Research map / Graph learning / Graph self-supervised learning
Graph Contrastive Learning Automated
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 |
|---|---|---|---|
| automatically optimize augmentation-pair selection instead of using fixed manually chosen pairs | GraphCL · TUDataset semi-supervised benchmarks | average rank | Mixed |
| add augmentation-aware projection heads (ablation) | JOAO · NCI1, augmentation strength 0.2 | accuracy: 61.77$\pm$1.61 -> 62.52$\pm$1.16 | Helped |
| add augmentation-aware projection heads (ablation) | JOAO · NCI1, augmentation strength 0.25 | accuracy: 60.95$\pm$0.55 -> 61.67$\pm$0.72 | Helped |
| add augmentation-aware projection heads (ablation) | JOAO · PROTEINS, augmentation strength 0.2 | accuracy: 71.45$\pm$0.89 -> 71.66$\pm$1.10 | Helped |
| add augmentation-aware projection heads (ablation) | JOAO · PROTEINS, augmentation strength 0.25 | accuracy: 71.61$\pm$1.65 -> 73.01$\pm$1.02 | Helped |
| automatically select augmentations on unseen bioinformatics datasets rather than use GraphCL's TUDataset-derived rules | JOAOv2 · bioinformatics transfer benchmarks | average rank | Helped |
| use automated augmentation selection without incorporating domain-specific expertise | JOAOv2 · bioinformatics transfer benchmarks | average rank | Mixed |
| automatically optimize augmentation selection | GraphCL · ogbg-ppa, 1% label rate | accuracy: 40.81$\pm$1.33 -> 47.19$\pm$1.30 | Helped |
| automatically optimize augmentation selection | GraphCL · ogbg-ppa, 10% label rate | accuracy: 57.77$\pm$1.25 -> 60.91$\pm$0.83 | Helped |
| add augmentation-aware projection heads (ablation) | JOAO · ogbg-ppa, 1% label rate | accuracy: 47.19$\pm$1.30 -> 44.30$\pm$1.67 | Hurt |
| add augmentation-aware projection heads (ablation) | JOAO · ogbg-ppa, 10% label rate | accuracy: 60.91$\pm$0.83 -> 59.32$\pm$1.11 | Hurt |
| add augmentation-aware projection heads (ablation) | JOAO · ogbg-code, 1% label rate | F1 score: 6.84$\pm$0.31 -> 7.74$\pm$0.24 | Helped |
Closest papers by meaning
- Graph Contrastive Learning with Augmentations (NeurIPS 2020 · similarity 0.94)
- Community-Invariant Graph Contrastive Learning (ICML 2024 · similarity 0.91)
- Perfect Alignment May be Poisonous to Graph Contrastive Learning (ICML 2024 · similarity 0.90)
- Unified Graph Augmentations for Generalized Contrastive Learning on Graphs (NeurIPS 2024 · similarity 0.90)
- Graph Contrastive Learning with Joint Spectral Augmentation of Attribute and Topology (AAAI 2025 · similarity 0.90)
- ReGCL: Rethinking Message Passing in Graph Contrastive Learning (AAAI 2024 · 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.