Research map / Label-efficient and robust learning / Self-supervised and contrastive
From Canonical Correlation Analysis to Self-supervised Graph Neural Networks
Hengrui Zhang, Qitian Wu, Junchi Yan, David Wipf, Philip S. Yu
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 CCA-SSG instead of MVGRL | self-supervised node representation learning · Cora | test accuracy: 83.5 ± 0.4 -> 84.2 ± 0.4 | Improved |
| Use CCA-SSG instead of MVGRL | self-supervised node representation learning · Citeseer | test accuracy: 73.3 ± 0.5 -> 73.1 ± 0.3 | No clear effect |
| Use CCA-SSG instead of GRACE | self-supervised node representation learning · Pubmed | test accuracy: 80.6 ± 0.4 -> 81.6 ± 0.4 | Improved |
| Use CCA-SSG instead of GCA | self-supervised node representation learning · Amazon Computer | test accuracy: 87.85 ± 0.31 -> 88.74 ± 0.28 | Improved |
| Use CCA-SSG instead of GCA | self-supervised node representation learning · Amazon Photo | test accuracy: 92.49 ± 0.09 -> 93.14 ± 0.14 | Improved |
| Use CCA-SSG instead of GCA | self-supervised node representation learning · Coauthor CS | test accuracy: 93.10 ± 0.01 -> 93.31 ± 0.22 | Improved |
| Use CCA-SSG instead of GCA | self-supervised node representation learning · Coauthor Physics | test accuracy: 95.68 ± 0.05 -> 95.38 ± 0.06 | No clear effect |
| Remove the invariance term (ablation) | CCA-SSG · node classification benchmarks | test accuracy | Hurt |
| Add the invariance term (ablation) | CCA-SSG without the invariance term · node classification benchmarks | test accuracy | Improved |
| Remove the decorrelation term (ablation) | CCA-SSG · node classification benchmarks | test accuracy | Hurt |
| Add the decorrelation term (ablation) | CCA-SSG without the decorrelation term · node classification benchmarks | test accuracy | Improved |
| Increase the decorrelation weight from too small to a suitable value (ablation) | CCA-SSG · Cora, Citeseer, Pubmed | test accuracy | Improved |
Closest papers by meaning
- Exploring Correlations of Self-Supervised Tasks for Graphs (ICML 2024 · similarity 0.89)
- Equivalence is All: A Unified View for Self-supervised Graph Learning (ICML 2025 · similarity 0.88)
- Uncovering the Redundancy in Graph Self-supervised Learning Models (NeurIPS 2024 · similarity 0.88)
- Revisiting Self-Supervised Heterogeneous Graph Learning from Spectral Clustering Perspective (NeurIPS 2024 · similarity 0.87)
- Through the Dual-Prism: A Spectral Perspective on Graph Data Augmentation for Graph Classifications (AAAI 2025 · similarity 0.87)
- Disentangled Graph Self-supervised Learning for Out-of-Distribution Generalization (ICML 2024 · similarity 0.87)
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.
