Research map / Graph learning / Knowledge and temporal graphs
Dynamic Graph Generation Network: Generating Relational Knowledge From Diagrams
Daesik Kim, YoungJoon Yoo, Jee-Soo Kim, SangKuk Lee, Nojun Kwak
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 DGGN graph-structured message passing instead of a vanilla GRU | Diagram relation inference · AI2D test | mAP: 39.28 -> 44.08 | Improved |
| Use DGGN graph-structured message passing instead of a vanilla GRU | Diagram relation inference · AI2D test | IoU_edge: 15.58 -> 25.86 | Improved |
| Use diagram graph extraction in the QA pipeline instead of Dqa-Net(Dsdp) | Dqa-Net · AI2D | accuracy: 38.47 -> 39.73 | Improved |
| Use diagram graph extraction in the QA pipeline instead of Dqa-Net(Dsdp) | Dqa-Net · FOODWEBS | accuracy: 59.30 -> 58.22 | Hurt |
| Add the global diagram feature to DGGN (ablation) | DGGN · AI2D test | mAP: 39.34 -> 44.08 | Improved |
| Add the global diagram feature to DGGN (ablation) | DGGN · AI2D test | IoU_edge: 14.44 -> 25.86 | Improved |
| Add the global diagram feature to DGGN (ablation) | DGGN · AI2D test | IoU_node: 70.95 -> 69.77 | Hurt |
| Use edge-probability weights in the mean pool (ablation) | DGGN · AI2D test | mAP: 42.15 -> 44.08 | Improved |
| Use edge-probability weights in the mean pool (ablation) | DGGN · AI2D test | IoU_edge: 24.84 -> 25.86 | Improved |
| Add a 2 × 2 ROI-pooled feature to DGGN local features (ablation) | DGGN · AI2D test | mAP: 44.08 -> 39.73 | Hurt |
| Add a 2 × 2 ROI-pooled feature to DGGN local features (ablation) | DGGN · AI2D test | IoU_edge: 25.86 -> 23.00 | Hurt |
| Replace the graph generation branch with fully connected layers | Diagram relation inference · AI2D test | mAP: 44.08 -> 8.87 | Hurt |
Closest papers by meaning
- CoG-DQA: Chain-of-Guiding Learning with Large Language Models for Diagram Question Answering (CVPR 2024 · similarity 0.85)
- Flatten Graphs as Sequences: Transformers are Scalable Graph Generators (NeurIPS 2025 · similarity 0.85)
- From Words to Structured Visuals: A Benchmark and Framework for Text-to-Diagram Generation and Editing (CVPR 2025 · similarity 0.85)
- A Deep Probabilistic Framework for Continuous Time Dynamic Graph Generation (AAAI 2025 · similarity 0.84)
- Towards Dynamic Message Passing on Graphs (NeurIPS 2024 · similarity 0.84)
- Coreference Graph Guidance for Mind-Map Generation (AAAI 2024 · 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.
