Research map / Graph learning / Graph algorithms and clustering
AgraSSt: Approximate Graph Stein Statistics for Interpretable Assessment of Implicit Graph Generators
Wenkai Xu, Gesine Reinert
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 AgraSSt instead of degree-variance statistic Deg to assess NetGAN | AgraSSt · ERGM-trained generators | rejection rate: 0.13 -> 0.81 | Improved |
| Use AgraSSt instead of Deg to assess GraphRNN | AgraSSt · ERGM-trained generators | rejection rate: 0.02 -> 0.42 | Improved |
| Use AgraSSt instead of MDdeg on NetGAN | AgraSSt · ERGM-trained generators | rejection rate: 0.61 -> 0.81 | Improved |
| Use AgraSSt instead of TV_deg on GraphRNN | AgraSSt · ERGM-trained generators | rejection rate: 0.27 -> 0.42 | Improved |
| Run on reliable MC generator: type-I error control | AgraSSt · MC baseline | rejection rate: 0.04 | No clear effect |
| Estimate q-hat with Edges statistic vs degree-based statistics (Sum_deg, Cum_deg, Bi_deg) (ablation) | AgraSSt · Synthetic E2ST | rejection rate (power) | Improved |
| Use 2-d vector of both vertex degrees instead of sum of degrees (ablation) | AgraSSt · Synthetic E2ST | power | Improved |
| Increase re-sampling size B (ablation) | AgraSSt · Synthetic E2ST | power | Improved |
| Compare AgraSSt with TV_deg and Deg on synthetic E2ST | AgraSSt · Synthetic E2ST | power | Improved |
| Use MDdeg and Param instead of AgraSSt on sparse alternatives | MDdeg/Param · Synthetic E2ST | rejection rate | Hurt |
| Use AgraSSt instead of Deg on Karate Club GraphRNN | AgraSSt · Karate Club | p-value: 0.01 -> 0.00 | Improved |
| Use AgraSSt instead of MDdeg on Karate Club NetGAN | AgraSSt · Karate Club | p-value: 0.59 -> 0.00 | Improved |
Closest papers by meaning
- A Non-Asymptotic Convergent Analysis for Scored-Based Graph Generative Model via a System of Stochastic Differential Equations (ICML 2025 · similarity 0.87)
- Quality Measures for Dynamic Graph Generative Models (ICLR 2025 · similarity 0.87)
- Generative Graph Pattern Machine (NeurIPS 2025 · similarity 0.87)
- GRAVER: Generative Graph Vocabularies for Robust Graph Foundation Models Fine-tuning (NeurIPS 2025 · similarity 0.86)
- On the Role of Edge Dependency in Graph Generative Models (ICML 2024 · similarity 0.86)
- Unifying Generation and Prediction on Graphs with Latent Graph Diffusion (NeurIPS 2024 · 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.
