Research map / Graph learning / Knowledge and temporal graphs
Relation-aware Graph Attention Model with Adaptive Self-adversarial Training
Xiao Qin, Nasrullah Sheikh, Berthold Reinwald, Lingfei Wu
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 RelGNN instead of R-GCN + ASA | heterogeneous relationship prediction · Amazon | AUC | Improved |
| Use RelGNN instead of R-GCN + ASA | heterogeneous relationship prediction · Youtube | AUC | Improved |
| Use RelGNN instead of R-GCN + ASA | heterogeneous relationship prediction · Company | AUC | Improved |
| Add ASA rather than random negative sampling (ablation) | RelGNN · Amazon | AUC: 98.84 -> 99.14 | Improved |
| Add ASA rather than random negative sampling (ablation) | RelGNN · Youtube | AUC: 94.39 -> 96.44 | Improved |
| Add ASA rather than random negative sampling (ablation) | RelGNN · Company | AUC: 74.94 -> 77.68 | Improved |
| Add ASA rather than random negative sampling (ablation) | RelGNN · Amazon | F1: 95.17 -> 90.44 | Hurt |
| Add ASA rather than random negative sampling (ablation) | RelGNN · Youtube | F1: 86.32 -> 83.89 | Hurt |
| Add relation-aware attention instead of the no-attention R-GCN variant (ablation) | RelGNN · Company | AUC | Improved |
| Use ASA instead of NSCaching at pool size 10 | negative sampling for RelGNN · Company | MRR: .0789 -> .0818 | Improved |
| Increase ASA pool size from 10 to 100 (ablation) | negative sampling for RelGNN · Company | Hit@30: 22.01 -> 23.71 | Improved |
| Increase NSCaching pool size from 10 to 100 (ablation) | negative sampling for RelGNN · Company | MRR: .0789 -> .0695 | Hurt |
Closest papers by meaning
- RelGNN: Composite Message Passing for Relational Deep Learning (ICML 2025 · similarity 0.88)
- Inductive Relation Prediction Using Analogy Subgraph Embeddings (ICLR 2022 · similarity 0.88)
- muxGNN: Multiplex Graph Neural Network for Heterogeneous Graphs (TPAMI 2023 · similarity 0.88)
- ARNS: Adaptive Relation-Aware Negative Sampling with Curriculum Learning for Inductive Knowledge Graph Completion (AAAI 2026 · similarity 0.87)
- Fully-inductive Node Classification on Arbitrary Graphs (ICLR 2025 · similarity 0.87)
- Learning Latent Graph Structures and their Uncertainty (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.
