Research map / Graph learning / Knowledge and temporal graphs
Scalable Spatiotemporal 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 |
|---|---|---|---|
| add spatial propagation to the encoder (ablation) | SGP · METR-LA | MAE (average): 3.22 { $\pm$ 0.00} -> 3.00 { $\pm$ 0.00} | Helped |
| add spatial propagation to the encoder (ablation) | SGP · PEMS-BAY | MAE (average): 1.62 { $\pm$ 0.00} -> 1.54 { $\pm$ 0.00} | Helped |
| use structured sparse decoder weights instead of a fully connected first layer (ablation) | SGP · METR-LA | MAE (average): 3.08 { $\pm$ 0.01} -> 3.00 { $\pm$ 0.00} | Helped |
| use structured sparse decoder weights instead of a fully connected first layer (ablation) | SGP · PEMS-BAY | MAE (average): 1.61 { $\pm$ 0.01} -> 1.54 { $\pm$ 0.00} | Helped |
| use spatial propagation scales beyond neighbor order K=1 (ablation) | SGP · METR-LA | MAE (average): 3.12 { $\pm$ 0.00} -> 3.00 { $\pm$ 0.00} | Helped |
| use spatial propagation scales beyond neighbor order K=1 (ablation) | SGP · PEMS-BAY | MAE (average): 1.59 { $\pm$ 0.01} -> 1.54 { $\pm$ 0.00} | Helped |
| precompute spatiotemporal embeddings and train the decoder with node-wise sampling | SGP · PV-US and CER-En large-scale benchmarks | Batch/s | Helped |
| precompute spatiotemporal embeddings and train the decoder with node-wise sampling | SGP · PV-US and CER-En large-scale benchmarks | Memory | Helped |
Closest papers by meaning
- Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling (ICML 2024 · similarity 0.92)
- Over-squashing in Spatiotemporal Graph Neural Networks (NeurIPS 2025 · similarity 0.90)
- Expand and Compress: Exploring Tuning Principles for Continual Spatio-Temporal Graph Forecasting (ICLR 2025 · similarity 0.89)
- Simple and Efficient Heterogeneous Temporal Graph Neural Network (NeurIPS 2025 · similarity 0.89)
- Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective (AAAI 2024 · similarity 0.89)
- Fully-Connected Spatial-Temporal Graph for Multivariate Time-Series Data (AAAI 2024 · similarity 0.88)
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.