Research map / Graph learning / Graph self-supervised learning
ImGCL: Revisiting Graph Contrastive Learning on Imbalanced Node Classification
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 |
|---|---|---|---|
| Apply ImGCL to GBT | GBT · Amazon-Computers, Amazon-Photo, Wiki-CS, and DBLP | accuracy: 42.17_{\pm3.74} -> 78.62_{\pm1.73}; 60.73_{\pm3.64} -> 75.13_{\pm7.13}; 44.95_{\pm2.63} -> 73.08_{\pm12.45}; 58.51_{\pm5 | Helped |
| Apply ImGCL to GRACE | GRACE · Wiki-CS | accuracy: 54.20_{\pm3.97} -> 73.86_{\pm2.78} | Helped |
| Balance sampling with ImGCL | GBT · Amazon-Computers, Exp factor 100 | middle accuracy: 41.92_{\pm3.28} -> 82.72_{\pm8.46} | Helped |
| Balance sampling with ImGCL | GBT · Amazon-Computers, Exp factor 100 | tail accuracy: 14.29_{\pm3.53} -> 94.45_{\pm6.07} | Helped |
| Balance sampling with ImGCL | GBT · Amazon-Computers, Exp factor 100 | head accuracy: 70.38_{\pm5.96} -> 57.32_{\pm7.29} | Hurt |
| Use iterative pseudo-labels and centrality-based PBS instead of the unmodified baseline | GBT · Amazon-Computers, Exp factor 100 | accuracy: 42.17_{\pm3.74} -> 78.62_{\pm1.73} | Helped |
| Use one-stage K-means pseudo-labels (ablation) | GBT · Amazon-Computers, Exp factor 100 | accuracy: 42.17_{\pm3.74} -> 75.99_{\pm6.75} | Helped |
| Use one-stage community-detection pseudo-labels (ablation) | GBT · Amazon-Computers, Exp factor 100 | accuracy: 42.17_{\pm3.74} -> 76.57_{\pm3.87} | Helped |
| Add node-centrality weighting to PBS (ablation) | GBT+PBS · Amazon-Computers, Exp factor 100 | accuracy: 76.70_{\pm6.94} -> 78.62_{\pm1.73} | Helped |
| Replace generated pseudo-labels with ground-truth labels (ablation) | GBT+PBS · Amazon-Computers, Exp factor 100 | accuracy: 78.62_{\pm1.73} -> 78.96_{\pm2.92} | No clear effect |
| Increase imbalance in the Exp setting | GBT · Amazon-Computers, Exp factors 20, 50, 100, and 200 | accuracy: 71.97_{\pm3.78}; 65.49_{\pm3.37}; 42.17_{\pm3.74}; 36.57_{\pm4.30} | Hurt |
Closest papers by meaning
- Cluster-guided Contrastive Class-imbalanced Graph Classification (AAAI 2025 · similarity 0.92)
- Does GCL Need a Large Number of Negative Samples? Enhancing Graph Contrastive Learning with Effective and Efficient Negative Sampling (AAAI 2025 · similarity 0.91)
- Hierarchical Topology Isomorphism Expertise Embedded Graph Contrastive Learning (AAAI 2024 · similarity 0.89)
- ReGCL: Rethinking Message Passing in Graph Contrastive Learning (AAAI 2024 · similarity 0.89)
- S3GCL: Spectral, Swift, Spatial Graph Contrastive Learning (ICML 2024 · similarity 0.89)
- IGL-Bench: Establishing the Comprehensive Benchmark for Imbalanced Graph Learning (ICLR 2025 · similarity 0.88)
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.