Abstract:Abstract Traffic accidents have become one of the biggest public health safety matters, which has raised many concerns from citizens and city managers. Accurate traffic accident prediction can not only assist the government in making decisions in advance but also enhance public trust in public safety. Conventional spatio-temporal prediction models, limited by the skewed distributions and sparse labels of traffic accident occurrence, are prone to overfitting. Inspired by hypergraph learning and self-supervised learning, this paper suggests a sparse spatio-temporal dynamic hypergraph learning (SST-DHL) framework to capture the higher-order dependencies in sparse traffic accidents. Specifically, a multi-view spatio-temporal convolution block is employed first to capture the local spatio-temporal correlation and inherent semantics of traffic accidents. Then we propose a cross-regional dynamic hypergraph learning model to capture global spatio-temporal dependencies beneath the entire urban landscape. In addition, a two-supervised self-learning paradigm is intended to strengthen the representation of sparse traffic occurrences by regional self-identification, which can capture local and global spatio-temporal traffic patterns. The proposed model is applicable to most sparse datasets for traffic forecasts. Extensive experiments was conducted on two heterogeneous accident datasets from New York City and London, and the results shows an average improvements of 7.21%-23.09% at different sparsity levels compared to the optimal baselines. More importantly, the proposed SST-DHL improves the interpretability of model results, which demonstrates that hypergraph learning can efficiently capture the complex higher-order spatio-temporal dependencies among different traffic accident instances.

Accident Risk Prediction based on Heterogeneous Sparse Data: New Dataset and Insights

Enhancing Accident Risk Prediction with Novel Data and Findings from Heterogeneous Sparse Sources

Real-Time Crash Risk Prediction using Long Short-Term Memory Recurrent Neural Network

A Countrywide Traffic Accident Dataset

Learning Deep Representation from Big and Heterogeneous Data for Traffic Accident Inference.

A Deep Learning Approach to the Prediction of Short-term Traffic Accident Risk

Urban Traffic Accident Risk Prediction Revisited: Regionality, Proximity, Similarity and Sparsity

Traffic Accident Prediction Based on Deep Spatio-Temporal Analysis.

Accident Impact Prediction based on a deep convolutional and recurrent neural network model

A Deep Learning Approach to the Citywide Traffic Accident Risk Prediction

Sparse Spatio-Temporal Dynamic Hypergraph Learning for Traffic Accident Prediction

Stack ResNet For Short-term Accident Risk Prediction Leveraging Cross-domain Data

Learning spatial patterns and temporal dependencies for traffic accident severity prediction: A deep learning approach

Foresee Urban Sparse Traffic Accidents: A Spatiotemporal Multi-Granularity Perspective

Advancing urban traffic accident forecasting through sparse spatio-temporal dynamic learning

ESCORT: Fine-Grained Urban Crime Risk Inference Leveraging Heterogeneous Open Data

CrashFormer: A Multimodal Architecture to Predict the Risk of Crash

Applying Deep Learning to Detect Traffic Accidents in Real Time Using Spatiotemporal Sequential Data

Road Car Accident Prediction Using a Machine-Learning-Enabled Data Analysis

HintNet: Hierarchical Knowledge Transfer Networks for Traffic Accident Forecasting on Heterogeneous Spatio-Temporal Data

Towards Big Data Analytics and Mining for UK Traffic Accident Analysis, Visualization & Prediction.