Research map / Adversarial robustness and security / Attacks on text and graphs
Practical Adversarial Attacks on Spatiotemporal Traffic Forecasting Models
Fan Liu, Hao Liu, Wenzhao Jiang
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 |
|---|---|---|---|
| Use TDNS to select victim nodes instead of random selection (ablation) | STPGD-TDNS · PeMS-BAY, grey-box | G-MAE: 5.9024 -> 6.1329 | Improved |
| Use STPGD rather than vanilla PGD noise on selected nodes (ablation) | STPGD-TDNS · PeMS-BAY, grey-box | G-MAE: 4.5969 -> 6.1329 | Improved |
| Use TDNS rather than random node selection (ablation) | STPGD-TDNS · PeMS-BAY, white-box | G-MAE: 6.1477 -> 6.4709 | Improved |
| Apply STPGD-TDNS instead of leaving the model un-attacked | GraphWaveNet · PeMS-BAY, grey-box | G-MAE: 1.975 -> 6.1329 | Hurt |
| Apply STMIM-TDNS instead of leaving the model un-attacked | GraphWaveNet · METR-LA, grey-box | G-MAE: 6.3504 -> 7.9381 | Hurt |
| Use STPGD-TDNS instead of the strongest listed baseline | GraphWaveNet · PeMS-BAY, grey-box | G-MAE: 5.5321 -> 6.1329 | Improved |
| Use PGD-TDNS rather than PGD-Centrality victim selection | GraphWaveNet · PeMS-BAY, black-box | G-MAE: 4.4859 -> 4.5636 | Improved |
| Train using adversarial examples generated by STPGD-TDNS rather than standard adversarial training | GraphWaveNet · PeMS-BAY, white-box defense under PGD-Random | G-MAE: 2.5436 (0.0249) -> 2.4695 (0.0036) | Improved |
| Increase the victim-node budget | STPGD-TDNS · PeMS-BAY | G-RMSE | Improved |
| Increase perturbation budget | STPGD-TDNS · PeMS-BAY | G-RMSE | Mixed |
| Increase batch size | STPGD-TDNS · PeMS-BAY | G-RMSE | Mixed |
| Increase the forecasting time interval | STPGD-TDNS · PeMS-BAY, 5 to 60 minutes | G-MAE: 3.9458 -> 6.1329 | Mixed |
Closest papers by meaning
- Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting (AAAI 2024 · similarity 0.87)
- Dynamic Spatial-Temporal Graph Convolutional Neural Networks for Traffic Forecasting (AAAI 2019 · similarity 0.86)
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.86)
- Spatiotemporal-aware Trend-Seasonality Decomposition Network for Traffic Flow Forecasting (AAAI 2025 · similarity 0.86)
- TESTAM: A Time-Enhanced Spatio-Temporal Attention Model with Mixture of Experts (ICLR 2024 · similarity 0.86)
- Decomposed Spatio-Temporal Mamba for Long-Term Traffic Prediction (AAAI 2025 · similarity 0.85)
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.
