BiST: A Lightweight and Efficient Bi-directional Model for Spatiotemporal Prediction
Summary: Introduces BiST, a bi-directional spatiotemporal predictor that injects label information via a spatiotemporal dynamic theory: MLP-only forward path plus a decoupled residual backward correction smoothed by diffusion. Matches SOTA (+8.13%) with far lower compute/memory. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
No non-self incoming citations found for this paper in this database.
Authors
- 1. Jiaming Ma
- 2. Binwu Wang
- 3. Pengkun Wang
- 4. Zhengyang Zhou
- 5. Xu Wang
- 6. Yang Wang
Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 2 of 2 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 4,206 | Decoupled Dynamic Spatial-Temporal Graph Neural Network for Traffic Forecasting | 2022 | VLDB | 6.3595566e-05 |
| 5,438 | Multiple Time Series Forecasting with Dynamic Graph Modeling | 2024 | VLDB | 5.5033018e-05 |
Previous
Page 1 / 1
Next