STEGNN: Spatial-Temporal Embedding Graph Neural Networks for Road Network Forecasting

Si, Jiaqi; Gan, Xinbiao; Xiao, Tiaojie; Yang, Bo; Dong, Dezun; Pang, Zhengbin

doi:10.1109/icpads56603.2022.00112

Cited by 1 publication

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…b) Baseline: Ten models were used as the baseline for this experiment. Among them, ARIMA ( 6 ) and SVR ( 8 ) are the traditional methods; FC-LSTM ( 11 ) is a classic deep learning method; GWNET ( 27 ), diffusion convolutional recurrent neural network (DCRNN) ( 24 ), STGCN ( 14 ), STSGCN ( 25 ), StemGNN ( 26 ), STDSGCN ( 28 ), STEGNN ( 29 ) are typical GNN-based deep learning methods. Among them, STEGNN and STDSGCN are the latest excellent traffic prediction methods.…”

Section: Methodsmentioning

confidence: 99%

“…Among them, STEGNN and STDSGCN are the latest excellent traffic prediction methods. The results of StemGNN are re-implemented by us, and other baseline results are cited from Zhang et al ( 28 ) and Si et al ( 29 ).…”

Section: Methodsmentioning

confidence: 99%

“…For instance, spatial-temporal dynamic semantic graph neural network (STDSGCN) captured two different semantic features through Pearson correlation coefficient and dynamic time warping algorithm to construct two semantic adjacency matrices ( 28 ). Spatial-temporal embedding graph neural network (STEGNN) employs cosine similarity to generate high-quality temporal maps to fill the gaps in the traffic data ( 29 ).…”

mentioning

confidence: 99%

See 2 more Smart Citations

Spatio-Temporal Graph Neural Network for Traffic Prediction Based on Adaptive Neighborhood Selection

Sun,

Tang,

et al. 2023

Transportation Research Record: Journal of the Transportation R

View full text Add to dashboard Cite

Traffic prediction is critical to intelligent transportation and smart cities. The prediction performance of many existing traffic prediction models is limited by the fixed original graph structure and inappropriate spatio-temporal dependency extraction. For this situation, this paper proposes a spatio-temporal graph neural network based on adaptive neighborhood selection (STGNN-ANS). To obtain more flexible graph structures, STGNN-ANS designs a neighbor selection mechanism to generate a new graph structure by filtering inappropriate neighbors. To further capture the spatio-temporal dependence of traffic data, a spatio-temporal serial module of STGNN-ANS adopts the bidirectional learning manner of bidirectional long short-term memory (BiLSTM) and the graph convolution network (GCN) enhanced by self-attention mechanism to reach excellent prediction accuracy in both short-range and long-range scenarios. In this paper, a new baseline comprehensive comparison metric (BCCM) is invented to cope with the complexity in the comparative analysis of large numbers of experimental results. Many experiments have been performed on four real-world traffic datasets, and the results show that the comprehensive prediction performance of STGNN-ANS is better than previous models.

show abstract