Research map / Graph learning / Graph self-supervised learning
GC-Flow: A Graph-Based Flow Network for Effective Clustering
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 |
|---|---|---|---|
| replace discriminative GCN layers with invertible graph flows and a Gaussian-mixture representation objective | GCN · Cora, Citeseer, Pubmed, Computers, Photo, Wiki-CS | Silhouette: GC-Flow: 0.734 $\pm$ 0.006; 0.538 $\pm$ 0.022; 0.669 $\pm$ 0.021; 0.487 $\pm$ 0.012; 0.655 $\pm$ 0.013; 0.717 $\pm$ 0.01 | Helped |
| use graph-aware GC-Flow instead of the strongest standard GNN baseline | GC-Flow · Computers | Micro-F1: 0.812 $\pm$ 0.016 -> 0.847 $\pm$ 0.007 | Helped |
| use graph-aware GC-Flow instead of the strongest standard GNN baseline | GC-Flow · Cora, Citeseer, Photo | Micro-F1: Cora: 0.825 $\pm$ 0.005 vs 0.815 $\pm$ 0.011; Citeseer: 0.702 $\pm$ 0.007 -> 0.714 $\pm$ 0.011; Photo: 0.900 $\pm$ 0.009 | Mixed |
| parameterize edge weights with attention (ablation) | GC-Flow · Pubmed, Computers, Photo | Silhouette: Pubmed: 0.669 $\pm$ 0.021 -> 0.804 $\pm$ 0.010; Computers: 0.487 $\pm$ 0.012 -> 0.706 $\pm$ 0.019; Photo: 0.655 $\pm$ 0. | Helped |
| parameterize edge weights with attention (ablation) | GC-Flow · Pubmed, Computers, Photo | Micro-F1: Pubmed: 0.791 $\pm$ 0.009 -> 0.790 $\pm$ 0.007; Computers: 0.847 $\pm$ 0.007 -> 0.841 $\pm$ 0.006; Photo: 0.917 $\pm$ 0. | No clear effect |
| learn graph structure by estimating edge presence (ablation) | GC-Flow · Pubmed, Computers, Photo | Silhouette: Pubmed: 0.669 $\pm$ 0.021 -> 0.856 $\pm$ 0.029; Computers: 0.487 $\pm$ 0.012 -> 0.582 $\pm$ 0.020; Photo: 0.655 $\pm$ 0. | Helped |
| learn graph structure by estimating edge presence (ablation) | GC-Flow · Pubmed, Computers, Photo | Micro-F1: Pubmed: 0.791 $\pm$ 0.009 -> 0.783 $\pm$ 0.009; Computers: 0.847 $\pm$ 0.007 -> 0.851 $\pm$ 0.009; Photo: 0.917 $\pm$ 0. | No clear effect |
| increase the number of flows/layers (ablation) | GC-Flow · Cora, Pubmed | Silhouette | Mixed |
Closest papers by meaning
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.91)
- Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning (AAAI 2018 · similarity 0.91)
- Effects of Graph Convolutions in Multi-layer Networks (ICLR 2023 · similarity 0.90)
- Deep Contrastive Graph Learning with Clustering-Oriented Guidance (AAAI 2024 · similarity 0.89)
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective (NeurIPS 2025 · similarity 0.89)
- GC4NC: A Benchmark Framework for Graph Condensation on Node Classification with New Insights (NeurIPS 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.