Research map / Graph learning / Graph self-supervised learning
Effective Decoding in Graph Auto-Encoder Using Triadic Closure
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 the pairwise inner-product decoder with a triad decoder | VGAE · Cora, Citeseer, and PubMed link prediction | AUC and AP: Cora: 94.4 -> 96.0 AUC; 95.9 -> 96.3 AP. Citeseer: 93.4 -> 96.2 AUC; 95.2 -> 96.5 AP. PubMed: 96.2 -> 98.5 AUC; 96.4 -> | Helped |
| Replace the pairwise inner-product decoder with a triad decoder | GAE · Cora, Citeseer, and PubMed link prediction | AUC and AP: Cora: 93.1 -> 95.6 AUC; 95.0 -> 96.2 AP. Citeseer: 91.5 -> 96.5 AUC; 92.6 -> 97.0 AP. PubMed: 97.1 -> 98.2 AUC; 97.4 -> | Helped |
| Replace the pairwise decoder with a triad decoder | VGAE · Cora node clustering | accuracy: 0.662 -> 0.753 | Helped |
| Replace the pairwise decoder with a triad decoder | VGAE · Citeseer node clustering | accuracy: 0.515 -> 0.591 | Helped |
| Replace the pairwise decoder with a triad decoder | GAE · Cora node clustering | accuracy: 0.602 -> 0.728 | Helped |
| Replace the pairwise decoder with a triad decoder | GAE · Citeseer node clustering | accuracy: 0.430 -> 0.611 | Helped |
| Use the triad decoder and triad-based graph generation | TVGA · Cora graph generation | average rank across graph statistics: 2.00 average rank | Helped |
| Use the triad decoder and triad-based graph generation | TVGA · Citeseer graph generation | average rank across graph statistics: 2.71 average rank | Helped |
| Use proposed balanced triad sampling instead of random triad sampling (ablation) | TVGA · Cora, Citeseer, and PubMed link prediction | AUC and AP: Cora: 68.5 -> 96.0 AUC; 68.5 -> 96.3 AP. Citeseer: 60.5 -> 96.2 AUC; 65.3 -> 96.5 AP. PubMed: 82.9 -> 98.5 AUC; 81.2 -> | Helped |
| Use proposed balanced triad sampling instead of random triad sampling (ablation) | TGA · Cora, Citeseer, and PubMed link prediction | AUC and AP: Cora: 76.3 -> 95.6 AUC; 76.9 -> 96.2 AP. Citeseer: 69.2 -> 96.5 AUC; 69.1 -> 97.0 AP. PubMed: 89.6 -> 98.2 AUC; 88.7 -> | Helped |
| Use deterministic GAE rather than variational VGAE with the triad decoder (ablation) | Triad auto-encoder · Cora node clustering | accuracy: 0.753 -> 0.728 | Hurt |
| Use deterministic GAE rather than variational VGAE with the triad decoder (ablation) | Triad auto-encoder · Citeseer node clustering | accuracy: 0.591 -> 0.611 | Helped |
Closest papers by meaning
- HC-GAE: The Hierarchical Cluster-based Graph Auto-Encoder for Graph Representation Learning (NeurIPS 2024 · similarity 0.89)
- GraphCroc: Cross-Correlation Autoencoder for Graph Structural Reconstruction (NeurIPS 2024 · similarity 0.89)
- Rethinking Graph Masked Autoencoders through Alignment and Uniformity (AAAI 2024 · similarity 0.87)
- The quest for the GRAph Level autoEncoder (GRALE) (NeurIPS 2025 · similarity 0.87)
- Teacher-guided Edge Discriminator for Personalized Graph Masked Autoencoder (AAAI 2025 · similarity 0.87)
- Self-supervised Masked Graph Autoencoder via Structure-aware Curriculum (ICML 2025 · similarity 0.86)
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.