Research map / Graph learning / GNN architectures
Learning Dynamic Stability Landscapes in Synchronization Networks
Christian Nauck, Junyou Zhu, Michael Lindner, Frank Hellmann
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 CNN decoder instead of MLP decoder (ablation) | DBGNN encoder · Heatmap SSIM, train 20 eval 100 | SSIM (%): 75.49 -> 81.03 | Improved |
| use CNN decoder instead of MLP decoder (ablation) | TAG encoder · Heatmap SSIM, train 20 eval 20 | SSIM (%): 81.37 -> 84.31 | Improved |
| use DBGNN encoder instead of TAG (ablation) | CNN decoder · Heatmap SSIM, train 20 eval 100 | SSIM (%): 74.36 -> 81.03 | Improved |
| use DBGNN encoder instead of TAG on real grids (ablation) | CNN decoder · Germany zero-shot | SSIM (%): 82.87 -> 53.10 | Hurt |
| predict landscapes then recover SNBS instead of direct SNBS regression | DBGNN · SNBS R2, train 20 eval 20 | R2 (%): 85.68 -> 83.99 | Mixed |
| predict landscapes then recover SNBS on OOD size shift | DBGNN · SNBS R2, train 20 eval 100 | R2 (%): 73.73 -> 70.66 | Hurt |
| train on 100-node graphs instead of 20-node graphs, evaluate on 100 (ablation) | DBGNN-CNN · Heatmap SSIM | SSIM (%): 81.03 -> 84.90 | Improved |
| MLP vs CNN decoder for contingency identification (ablation) | TAG encoder · IoU, train 100 eval 100 | IoU: 0.41 -> 0.47 | Improved |
| MLP vs CNN decoder for contingency identification with TAG, train 20 eval 20 (ablation) | TAG encoder · IoU, train 20 eval 20 | IoU: 0.58 -> 0.58 | No clear effect |
Closest papers by meaning
- Robust Angular Synchronization via Directed Graph Neural Networks (ICLR 2024 · similarity 0.86)
- PowerGraph: A power grid benchmark dataset for graph neural networks (NeurIPS 2024 · similarity 0.85)
- SGNNBench: A Holistic Evaluation of Spiking Graph Neural Networks on Large-scale Graphs (NeurIPS 2026 · similarity 0.85)
- Delay-DSGN: A Dynamic Spiking Graph Neural Network with Delay Mechanisms for Evolving Graph (ICML 2025 · similarity 0.84)
- Benchmarking Structural Inference Methods for Interacting Dynamical Systems with Synthetic Data (NeurIPS 2024 · similarity 0.84)
- Optimal Recurrent Network Topologies for Dynamical Systems Reconstruction (ICML 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.
