Smart City Gnosys

Smart city article details

Title	Mfstgn: A Multi-Scale Spatial-Temporal Fusion Graph Network For Traffic Prediction
ID_Doc	36944
Authors	Tian R.; Wang C.; Hu J.; Ma Z.
Year	2023
Published	Applied Intelligence, 53, 19
DOI	http://dx.doi.org/10.1007/s10489-023-04703-4
Abstract	Traffic prediction is an important link in building a smart city, accurate prediction of future traffic conditions can provide a reference for people to travel and facilitate the work of relevant departments, and is vital for urban applications. Nevertheless, due to the cyclical nature of traffic conditions and uncertain factors, as well as the spatial heterogeneity in the traffic network, it is difficult to model the dynamic temporal and spatial correlation, so traffic prediction is facing severe challenges. Existing methods usually use a dynamic graph to model and analyze traffic conditions at a certain moment, lacking consideration of spatial heterogeneity and trend changes in traffic flow, so prediction results are often unsatisfactory. To address the above challenges, we propose a Multi-scale Spatial-temporal Fusion Graph Network for Traffic Prediction framework (MFSTGN). Specifically, we design a spatial-temporal graph convolution network module (STGCN) that dynamically models spatial-temporal correlations while preserving the inherent structure of the traffic network, and describes the trend variation of traffic flows through a kind of trend graph convolution, while modeling the spatial heterogeneity of the traffic network using spatial-temporal embedding. STGCN allows MFSTGN to enjoy a perceptual field that facilitates long-term prediction. In addition, we have developed a Gated Attention mechanism that adaptively fuses cyclical dependence and trend dependence, enabling MFSTGN to enjoy multiple scale information. Experimental results on four datasets for two types of traffic prediction tasks show that MFSTGN outperforms the state-of-the-art baseline. © 2023, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.
Author Keywords	Attention mechanism; Graph convolution network; Periodism; Spatial heterogeneity; Traffic prediction

Similar Articles

Id	Similarity	Authors	Title	Published
38047	0.959	Yang S.; Wu Q.; Wang Y.; Zhou Z.	Mstdfgrn: A Multi-View Spatio-Temporal Dynamic Fusion Graph Recurrent Network For Traffic Flow Prediction	Computers and Electrical Engineering, 123 (2025)
53044	0.956	Meng X.; Xie W.; Cui J.	Stmgfn: Spatio-Temporal Multi-Graph Fusion Network For Traffic Flow Prediction	Lecture Notes in Computer Science, 15291 LNCS (2025)
21123	0.945	Hu J.; Lin X.; Wang C.	Dstgcn: Dynamic Spatial-Temporal Graph Convolutional Network For Traffic Prediction	IEEE Sensors Journal, 22, 13 (2022)
38415	0.938	Zhao W.; Zhang S.; Zhou B.; Wang B.	Multi-Spatio-Temporal Fusion Graph Recurrent Network For Traffic Forecasting	Engineering Applications of Artificial Intelligence, 124 (2023)
2033	0.938	Xu J.; Li Y.; Lu W.; Wu S.; Li Y.	A Heterogeneous Traffic Spatio-Temporal Graph Convolution Model For Traffic Prediction	Physica A: Statistical Mechanics and its Applications, 641 (2024)
20800	0.938	Diao Z.; Wang X.; Zhang D.; Xie G.; Chen J.; Pei C.; Meng X.; Xie K.; Zhang G.	Dmstg: Dynamic Multiview Spatio-Temporal Networks For Traffic Forecasting	IEEE Transactions on Mobile Computing, 23, 6 (2024)
37213	0.938	Meng X.; Xie W.; Cui J.	Mmgcrn: Multimodal And Multiview Graph Convolutional Recurrent Network For Traffic Prediction	Proceedings of the International Conference on Software Engineering and Knowledge Engineering, SEKE (2024)
28936	0.936	Mao H.; Sun Z.; Qin N.	Heterogeneous Augmentation Based Spatio-Temporal Graph Convolutional Network For Traffic Forecasting	Proceedings of the International Joint Conference on Neural Networks (2024)
37214	0.935	Yu W.; Wu S.; Huang M.	Mmgfra: A Multiscale Multigraph Learning Framework For Traffic Prediction In Smart Cities	Earth Science Informatics, 16, 3 (2023)
4883	0.934	Cao S.; Wu L.; Wu J.; Wu D.; Li Q.	A Spatio-Temporal Sequence-To-Sequence Network For Traffic Flow Prediction	Information Sciences, 610 (2022)