Toward an Integrated Cross-Urban Accident Prevention System: A Multi-Task Spatial-Temporal Learning Framework for Urban Safety Management

2026-01-09

arXiv.org

AI Summary

This study addresses the challenges posed by heterogeneity, inconsistent reporting standards, and sparse noise in cross-city traffic accident data, which hinder the development of unified accident prevention systems. To overcome these issues, the authors propose MLA-STNet, a novel framework that formulates cross-city accident risk prediction as a multi-task learning problem. The approach innovatively integrates Mamba with attention mechanisms through two dedicated modules—spatiotemporal-geographic (STG-MA) and spatiotemporal-semantic (STS-MA)—enabling joint training across multiple cities under shared parameters while preserving both global consistency and city-specific characteristics. Experiments on real-world datasets from New York and Chicago demonstrate that MLA-STNet outperforms state-of-the-art models, achieving a 6% reduction in RMSE, an 8% increase in Recall, and a 5% improvement in mAP. Moreover, the model exhibits exceptional robustness, with performance variation of less than 1% under 50% input noise.

Technology Category

• Machine Learning: Multimodal Learning
• Natural Language Processing: Safety and Robustness
• Computer Vision: Multi-modal Vision

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

The development of a cross-city accident prevention system is particularly challenging due to the heterogeneity, inconsistent reporting, and inherently clustered, sparse, cyclical, and noisy nature of urban accident data. These intrinsic data properties, combined with fragmented governance and incompatible reporting standards, have long hindered the creation of an integrated, cross-city accident prevention framework. To address this gap, we propose the Mamba Local-ttention Spatial-Temporal Network MLA-STNet, a unified system that formulates accident risk prediction as a multi-task learning problem across multiple cities. MLA-STNet integrates two complementary modules: (i)the Spatio-Temporal Geographical Mamba-Attention (STG-MA), which suppresses unstable spatio-temporal fluctuations and strengthens long-range temporal dependencies; and (ii) the Spatio-Temporal Semantic Mamba-Attention (STS-MA), which mitigates cross-city heterogeneity through a shared-parameter design that jointly trains all cities while preserving individual semantic representation spaces. We validate the proposed framework through 75 experiments under two forecasting scenarios, full-day and high-frequency accident periods, using real-world datasets from New York City and Chicago. Compared with the state-of-the-art baselines, MLA-STNet achieves up to 6% lower RMSE, 8% higher Recall, and 5% higher MAP, while maintaining less than 1% performance variation under 50% input noise. These results demonstrate that MLA-STNet effectively unifies heterogeneous urban datasets within a scalable, robust, and interpretable Cross-City Accident Prevention System, paving the way for coordinated and data-driven urban safety management.

Problem

• cross-city accident prevention
• urban accident data heterogeneity
• spatio-temporal learning
• integrated safety management
• multi-task learning

Innovation

• multi-task learning
• Mamba-Attention
• cross-city accident prediction
• spatio-temporal modeling
• urban safety management