Capsules TCN Network for Urban Computing and Intelligence in Urban Traffic Prediction
Dazhou Li,Chuan Lin,Wei Gao,Zeying Chen,Zeshen Wang,Guangqi Liu
DOI: https://doi.org/10.1155/2020/6896579
2020-06-04
Wireless Communications and Mobile Computing
Abstract:Predicting urban traffic is of great importance to smart city systems and public security; however, it is a very challenging task because of several dynamic and complex factors, such as patterns of urban geographical location, weather, seasons, and holidays. To tackle these challenges, we are stimulated by the deep-learning method proposed to unlock the power of knowledge from urban computing and proposed a deep-learning model based on neural network, entitled Capsules TCN Network, to predict the traffic flow in local areas of the city at once. Capsules TCN Network employs a Capsules Network and Temporal Convolutional Network as the basic unit to learn the spatial dependence, time dependence, and external factors of traffic flow prediction. In specific, we consider some particular scenarios that require accurate traffic flow prediction (e.g., smart transportation, business circle analysis, and traffic flow assessment) and propose a GAN-based superresolution reconstruction model. Extensive experiments were conducted based on real-world datasets to demonstrate the superiority of Capsules TCN Network beyond several state-of-the-art methods. Compared with HA, ARIMA, RNN, and LSTM classic methods, respectively, the method proposed in the paper achieved better results in the experimental verification.
computer science, information systems,telecommunications,engineering, electrical & electronic