CS Events

Abstract: With the increasing population density in the city, people rely more on urban mobility to access services such as education and entertainment. Traffic accident prediction is a crucial problem for public safety, emergency treatment, and urban mobility management. Existing works use rich data collected from city infrastructures to achieve encouraging performance based on various machine learning techniques but cannot achieve a good performance in situations with limited data (i.e., data scarcity). Recent developments in transfer learning bring a new opportunity to solve the data scarcity problem. In this work, we design a novel cross-city transfer learning framework, named CARPG, to predict traffic accidents in data-scarce cities. We address the unique challenge of predicting traffic accidents because of its two fundamental characteristics, i.e., spatial heterogeneity and inherent rareness, which result in the biased performance of the state-of-the-art methods. Specifically, we jointly learn the spatial region representations for both source and target cities with an inter-city global graph knowledge transfer process. Further, we design an efficient attention-based parameter-generating mechanism to ensure that only relevant patterns are transferred and fine-tuned for each target region during the knowledge transfer process. We conduct extensive experiments on three real-world datasets, and the evaluation results demonstrate the superiority of our framework compared with state-of-the-art baseline models.