Research map / Graph learning / GNN architectures
Graph Cross Networks with Vertex Infomax Pooling
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 |
|---|---|---|---|
| Use GXN with VIPool and intermediate feature crossing | Graph classification · Six graph-classification datasets | average accuracy: 2.12% | Helped |
| Use GXN with VIPool and intermediate feature crossing | Vertex classification · Cora, Citeseer, Pubmed | average accuracy: 1.15% | Helped |
| Replace VIPool with gPool (ablation) | GXN · Graph classification | accuracy | Hurt |
| Replace VIPool with SAGPool (ablation) | GXN · Graph classification | accuracy | Hurt |
| Replace VIPool with AttPool (ablation) | GXN · Graph classification | accuracy | Hurt |
| Replace VIPool with gPool, SAGPool, or AttPool (ablation) | GXN · Cora, Citeseer, Pubmed vertex classification | average accuracy | Hurt |
| Remove feature-crossing layers (ablation) | GXN · Graph classification | accuracy | Hurt |
| Remove feature-crossing layers (ablation) | GXN · Vertex classification | accuracy: 1.10% on average | Hurt |
| Use more feature-crossing across layers rather than one direction or no crossing (ablation) | GXN · Graph classification | accuracy | Helped |
| Increase graph scales from 1 to 2 (ablation) | GXN · IMDB-B | accuracy: 2.53% on average | Helped |
| Use more than 3 graph scales (ablation) | GXN · IMDB-B | accuracy | No clear effect |
| Use more than 2 feature-crossing layers (ablation) | GXN · IMDB-B | accuracy | Hurt |
Closest papers by meaning
- ENAHPool: The Edge-Node Attention-based Hierarchical Pooling for Graph Neural Networks (ICML 2025 · similarity 0.88)
- Geometry-Aware Edge Pooling for Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- SkipPool: Improved Sparse Hierarchical Graph Pooling with Differentiable Exploration (AAAI 2025 · similarity 0.86)
- MaxCutPool: differentiable feature-aware Maxcut for pooling in graph neural networks (ICLR 2025 · similarity 0.85)
- Improving Graph Neural Networks by Learning Continuous Edge Directions (ICLR 2025 · similarity 0.85)
- Integrating Co-Training with Edge Discrimination to Enhance Graph Neural Networks Under Heterophily (AAAI 2025 · similarity 0.85)
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.