-1 accuracy: 92.4 -> 93.1 | Improved |
| use GIG network instead of ST-GCN | ST-GCN · NTU RGB+D X-View | top-1 accuracy: 88.3 -> 91.4 | Improved |
| add GSG, GVU and GGU modules (graph vertices vs vector vertices) (ablation) | GIG · MNIST | accuracy: 26.32 -> 98.80 | Improved |
| add GVU (local updating) to GSG+GGU (ablation) | GIG · NTU RGB+D X-Sub | top-1 accuracy: 57.0 -> 83.6 | Improved |
| add GGU (global updating) to GSG+GVU (ablation) | GIG · NTU RGB+D X-Sub | top-1 accuracy: 81.5 -> 83.6 | Improved |
| use local-global GVU+GGU updating instead of one large graph (ablation) | GIG · MNIST | accuracy: 94.75 -> 98.80 | Improved |
| connect 10% instead of 100% of graph vertices to proxy vertices (ablation) | GIG · TSP | F1: 0.863 -> 0.857 | No clear effect |
| increase GIG layers from 1 to 3 (ablation) | GIG · MNIST | accuracy: 97.82 -> 98.80 | Improved |
| increase GIG layers from 1 to 3 (ablation) | GIG · PATTERN | accuracy: 86.79 -> 86.79 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- EXPHORMER — Exphormer: Sparse Transformers for Graphs (ICML 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VQ-GNN: A Universal Framework to Scale up Graph Neural Networks using Vector Quantization (NeurIPS 2021 · similarity 0.88)
- SizeShiftReg: a Regularization Method for Improving Size-Generalization in Graph Neural Networks (NeurIPS 2022 · similarity 0.88)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.87)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.86)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.86)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (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.
