Dynamic and Multi-faceted Spatio-temporal Deep Learning for Traffic Speed Forecasting

TLDR

Dynamic Graph Neural Networks are promising for traffic speed forecasting, yet most approaches rely on static adjacency matrices and ignore that spatial relationships and traffic volumes can change over time. This study aims to exploit dynamic, multi‑faceted spatio‑temporal characteristics in traffic data to enhance DGNN performance for speed prediction. We introduce a dynamic graph construction that learns time‑specific spatial dependencies, a graph convolution module that aggregates neighbor states via message passing on these dynamic adjacencies, and a fusion module that integrates traffic‑volume‑derived auxiliary states with speed‑derived primary states. Experiments on real‑world data show the method achieves state‑of‑the‑art accuracy and yields explicit, interpretable dynamic spatial relationships among road segments.

Abstract

Dynamic Graph Neural Networks (DGNNs) have become one of the most promising methods for traffic speed forecasting. However, when adapting DGNNs for traffic speed forecasting, existing approaches are usually built on a static adjacency matrix (no matter predefined or self-learned) to learn spatial relationships among different road segments, even if the impact of two road segments can be changeable dynamically during a day. Moreover, the future traffic speed cannot only be related with the current traffic speed, but also be affected by other factors such as traffic volumes. To this end, in this paper, we aim to explore these dynamic and multi-faceted spatio-temporal characteristics inherent in traffic data for further unleashing the power of DGNNs for better traffic speed forecasting. Specifically, we design a dynamic graph construction method to learn the time-specific spatial dependencies of road segments. Then, a dynamic graph convolution module is proposed to aggregate hidden states of neighbor nodes to focal nodes by message passing on the dynamic adjacency matrices. Moreover, a multi-faceted fusion module is provided to incorporate the auxiliary hidden states learned from traffic volumes with the primary hidden states learned from traffic speeds. Finally, experimental results on real-world data demonstrate that our method can not only achieve the state-of-the-art prediction performances, but also obtain the explicit and interpretable dynamic spatial relationships of road segments.