Research map / Video / Video segmentation and tracking
GSNet: Learning Spatial-Temporal Correlations from Geographical and Semantic Aspects for Traffic Accident Risk Forecasting
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| combine geographical and semantic spatial-temporal modeling | GSNet · NYC | RMSE: 7.7358 -> 7.5158 | Helped |
| combine geographical and semantic spatial-temporal modeling | GSNet · Chicago | Recall: 18.95% -> 20.69% | Helped |
| add temporal attention to the geographical and semantic modules (ablation) | GSNet · NYC | overall performance | Helped |
| replace dynamic feature fusion with simple concatenation (ablation) | GSNet · NYC | overall performance | Hurt |
| remove the geographical module (ablation) | GSNet · NYC | overall performance | Hurt |
| remove the semantic module (ablation) | GSNet · NYC | overall performance | Hurt |
| add multiple semantic similarity graphs instead of using a single graph (ablation) | GSNet · NYC | RMSE: 7.7135/6.8858 -> 7.5158/6.7206 | Helped |
| use only the risk similarity graph (ablation) | GSNet · NYC | RMSE / RMSE*: 7.5158/6.7206 -> 7.7135/6.8858 | Hurt |
| use only the road similarity graph (ablation) | GSNet · NYC | RMSE / RMSE*: 7.5158/6.7206 -> 7.7537/6.9000 | Hurt |
| use only the POI similarity graph (ablation) | GSNet · NYC | RMSE / RMSE*: 7.5158/6.7206 -> 7.8317/6.9748 | Hurt |
| weight samples by accident-risk level in the loss (ablation) | GSNet · NYC | RMSE / RMSE*: 10.2113/9.3556 -> 7.5158/6.7206 | Helped |
| weight samples by accident-risk level in the loss (ablation) | GSNet · NYC | Recall / Recall*: 34.22/34.46 -> 34.20/35.30 | Mixed |
Closest papers by meaning
- GeoPro-Net: Learning Interpretable Spatiotemporal Prediction Models Through Statistically-Guided Geo-Prototyping (AAAI 2025 · similarity 0.86)
- Spatiotemporal-aware Trend-Seasonality Decomposition Network for Traffic Flow Forecasting (AAAI 2025 · similarity 0.85)
- TraffiDent: A Dataset for Understanding the Interplay Between Traffic Dynamics and Incidents (NeurIPS 2025 · similarity 0.84)
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.84)
- Accident Anticipation via Temporal Occurrence Prediction (NeurIPS 2025 · similarity 0.84)
- Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting (AAAI 2024 · similarity 0.84)
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.