Research map / Graph learning / Graph algorithms and clustering
DAG-GNN: DAG Structure Learning with Graph Neural 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 DAG-GNN instead of GOPNILP | DAG structure learning · Child | BIC: -1.27e+4 -> -1.38e+4 | Hurt |
| Use DAG-GNN instead of GOPNILP | DAG structure learning · Alarm | BIC: -1.12e+4 -> -1.28e+4 | Hurt |
| Use DAG-GNN instead of GOPNILP | DAG structure learning · Pigs | BIC: -3.50e+5 -> -3.69e+5 | Hurt |
| Use DAG-GNN instead of FGS | Protein signaling network · Protein signaling network | SHD: 22 -> 19 | Helped |
| Use DAG-GNN instead of NOTEARS | Protein signaling network · Protein signaling network | SHD: 22 -> 19 | Helped |
| Use a nonlinear graph-generative model rather than the linear SEM model | Synthetic nonlinear data · Synthetic data generated by x=A^T cos(x+1)+z | FDR | Helped |
| Use a nonlinear graph-generative model rather than the linear SEM model | Synthetic nonlinear data · Synthetic data generated by x=2 sin(A^T(x+0.5·1))+A^T(x+0.5·1)+z | SHD | Helped |
| Use the DAG-GNN vector-valued model with d=5 and d_Z=1 rather than DAG-NOTEARS | Synthetic vector-valued data · Synthetic vector-valued data | SHD | Helped |
| Switch the MLP and identity mappings inside each encoder/decoder (ablation) | DAG-GNN · Synthetic data | Performance | Hurt |
Closest papers by meaning
- D-VAE: A Variational Autoencoder for Directed Acyclic Graphs (NeurIPS 2019 · similarity 0.91)
- DAGs with No Curl: An Efficient DAG Structure Learning Approach (ICML 2021 · similarity 0.91)
- On the Role of Sparsity and DAG Constraints for Learning Linear DAGs (NeurIPS 2020 · similarity 0.91)
- Directed Graph Grammars for Sequence-based Learning (ICML 2025 · similarity 0.88)
- Analytic DAG Constraints for Differentiable DAG Learning (ICLR 2025 · similarity 0.87)
- CoLiDE: Concomitant Linear DAG Estimation (ICLR 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.