Spatial-Temporal Synchronous Graph Convolutional Networks: A New Framework for Spatial-Temporal Network Data Forecasting

By Chao Song, Youfang Lin, Shengna Guo, Huaiyu Wan. AAAI Conference on Artificial Intelligence, 2020

1,748 citations recorded by the source corpus.

Spatial-temporal network data forecasting is of great importance in a huge amount of applications for traffic management and urban planning. However, the underlying complex spatial-temporal correlations and heterogeneities make this problem challenging. Existing methods usually use separate components to capture spatial and temporal correlations and ignore the heterogeneities in spatial-temporal data. In this paper, we propose a novel model, named Spatial-Temporal Synchronous Graph Convolutional Networks (STSGCN), for spatial-temporal network data forecasting. The model is able to effectively capture the complex localized spatial-temporal correlations through an elaborately designed spatial-temporal synchronous modeling mechanism. Meanwhile, multiple modules for different time periods are designed in the model to effectively capture the heterogeneities in localized spatial-temporal graphs. Extensive experiments are conducted on four real-world datasets, which demonstrates that our method achieves the state-of-the-art performance and consistently outperforms other baselines.

Subfields: Artificial Intelligence, Building and Construction, Transportation

View Paper: https://www.semanticscholar.org/paper/f420663fe8c69ed5ea5236201a1f4c734cd145a7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Building and Construction

In Transportation

Citations in the corpus

Learning Dynamics and Heterogeneity of Spatial-Temporal Graph Data for Traffic Forecasting (2021, methodology): “CNN-based STGNNs [9], [10], [11], [12] utilize 1D CNN along the temporal dimension to capture temporal features, and use graph convolutions to capture spatial features.”
Spatio-Temporal Self-Supervised Learning for Traffic Flow Prediction (2022, background): “Nevertheless, the parameter space differentiation strategy adopted in (Song et al. 2020; Li and Zhu 2021) assumes that the temporal heterogeneity is static across the entire time periods, which is not always held, e.g., evening traffic patterns can be significantly different for workdays and…”
PDFormer: Propagation Delay-aware Dynamic Long-range Transformer for Traffic Flow Prediction (2023, background): “…structure of traffic data (Li et al. 2018; Yu, Yin, and Zhu 2018), and thus GNN-based methods have been widely explored in traffic prediction (Wu et al. 2019; Song et al. 2020; Wu et al. 2020; Bai et al. 2020; Chen et al. 2020; Ye et al. 2021; Li and Zhu 2021; Fang et al. 2021; Choi et al. 2022).”
Graph Neural Network for Traffic Forecasting: A Survey (2021, background): “Problem Relevant Studies Road Traffic Flow [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62],”
Decoupled Dynamic Spatial-Temporal Graph Neural Network for Traffic Forecasting (2022, positive): “• Graph WaveNet [41]: Graph WaveNet stacks Gated TCN • ASTGCN [11]: ASTGCN combines the spatial-temporal attention • STSGCN [31]: STSGCN is proposed to effectively capture • GMAN [51]: GMAN is an attention-based model which • MTGNN [40]: MTGNN extends Graph WaveNet through • DGCRN [20]: DGCRN models”
Pre-training Enhanced Spatial-temporal Graph Neural Network for Multivariate Time Series Forecasting (2022, methodology): “Even compared with many newer works such as ASTGCN and STSGCN, their performance is still very promising.”
A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection (2023, methodology): “For example, STS-GCN [92] proposed spatial-temporal synchronous graph convolution, extending GCN to model spatial and temporal dependencies on localized spatial-temporal graphs.”
Spatial-Temporal Graph ODE Networks for Traffic Flow Forecasting (2021, methodology): “But STSGCN only concentrates on localized spatial-temporal correlations, and turns turtle in global dependencies.”
Dynamic Graph Convolutional Recurrent Network for Traffic Prediction: Benchmark and Solution (2021, background): “[18] propose spatialtemporal synchronous Graph Convolution Module to capture spatial correlations and temporal correlations simultaneously.”
Graph Neural Controlled Differential Equations for Traffic Forecasting (2021, background): “STSGCN (Song et al. 2020) utilizes multiple localized spatial-temporal subgraph modules to synchronously capture the localized spatial-temporal correlations directly.”
SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction (2021, methodology): “Most recent approaches: DCRNN [22], STGCN [41], ASTGCN [11], GraphWaveNet [39], STSGCN [35], AGCRN [3], LSGCN [15] and STFGNN [20] use graph neural networks to capture spatial relations while modeling temporal dependencies via conventional TCN or RNN/LSTM architectures.”
Spatio-Temporal Graph Neural Networks for Predictive Learning in Urban Computing: A Survey (2023, methodology): “In recent years, some literature focused on jointly modeling spatial-temporal dependencies based on 3D GCN [179], Spatio-Temporal Joint GCN (STJGCN) [180] and Spatio-Temporal SynchronousGCN(STSGCN)[53].”
Deep Learning on Traffic Prediction: Methods, Analysis, and Future Directions (2021, methodology): “GCN. Song et al. ﬁrst constructed a localized spatio-temporal graph that includes both temporal and spatial attributes, and then used the proposed spatial-based GCN method to model the spatio-temporal correlations simultaneously [56].”
Graph Neural Networks for Intelligent Transportation Systems: A Survey (2023, methodology): “The performance of their framework is evaluated using two real datasets and baseline models including ARIMA, LSTM, Hetero-ConvLSTM [204], STGCN [27], STG2seq [51], STSGCN [205], STDN [206], DFN [207], MTPSO [208].”
DSTAGNN: Dynamic Spatial-Temporal Aware Graph Neural Network for Traffic Flow Forecasting (2022, methodology): “On the basis of (Song et al., 2020), STFGCN (Li & Zhu, 2021) constructed a spatial-temporal fusion graph for tra ffi c prediction, in which the static adjacent graph was supplemented by the information derived from historic sequences, similar to DTW (Berndt & Cli ff ord, 1994).”