Attention-Based Spatial-Temporal Fusion Graph Network for Traffic Flow
Researchers propose a novel Attention-Based Spatial-Temporal Fusion Graph Convolution Network to address propagation delay in traffic flow prediction. Existing graph convolution models focus on spatial-temporal and semantic correlations in topological relationships but neglect information propagation delays between adjacent nodes. They also stack complex structures, increasing computational time and reducing timeliness. The new network aims to fuse spatial-temporal features while accounting for propagation delays, improving efficiency and accuracy. The paper is available on arXiv.
Key facts
- The paper is titled 'Eliminating Propagation Delay: Attention-Based Spatial-Temporal Fusion Graph Convolution Network for Traffic Flow Prediction'.
- It is published on arXiv with ID 2607.24885.
- The model addresses two problems: neglect of propagation delays and high computational time.
- The network is called Attention-Based Spatial-Temporal Fusion Graph Convolution Network.
- Traffic flow prediction is crucial for optimizing transportation and urban mobility.
- Existing models focus on static spatial dependencies and spatial-temporal relationships.
- The proposed model aims to improve timeliness and accuracy.
- The paper is categorized as a cross-type announcement.
Entities
Institutions
- arXiv