Research map / Video / Video segmentation and tracking
AutoST: Towards the Universal Modeling of Spatio-temporal Sequences
Jianxin Li, Shuai Zhang, Hui Xiong, Haoyi Zhou
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 |
|---|---|---|---|
| search layer combination with AutoST2 instead of stacking identical layers (UniSTT) | UniST · METR-LA 60 min | MAE: 2.89 -> 2.78 | Improved |
| search layer combination with AutoST2 instead of stacking identical layers (UniSTT) | UniST · PEMS-08 60 min | MAE: 15.44 -> 14.72 | Improved |
| single-layer complex internal connection search (AutoST2) vs multi-layer search (AutoST1) | AutoST · METR-LA 60 min | MAE: 2.85 -> 2.78 | Improved |
| UniST layers vs strongest baseline STFGNN | UniST · METR-LA 60 min | MAE: 3.18 -> 2.91 | Improved |
| UniST vs GMAN, self-attention baseline | UniST · short-term forecasting | MAE decrease: 18.41% | Improved |
| UniST vs GMAN, self-attention baseline | UniST · medium-term forecasting | MAE decrease: 15.31% | Improved |
| UniST vs GMAN, self-attention baseline | UniST · long-term forecasting | MAE decrease: 13.82% | Improved |
| UniST vs STFGNN | UniST · short-term forecasting | MAE decrease: 12.06% | Improved |
| UniST vs STFGNN | UniST · medium-term forecasting | MAE decrease: 10.61% | Improved |
| UniST vs STFGNN | UniST · long-term forecasting | MAE decrease: 9.43% | Improved |
| use AutoST candidate layers with PAS search instead of AutoSTG (ablation) | AutoST2 · METR-LA 60 min | MAE: 3.47 -> 2.78 | Improved |
| use AutoST2 with random search instead of AutoSTG (ablation) | AutoST2 · PEMS-BAY 60 min | MAE: 1.92 -> 1.62 | Improved |
Closest papers by meaning
- UniSTD: Towards Unified Spatio-Temporal Learning across Diverse Disciplines (CVPR 2025 · similarity 0.86)
- Diffusion Transformers as Open-World Spatiotemporal Foundation Models (NeurIPS 2025 · similarity 0.84)
- FACTS: A Factored State-Space Framework for World Modelling (ICLR 2025 · similarity 0.84)
- Seahorse: A Unified Benchmarking Framework for Spatiotemporal Event Modeling (NeurIPS 2026 · similarity 0.83)
- STRAP: Spatio-Temporal Pattern Retrieval for Out-of-Distribution Generalization (NeurIPS 2025 · similarity 0.83)
- FlashST: A Simple and Universal Prompt-Tuning Framework for Traffic Prediction (ICML 2024 · similarity 0.83)
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.
