Research map / Graph learning / Knowledge and temporal graphs
A Retrieval Augmented Spatio-Temporal Framework for Traffic Prediction
Weilin Ruan, Xilin Dang, Ziyu Zhou, Sisuo Lyu, Yuxuan Liang
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 the fusion query (query generator) (ablation) | RAST · PEMS07 | MAE: 24.53 -> 19.52 | Improved |
| add ST-Retriever (ablation) | RAST · PEMS07 | MAE: 21.71 -> 19.52 | Improved |
| add spatial encoder (ablation) | RAST · PEMS07 | MAE: 22.88 -> 19.52 | Improved |
| add temporal encoder (ablation) | RAST · PEMS07 | MAE: 23.66 -> 19.52 | Improved |
| add retrieval embedding to query embedding (cross-attention fusion) (ablation) | RAST · PEMS07 | MAE: 23.59 -> 19.52 | Improved |
| fuse query with retrieval embedding instead of retrieval embedding alone (ablation) | RAST · PEMS07 | MAE, RMSE, MAPE: MAE 19.38 vs 19.52; MAPE 13.32 vs 8.23 | Mixed |
| add MLP predictor (ablation) | RAST · PEMS07 | MAE: 21.67 -> 19.52 | Improved |
| set retrieval store update interval to 10 epochs rather than 20 or 50 (ablation) | RAST · PEMS04 | MAE, MAPE | Improved |
| increase query embedding dimension from 32 to 128 (ablation) | RAST · PEMS04 | MAE | Improved |
| use more than 3 encoder layers (ablation) | RAST · PEMS04 | MAE: MAE 19.19 at 3 layers | No clear effect |
| use RAST instead of STGCN for efficiency | large-scale traffic forecasting · GBA | training time per epoch (s): 852.57 -> 154.08 | Improved |
| RAST versus baselines on PEMS03 | RAST · PEMS03 | MAE, RMSE, MAPE: MAE 15.20 vs DSTAGNN 15.57 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DSTAGNN — DSTAGNN: Dynamic Spatial-Temporal Aware Graph Neural Network for Traffic Flow Forecasting (ICML 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.89)
- FlashST: A Simple and Universal Prompt-Tuning Framework for Traffic Prediction (ICML 2024 · similarity 0.88)
- Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting (AAAI 2024 · similarity 0.88)
- Spatiotemporal-aware Trend-Seasonality Decomposition Network for Traffic Flow Forecasting (AAAI 2025 · similarity 0.88)
- Decomposed Spatio-Temporal Mamba for Long-Term Traffic Prediction (AAAI 2025 · similarity 0.87)
- TESTAM: A Time-Enhanced Spatio-Temporal Attention Model with Mixture of Experts (ICLR 2024 · similarity 0.87)
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.
