Trasnet: A lightweighting Spatio-temporal Attention Network for Traffic Flow Prediction

Accurate traffic prediction is a significant challenge in the development of intelligent transportation systems. The ability to capture crucial spatio-temporal traffic information plays a crucial role in the accuracy of predictions. In recent years, many complex neural networks have been proposed to address this issue. However, intricate network architectures have resulted in lower performance. In this paper, we introduce a spatio-temporal graph neural network, Trasnet, which employs spatio-temporal attention convolution to capture complex spatio-temporal information. Additionally, we propose a graph learning module to learn spatio-temporal dependencies from both global and local perspectives. We conduct extensive experiments on real-world datasets to validate the effectiveness of our approach.