Deep Multi-View Channel-Wise Spatio-Temporal Network for Traffic Flow Prediction

TL;DR

This work tackles traffic flow forecasting under dynamic mobility and complex spatio-temporal dependencies. It introduces MVC-STNet, which combines localized and global spatial graphs with a channel-wise graph convolutional network (CGCN) to model channel-specific influences, followed by LSTM-based temporal modeling and integration of external features. Key components include the local adjacency $A_g$, global adaptive adjacency $A_s$, multi-view fusion across channels, and an objective that minimizes $\|\hat{Y}^t-Y^t\|^2$; these enable effective joint spatial-temporal and cross-channel reasoning. Experiments on PeMS04 and PeMS08 demonstrate state-of-the-art performance, with ablations confirming the contributions of CGCN and multi-view fusion to robust predictions in real-world traffic networks.

Abstract

Accurately forecasting traffic flows is critically important to many real applications including public safety and intelligent transportation systems. The challenges of this problem include both the dynamic mobility patterns of the people and the complex spatial-temporal correlations of the urban traffic data. Meanwhile, most existing models ignore the diverse impacts of the various traffic observations (e.g. vehicle speed and road occupancy) on the traffic flow prediction, and different traffic observations can be considered as different channels of input features. We argue that the analysis in multiple-channel traffic observations might help to better address this problem. In this paper, we study the novel problem of multi-channel traffic flow prediction, and propose a deep \underline{M}ulti-\underline{V}iew \underline{C}hannel-wise \underline{S}patio-\underline{T}emporal \underline{Net}work (MVC-STNet) model to effectively address it. Specifically, we first construct the localized and globalized spatial graph where the multi-view fusion module is used to effectively extract the local and global spatial dependencies. Then LSTM is used to learn the temporal correlations. To effectively model the different impacts of various traffic observations on traffic flow prediction, a channel-wise graph convolutional network is also designed. Extensive experiments are conducted over the PEMS04 and PEMS08 datasets. The results demonstrate that the proposed MVC-STNet outperforms state-of-the-art methods by a large margin.

Figures (5)

• Figure 1: Relationship between traffic flow and two traffic observation features (vehicle speed and road occupancy)
• Figure 2: The framework of MVC-STNet.(Relu is a non-linear activation function)
• Figure 3: The illustration of channel-wise graph convolutional network.(Relu is a non-linear activation function)
• Figure 4: Loss curves of MVC-STNet on the two datasets
• Figure 5: Prediction vs ground truth on two datasets(top to down: node 7 and node 15 in PeMS04, node 20 and node 29 in PeMS08)

Theorems & Definitions (2)

• Definition 1
• Definition 2