Research map / Graph learning / Knowledge and temporal graphs
SIMBAD: Spatio-Temporal Traffic Forecasting Robust to Aperiodicity
Daniel Yoonhwan Lee, Seungwon Shin, Seunghoon Han, Sungsu Lim, Susik Yoon
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes SIMBAD, a spatio-temporal traffic forecasting approach for STGNNs aimed at robustness to aperiodic, event-driven fluctuations.
Problem
Traffic forecasting becomes unreliable when irregular patterns from unexpected events break periodicity, which STGNNs handle poorly.
Approach
Improves generalization on irregular traffic patterns via adaptive regulation (details truncated in the abstract) combined with a second mechanism not specified in the available text.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add masking matrix M that gates historical input by similarity to current traffic (ablation) | SIMBAD · PEMS08 | MAE: 13.10 -> 12.58 | Improved |
| add masking matrix M (ablation) | SIMBAD · PEMS03 | MAE: 15.94 -> 14.77 | Improved |
| add uncertainty estimation and loss L_u (ablation) | SIMBAD · PEMS07 | MAE: 18.23 -> 17.86 | Improved |
| add similarity-weighted adjacency matrices (ablation) | SIMBAD · PEMS03 | MAE: 15.65 -> 14.77 | Improved |
| use weekly historical data (ablation) | SIMBAD · all four PEMS datasets | MAE: 9.27% average MAE increase when removed; PEMS08 15.66% | Improved |
| use daily historical data (ablation) | SIMBAD · all four PEMS datasets | MAE: 1.86% average MAE increase when removed | Improved |
| plug temporal regulation module into DSTAGNN | DSTAGNN · PEMS08 | MAE: 16.19 -> 14.89 | Improved |
| plug temporal regulation module into STIDGCN | STIDGCN · PEMS08 | MAE: 13.71 -> 13.12 | Improved |
| plug temporal regulation module into LHMformer | LHMformer · PEMS04, PEMS07 | MAE: PEMS04 18.45 -> 18.83; PEMS07 19.27 -> 19.76 | Hurt |
| plug temporal regulation module into DSTAGNN | DSTAGNN · PEMS03 | MAE: 15.99 -> 16.44 | Hurt |
| raise softmax temperature tau (ablation) | SIMBAD · four PEMS datasets | MAE | Hurt |
| raise uncertainty regularization factor lambda_u (ablation) | SIMBAD · four PEMS datasets | MAE | Hurt |
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 1 of 1 shared result
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
- TESTAM: A Time-Enhanced Spatio-Temporal Attention Model with Mixture of Experts (ICLR 2024 · similarity 0.87)
- Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective (AAAI 2024 · similarity 0.87)
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.87)
- Spatiotemporal-aware Trend-Seasonality Decomposition Network for Traffic Flow Forecasting (AAAI 2025 · similarity 0.87)
- Decomposed Spatio-Temporal Mamba for Long-Term Traffic Prediction (AAAI 2025 · similarity 0.87)
- Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting (AAAI 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.
