Research map / Probabilistic and scientific modeling / Variational and generative sampling
Scalable Deep Generative Relational Model with High-Order Node Dependence
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 |
|---|---|---|---|
| Use the simplified SDREM with data augmentation instead of MMSB | SDREM with F=0 and L=1, compared with MMSB · Citeer, Cora, Pubmed, and PPI | AUC | Helped |
| Use the simplified SDREM with data augmentation instead of MMSB | SDREM with F=0 and L=1, compared with MMSB · Citeer, Cora, Pubmed, and PPI | per-iteration runtime: at least two orders of magnitude lower | Helped |
| Increase the number of membership dimensions K (ablation) | SDREM · Citeer, Cora, Pubmed, and PPI | AUC | Helped |
| Increase the number of layers L (ablation) | SDREM · Citeer, Cora, Pubmed, and PPI | AUC | Helped |
| Restrict information propagation to graph-neighbor connections (ablation) | SDREM versus Full-SDREM and Independent-SDREM · Citeer, Cora, Pubmed, and PPI | AUC | Helped |
| Restrict information propagation to graph-neighbor connections (ablation) | SDREM versus Full-SDREM and Independent-SDREM · Citeer, Cora, Pubmed, and PPI | negative log-likelihood | Helped |
| Allow propagation between every pair of nodes, including unconnected nodes (ablation) | Full-SDREM · Citeer, Cora, Pubmed, and PPI | AUC | Hurt |
| Remove propagation between distinct nodes, retaining only self-propagation (ablation) | Independent-SDREM · Citeer, Cora, Pubmed, and PPI | AUC | Hurt |
| Use SDREM instead of alternative methods | SDREM compared with HLFM, NARM, HGP-EPM, and GCN · Four real-world datasets with training-data ratio greater than 0.5 | AUC | Helped |
| Use SDREM instead of alternative methods | SDREM compared with HLFM, NARM, and HGP-EPM · Four real-world datasets | negative log-likelihood | Mixed |
| Use VGAE instead of SDREM on datasets with sparse linkages | SDREM versus VGAE · Datasets with sparse linkages | AUC | Hurt |
| Use VGAE instead of SDREM on other dataset types | SDREM versus VGAE · Other dataset types | AUC | No clear effect |
Closest papers by meaning
- Deep Generative Models for Relational Data with Side Information (ICML 2017 · similarity 0.90)
- Relational Deep Learning: A Deep Latent Variable Model for Link Prediction (AAAI 2017 · similarity 0.89)
- Poisson-Randomised DirBN: Large Mutation is Needed in Dirichlet Belief Networks (ICML 2021 · similarity 0.87)
- Discriminative Nonparametric Latent Feature Relational Models with Data Augmentation (AAAI 2016 · similarity 0.87)
- Deep Amortized Relational Model with Group-Wise Hierarchical Generative Process (AAAI 2022 · similarity 0.87)
- GLAD: Improving Latent Graph Generative Modeling with Simple Quantization (AAAI 2025 · similarity 0.84)
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.