Research map / Video / Video segmentation and tracking
GPT-ST: Generative Pre-Training of Spatio-Temporal Graph Neural Networks
Zhonghang Li, Lianghao Xia, Yong Xu, Chao Huang
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 |
|---|---|---|---|
| add GPT-ST pre-trained representations to STGCN | STGCN · PEMS08 | MAE: 17.85 -> 16.24 | Improved |
| add GPT-ST to TGCN | TGCN · NYC Taxi | MAE: 9.73 -> 6.88 | Improved |
| add GPT-ST to GWN | GWN · METR-LA | MAE: 3.15 -> 3.02 | Improved |
| add GPT-ST to advanced baseline MSDR vs classical STGCN | MSDR · PEMS08 | MAE: 16.47 -> 15.92 | Improved |
| add GPT-ST (short-term pretraining) vs STEP (long-term pretraining) on GWN* | GWN* · NYC Taxi | MAE: 5.33 -> 5.07 | Improved |
| add GPT-ST to GWN*, which already uses long-term input | GWN* · PEMS08 | MAE: 13.97 -> 13.96 | No clear effect |
| remove hypergraph capsule clustering network (-C) (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
| add customized parameter learner (vs removing it, -P) (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
| add cross-cluster relation learning (vs removing it, -T) (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
| adaptive cluster-aware mask instead of random mask (ratio 0.25 or 0.75) (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
| adaptive mask instead of GraphMAE or AdaMAE masks (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
| mask-reconstruction pre-training instead of DGI or GraphCL pre-training (ablation) | GPT-ST with GWN · METR-LA and NYC Taxi | MAE/RMSE/MAPE | Improved |
Closest papers by meaning
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.90)
- Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective (AAAI 2024 · similarity 0.89)
- Efficient Unlearning for Spatio-temporal Graph (Student Abstract) (AAAI 2025 · similarity 0.88)
- Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting (AAAI 2024 · similarity 0.88)
- STRAP: Spatio-Temporal Pattern Retrieval for Out-of-Distribution Generalization (NeurIPS 2025 · similarity 0.88)
- Spatio-Temporal Few-Shot Learning via Diffusive Neural Network Generation (ICLR 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.
