Research map / Graph learning / Graph self-supervised learning
Efficient Graph Generation with Graph Recurrent Attention Networks
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 GRAN instead of GraphRNN | graph generation · Grid | degree MMD: 1.12$e^{-2}$ -> 8.23$e^{-4}$ | Helped |
| Use GRAN instead of GraphRNN | graph generation · Protein | spectrum MMD: 1.88$e^{-2}$ -> 5.13$e^{-3}$ | Helped |
| Use GRAN instead of GraphRNN | graph generation · Lobster | lobster validity accuracy: 1.00 -> 0.88 | Hurt |
| Increase Bernoulli mixture components from 1 to 20 (ablation) | GRAN · Grid | orbit MMD: 2.66$e^{-5}$ -> 4.27$e^{-6}$ | Helped |
| Increase Bernoulli mixture components from 20 to 50 (ablation) | GRAN · Grid | orbit MMD: 4.27$e^{-6}$ -> 9.56$e^{-7}$ | Helped |
| Use DFS ordering rather than the default ordering (ablation) | GRAN · Grid | orbit MMD: 6.66$e^{-4}$ -> 4.27$e^{-6}$ | Helped |
| Use all five canonical orderings instead of DFS alone (ablation) | GRAN · Grid | orbit MMD: 4.27$e^{-6}$ -> 6.66$e^{-4}$ | Helped |
| Increase block size from 1 to 16, keeping stride 1 (ablation) | GRAN · Grid | orbit MMD: 2.66$e^{-5}$ -> 9.32$e^{-3}$ | Hurt |
| Increase sampling stride from 1 to 16 at block size 16 (ablation) | GRAN · Grid | run time: 33.1 -> 2.2 | Helped |
| Increase sampling stride from 1 to 16 at block size 16 (ablation) | GRAN · Grid | orbit MMD: 9.32$e^{-3}$ -> 1.03$e^{-2}$ | Hurt |
Closest papers by meaning
- A Large-scale Training Paradigm for Graph Generative Models (ICLR 2025 · similarity 0.89)
- Towards Deep Attention in Graph Neural Networks: Problems and Remedies (ICML 2023 · similarity 0.89)
- Generative Graph Pattern Machine (NeurIPS 2025 · similarity 0.88)
- GraphTOP: Graph Topology-Oriented Prompting for Graph Neural Networks (NeurIPS 2025 · similarity 0.87)
- A Dynamical Systems-Inspired Pruning Strategy for Addressing Oversmoothing in Graph Attention Networks (ICML 2025 · similarity 0.87)
- Towards Addressing Frontiers in Graph Generation (AAAI 2025 · similarity 0.87)
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.