Abstract:Forecasting traffic flow is crucial for Intelligent Traffic Systems (ITS), traffic control, and traffic management systems. Complex spatial and temporal interactions of traffic networks make traffic forecasting tasks challenging. Recently, Graph Convolutional Network (GCN) has attracted researchers' attention as it can better represent graph-shaped road networks and extract spatial features of traffic. However, traditional GCN has some drawbacks since it uses a static adjacency matrix which is unable to capture the time-varying features of traffic propagation. To overcome this, we represent the traffic road network as a dynamic graph and use a probabilistic spatiotemporal adjacency matrix to identify the time-varying impacts of adjacent roads on target roads in GCN. In addition, to find the similarity among the nonadjacent nodes, we have employed node-specific learning in GCN rather than sharing parameters in traditional GCN. This node-specific learning helps our model to learn detailed characteristics of road networks. For temporal feature extractions, we used a Gated Recurrent Unit (GRU) that captures the local trend of traffic flow and an attention mechanism to capture the global trend of traffic flow. We compared the performance of our model with baseline models using two real-world datasets. Experimental results show that our model is effective in forecasting both short and long-term traffic flow. Source code of our model is available at https://github.com/atkia/PSTGCN

Hierarchical Graph Convolution Network for Traffic Forecasting

A Spatial-Temporal Gated Hypergraph Convolution Network for Traffic Prediction

Attention Based Graph Covolution Networks for Intelligent Traffic Flow Analysis

Directed Hypergraph Attention Network for Traffic Forecasting

Hierarchical Spatio–Temporal Graph Convolutional Networks and Transformer Network for Traffic Flow Forecasting

Graph learning-based spatial-temporal graph convolutional neural networks for traffic forecasting

Adaptive Hierarchical SpatioTemporal Network for Traffic Forecasting

Dynamic Spatio-Temporal Hypergraph Convolutional Network for Traffic Flow Forecasting

Traffic Graph Convolutional Recurrent Neural Network: A Deep Learning Framework for Network-Scale Traffic Learning and Forecasting

Dual Dynamic Spatial-Temporal Graph Convolution Network for Traffic Prediction

T-GCN: A Temporal Graph ConvolutionalNetwork for Traffic Prediction

T-GCN: A Temporal Graph Convolutional Network for Traffic Prediction

Probabilistic spatio-temporal graph convolutional network for traffic forecasting

Optimized Graph Convolution Recurrent Neural Network for Traffic Prediction

Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting

Local-Global Spatial-Temporal Graph Convolutional Network for Traffic Flow Forecasting

Bayesian Spatio-Temporal Graph Convolutional Network for Traffic Forecasting

Attention Based Spatial-Temporal Graph Convolutional Networks for Traffic Flow Forecasting

Constructing Geographic and Long-term Temporal Graph for Traffic Forecasting

A Graph Convolutional Method for Traffic Flow Prediction in Highway Network