Dynamic Graph Convolutional Recurrent Network for Traffic Prediction: Benchmark and Solution

Abstract

Traffic prediction is the cornerstone of intelligent transportation system. Accurate traffic forecasting is essential for the applications of smart cities, i.e., intelligent traffic management and urban planning. Although various methods are proposed for spatio-temporal modeling, they ignore the dynamic characteristics of correlations among locations on road network. Meanwhile, most Recurrent Neural Network (RNN) based works are not efficient enough due to their recurrent operations. Additionally, there is a severe lack of fair comparison among different methods on the same datasets. To address the above challenges, in this paper, we propose a novel traffic prediction framework, named Dynamic Graph Convolutional Recurrent Network (DGCRN). In DGCRN, hyper-networks are designed to leverage and extract dynamic characteristics from node attributes, while the parameters of dynamic filters are generated at each time step. We filter the node embeddings and then use them to generate dynamic graph, which is integrated with pre-defined static graph. As far as we know, we are first to employ a generation method to model fine topology of dynamic graph at each time step. Further, to enhance efficiency and performance, we employ a training strategy for DGCRN by restricting the iteration number of decoder during forward and backward propagation. Finally, a reproducible standardized benchmark and a brand new representative traffic dataset are opened for fair comparison and further research. Extensive experiments on three datasets demonstrate that our model outperforms 15 baselines consistently. Source codes are available at https://github.com/tsinghua-fib-lab/Traffic-Benchmark.

STONet: A Neural-Operator-Driven Spatio-temporal Network

“…Graph Spatio-temporal networks. Spatio-temporal forecasting networks are mostly graph-based thanks to their ability to learn representations of spatially-irregular distributed signals, such as traffic flows recorded by sensors (Yu et al, 2018;Li et al, 2018;Guo et al, 2019b;Bai et al, 2020;Zhao et al, 2020;Li et al, 2021a). These works usually regard signals' location as nodes, and establish graphs to describe the nodes' dependency according to their spatial distance.…”

“…Studying the spatio-temporal patterns of physical quantities is of great scientific interest. Significant progress has been achieved thanks to immense research efforts in deep neural networks for modeling the spatial dependency and temporal dynamics (Shi et al, 2015;Guo et al, 2019b;Zhao et al, 2020;Bai et al, 2020;Li et al, 2021a). Most of them are established for spatially-discrete nodes, such as sensors' signals of traffic flow located on discretized roads, and graph neural networks (GNNs) are usually employed to handle signals with spatially-irregular distribution and establish dependency between nodes (Seo et al, 2016;Yu et al, 2018;Li et al, 2018;Rozemberczki et al, 2021).…”

ST-CRMF: Compensated Residual Matrix Factorization with Spatial-Temporal Regularization for Graph-Based Time Series Forecasting

“…Graph WaveNet [ 4 ] integrated GCN with dilated causal convolution for saving the computational cost of dealing with long sequences. Moreover, Li et al [ 33 ] built the dynamic graph convolutional recurrent network (DGCRN) which generated the dynamic adjacency matrices and extracted features from node attributes. Bai et al [ 3 ] established the adaptive GCRN (AGCRN) for automatically capturing the node-specific spatial-temporal dependencies.…”

“…DGCRN: Dynamic Graph Convolutional Recurrent Network indicates that their dynamic graph can cooperate effectively with pre-defined graph while improving the prediction performance [ 33 ];…”

“…For reference [ 3 ], the dimension of the node embedding was one of the crucial parameters in AGCRN and its optimal value was 10. For reference [ 33 ], the layers, the size of the hidden state, and batch size of DGCRN were set to 1, 64, and 64, respectively.…”