Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Geographical Information Alignment Boosts Traffic Analysis via Transpose Cross-attention

Xiangyu Jiang, Xiwen Chen, Hao Wang, Abolfazl Razi

TL;DR

The paper addresses the limitation of GNNs in traffic analysis that overlook absolute geographic coordinates. It proposes Geographic Information Alignment (GIA), a plug-in module that fuses node features with geographic positions using a Transpose Cross-attention mechanism, dramatically reducing computation from $ ext{O}(N^2)$ to $ ext{O}(D_n^2)$. Across large city-scale datasets and multiple baseline GNNs, GIA yields consistent gains in accident occurrence (up to $ ext{F1}$ improvement of 10.9%) and severity prediction (up to $ ext{AUC}$ improvement of 4.8%), validating its effectiveness and efficiency. The approach is demonstrated to be broadly compatible with existing GNN architectures and provides a practical, scalable means to leverage geographic information for traffic safety analytics.

Abstract

Traffic accident prediction is crucial for enhancing road safety and mitigating congestion, and recent Graph Neural Networks (GNNs) have shown promise in modeling the inherent graph-based traffic data. However, existing GNN- based approaches often overlook or do not explicitly exploit geographic position information, which often plays a critical role in understanding spatial dependencies. This is also aligned with our observation, where accident locations are often highly relevant. To address this issue, we propose a plug-in-and-play module for common GNN frameworks, termed Geographic Information Alignment (GIA). This module can efficiently fuse the node feature and geographic position information through a novel Transpose Cross-attention mechanism. Due to the large number of nodes for traffic data, the conventional cross-attention mechanism performing the node-wise alignment may be infeasible in computation-limited resources. Instead, we take the transpose operation for Query, Key, and Value in the Cross-attention mechanism, which substantially reduces the computation cost while maintaining sufficient information. Experimental results for both traffic occurrence prediction and severity prediction (severity levels based on the interval of recorded crash counts) on large-scale city-wise datasets confirm the effectiveness of our proposed method. For example, our method can obtain gains ranging from 1.3% to 10.9% in F1 score and 0.3% to 4.8% in AUC.

Geographical Information Alignment Boosts Traffic Analysis via Transpose Cross-attention

TL;DR

The paper addresses the limitation of GNNs in traffic analysis that overlook absolute geographic coordinates. It proposes Geographic Information Alignment (GIA), a plug-in module that fuses node features with geographic positions using a Transpose Cross-attention mechanism, dramatically reducing computation from to . Across large city-scale datasets and multiple baseline GNNs, GIA yields consistent gains in accident occurrence (up to improvement of 10.9%) and severity prediction (up to improvement of 4.8%), validating its effectiveness and efficiency. The approach is demonstrated to be broadly compatible with existing GNN architectures and provides a practical, scalable means to leverage geographic information for traffic safety analytics.

Abstract

Traffic accident prediction is crucial for enhancing road safety and mitigating congestion, and recent Graph Neural Networks (GNNs) have shown promise in modeling the inherent graph-based traffic data. However, existing GNN- based approaches often overlook or do not explicitly exploit geographic position information, which often plays a critical role in understanding spatial dependencies. This is also aligned with our observation, where accident locations are often highly relevant. To address this issue, we propose a plug-in-and-play module for common GNN frameworks, termed Geographic Information Alignment (GIA). This module can efficiently fuse the node feature and geographic position information through a novel Transpose Cross-attention mechanism. Due to the large number of nodes for traffic data, the conventional cross-attention mechanism performing the node-wise alignment may be infeasible in computation-limited resources. Instead, we take the transpose operation for Query, Key, and Value in the Cross-attention mechanism, which substantially reduces the computation cost while maintaining sufficient information. Experimental results for both traffic occurrence prediction and severity prediction (severity levels based on the interval of recorded crash counts) on large-scale city-wise datasets confirm the effectiveness of our proposed method. For example, our method can obtain gains ranging from 1.3% to 10.9% in F1 score and 0.3% to 4.8% in AUC.

Paper Structure

This paper contains 8 sections, 3 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. RELATED WORK
  3. Method
  4. Problem Formulation
  5. Proposed Method
  6. Position Encoding
  7. Experiments
  8. Conclusion

Figures (3)

  • Figure 1: Top: Traffic Accident Map in Los Angeles shows a strong correlation between traffic accidents and location information. Positive denotes the crash/accident occurrence. We mark some exemplary accident clusters. Bottom: The proposed Geographic Information Alignment (GIA) through the Transpose Cross-attention (TCA) mechanism.
  • Figure 2: Comparison of different fusion strategies. Left: Simply take addition, Middle: The conventional Cross-attention, and Right: Our proposed Transpose Cross-attention (TCA) for feature-alignment.
  • Figure 3: Comparison of traffic accident predictions in the Houston area using the GEN model(left), GEN-GIA model(middle), and the ground truth(right). The GEN-GIA model shows improved accuracy in predicting high-accident areas by integrating positional encoding.