Urban Traffic Prediction from Spatio-Temporal Data Using Deep Meta Learning

By Zheyi Pan, Yuxuan Liang, Weifeng Wang, Yong Yu, Yu Zheng, Junbo Zhang. Knowledge Discovery and Data Mining, 2019

632 citations recorded by the source corpus.

Predicting urban traffic is of great importance to intelligent transportation systems and public safety, yet is very challenging because of two aspects: 1) complex spatio-temporal correlations of urban traffic, including spatial correlations between locations along with temporal correlations among timestamps; 2) diversity of such spatio-temporal correlations, which vary from location to location and depend on the surrounding geographical information, e.g., points of interests and road networks. To tackle these challenges, we proposed a deep-meta-learning based model, entitled ST-MetaNet, to collectively predict traffic in all location at once. ST-MetaNet employs a sequence-to-sequence architecture, consisting of an encoder to learn historical information and a decoder to make predictions step by step. In specific, the encoder and decoder have the same network structure, consisting of a recurrent neural network to encode the traffic, a meta graph attention network to capture diverse spatial correlations, and a meta recurrent neural network to consider diverse temporal correlations. Extensive experiments were conducted based on two real-world datasets to illustrate the effectiveness of ST-MetaNet beyond several state-of-the-art methods.

Subfields: Building and Construction, Transportation

View Paper: https://www.semanticscholar.org/paper/06b8e82542d1873928d007548a23d3b77daa11f8

How researchers see this paper

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

In Building and Construction

1. Haitao Yuan
2. Guoliang Li
3. Shirui Pan
4. Junbo Zhang
5. Zhifeng Bao
6. Yu Zheng
7. Nizar Bouguila
8. Saeed Rahmani

In Transportation

1. Shirui Pan
2. Guoliang Li
3. Haitao Yuan
4. Junbo Zhang
5. Zhifeng Bao
6. Yu Zheng
7. Chengqi Zhang
8. Guodong Long

Citations in the corpus

• Deep Learning on Traffic Prediction: Methods, Analysis, and Future Directions (2021, methodology): “To model these properties, the spatial attention mechanism is often used to adaptively capture the correlations between regions in the road network ( [58]–[66] ).”
• A Survey on Modern Deep Neural Network for Traffic Prediction: Trends, Methods and Challenges (2020, methodology): “[57], [71], [69], [72] and [73], and has demonstrated state-ofthe-art performance.”
• Process (2020, background): “Most of the previous works [10, 11, 12, 13, 14, 15, 16, 17, 18, 19] on traffic modeling focus on predicting speed, volume, density and travel time, which have achieved remarkable success in this field.”
• Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks (2020, background): “• ST-MetaNet: A sequence-to-sequence architecture, which employs meta networks to generate parameters [15].”
• Dynamic Graph Convolutional Recurrent Network for Traffic Prediction: Benchmark and Solution (2021, methodology): “[7] uses meta learning to generate the parameters of the graph attention network (GAT [29]).”
• A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection (2023, background): “In contrast, STGCN (1 st ) (a second version of STGCN [54]) and ST-MetaNet [76] modeled spatial dependencies with GCN [119] and GAT [120] to aggregate information from adjacent time series.”
• Spatial-Temporal Transformer Networks for Traffic Flow Forecasting (2020, methodology): “Compared with other attention-based dynamical spatial models such as GraphWaveNet [20] which limited their dynamical spatial dependencies in k -nearest neighbor nodes, we can capture long-range spatial dependencies beyond local nodes, thus can better to capture sudden changes, resultsing our…”
• Spatial-Temporal Graph ODE Networks for Traffic Flow Forecasting (2021, methodology): “Most of them combine GNNs with RNNs to obtain spatial representations and temporal representation respectively [24, 35], and multiple works improve recurrent structure with convolution structure for better training stability and efficiency [9, 34].”
• Spatio-Temporal Graph Neural Networks for Predictive Learning in Urban Computing: A Survey (2023, neutral): “Overview of ST-MetaNet [197].”
• PDFormer: Propagation Delay-aware Dynamic Long-range Transformer for Traffic Flow Prediction (2023, background): “We verify the performance of PDFormer on six real-world public traffic datasets, including three graph-based highway traffic datasets, i.e., PeMS04, PeMS07, PeMS08 (Song et al. 2020), and three grid-based citywide traffic datasets, i.e., NYTaxi (Liu et al. 2021a), CHBike (Wang et al. 2021), TDrive (”
• Spatio-Temporal Self-Supervised Learning for Traffic Flow Prediction (2022, methodology): “In addition, meta-learning has been used in recent approaches (Pan et al. 2019a; Ye et al. 2022) to con-a sider the difference of region traffic distributions.”
• Graph Neural Networks for Intelligent Transportation Systems: A Survey (2023, methodology): “Inspired by graph attention networks and encoder-decoders, Pan et al. [33] proposed ST-MetaNet, a spatial-temporal meta graph attention network for multi-step traffic forecasting.”
• Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting (2020, background): “1), they may appear similar, dissimilar, and even contradictory owning to the distinct attributes across a variety of data sources [7, 13].”
• A Survey of Traffic Prediction: from Spatio-Temporal Data to Intelligent Transportation (2021, methodology): “At last, the meta-learning method is also applied for capturing the dynamic dependency among traffic flow data [164].”
• Graph Neural Network for Traffic Forecasting: A Survey (2021, neutral): “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],”