Abstract:Traffic flow forecasting, which requires modelling involuted spatial and temporal dependence and uncertainty regarding road networks and traffic conditions, is a challenge for intelligent transportation systems (ITS). Recent studies have mainly focused on modelling spatial-temporal dependence through a fixed weighted graph based on prior knowledge. However, collecting up-to-date and accurate road information is costly. Moreover, is a single fixed graph enough to describe the correlation between sensors? The fixed weighted graph cannot directly relate to prediction tasks, which may result in considerable biases. To tackle this issue, in this paper, we propose a novel deep learning model framework: an adaptive spatial-temporal graph attention network (ASTGAT). Our ASTGAT simultaneously learns the dynamic graph structure and spatial-temporal dependency for traffic flow forecasting. Specifically, our framework consists of two joint training parts: a Network Generator model that generates a discrete graph with the Gumbel-Softmax technique and a Spatial-Temporal model that utilizes the generated network to predict traffic speed. Our Network Generator can adaptively infer the hidden correlations from data. Moreover, we propose a graph talking-heads attention layer (GTHA) for capturing spatial dependencies and design a gate temporal convolution (GTCN) layer for handling long temporal sequences. We evaluated our ASTGAT on two public datasets: METR-LA is collected in Los Angeles and PEMS-BAY is collected in California. Experimental results indicate that our ASTGAT outperforms the state-of-the-art (SOTA) baselines. Finally, to further describe our model, we visualize the forecasting results and the generated graph.

Adaptive spatio-temporal graph convolutional network with attention mechanism for mobile edge network traffic prediction

A Spatiotemporal Graph Neural Network with Graph Adaptive and Attention Mechanisms for Traffic Flow Prediction

Graph Attention Spatial-Temporal Network for Deep Learning Based Mobile Traffic Prediction.

Adaptive Hybrid Spatial-Temporal Graph Neural Network for Cellular Traffic Prediction

Adaptive Graph Convolutional Recurrent Neural Networks for System-Level Mobile Traffic Forecasting

Adaptive Multi-receptive Field Spatial-Temporal Graph Convolutional Network for Traffic Forecasting

Network traffic prediction with Attention-based Spatial-Temporal Graph Network

A mobility aware network traffic prediction model based on Dynamic Graph Attention Spatio-temporal Network

Spatial-Temporal Adaptive Graph Convolution with Attention Network for Traffic Forecasting

Enhanced edge convolution-based spatial-temporal network for network traffic prediction

Adaptive Spatial-Temporal Graph Convolution Networks for Collaborative Local-Global Learning in Traffic Prediction

A Self-Adaptive Spatio-Temporal Graph Convolutional Network for Traffic Prediction

Traffic Flow Forecasting with Spatial-Temporal Graph Convolutional Networks in Edge-Computing Systems

Spatio-Temporal Attention-based Graph Convolution Networks for Traffic Prediction

Adaptive spatial-temporal graph attention networks for traffic flow forecasting

Spatial-Temporal Adaptive Convolutional Network with External Factors for Cellular Traffic Prediction

State Transition Graph-Based Spatial-Temporal Attention Network for Cell-Level Mobile Traffic Prediction

Adaptive Spatio-Temporal Convolutional Network for Traffic Prediction.

ESGCN: Edge Squeeze Attention Graph Convolutional Network for Traffic Flow Forecasting

An adaptive spatio-temporal graph neural network for traffic prediction

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