Dynamic Jacobi graph and trend-aware flow attention convolutional network for traffic forecasting

Yang, Yongpeng; Yang, Zhenzhen; Yang, Zhen

doi:10.1016/j.dsp.2023.104156

Cited by 2 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…-DSTAGNN [13] utilizes Wasserstein distance [18] and spatial-temporal at tention mechanism to capture dynamic spatial-temporal correlations.…”

Section: Methodsmentioning

confidence: 99%

“…STFGNN [14] exploits DTW technique to compute the similarity between time sequences of nodes to construct a spatial-temporal fusion graph. DSTAGNN [13] emerges Wasserstein distance [25] to measure differences of traffic flow distributions in different roads and construct a new graph to incorporate the spatial-temporal attention mechanism. Nevertheless, none of these methods focus on turn-level traffic forecasting, and neglect the global, non-pairwise correlation between road segments.…”

Section: Graph Based Traffic Forecastingmentioning

confidence: 99%

“…First, the correlations between roads change dynamically over time, but most existing GNNbased works [10,11,12] employ a static adjacent matrix to represent spatial dependencies between roads, which cannot reflect this dynamicity. Though some works attempt to construct a new graph structure that contains both spatial and temporal information [13,14,15], these manually predefined adjacency matrices may introduce bias, further reducing the generalizability of the model. Second, as shown in Figure 1(b), traffic flow distribution between different road segment pairs can have significant discrepancies in different turn directions and time segments in a day.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A GNN Approach for Turn-Level Traffic Prediction: Dynamic Relation Awareness and Hypergraph Modeling

2024

jeis

View full text Add to dashboard Cite

It cannot be emphasized too much to predict traffic flow accurately in modern intelligent transportation systems. Though big progress has been made, few works focus on the turn-level traffic flow prediction, which is important to inspect fine-grained urban traffic dynamics closely. In this work, we develop a GNN (Graph Neural Network) approach built upon Dynamic Relation Awareness and Hypergraph modeling toward turnlevel traffic flow prediction, namely DrahGNN. First, we construct a dynamic graph sequence where each snapshot denotes a turn-level traffic flow picture on top of a realworld road network. Second, we innovate a relation-aware spatiotemporal diffusion convolution network to capture road segments' differences and relatedness explicitly. Third, we construct a hypergraph in each time frame to capture high-order and manifold correlations between road segments and design an attentive two-stage message-passing mechanism for aggregating infor-mation from non-directly connected nodes. We conduct empirical studies on real-world data which demonstrate the effectiveness of our proposed framework.

show abstract

“…-DSTAGNN [13] utilizes Wasserstein distance [18] and spatial-temporal at tention mechanism to capture dynamic spatial-temporal correlations.…”

Section: Methodsmentioning

confidence: 99%

Section: Graph Based Traffic Forecastingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A GNN Approach for Turn-Level Traffic Prediction: Dynamic Relation Awareness and Hypergraph Modeling

2024

jeis

View full text Add to dashboard Cite

show abstract

Gated Spatial–Temporal Merged Transformer Inspired by Multimask and Dual Branch for Traffic Forecasting

Yang,

Yang

2024

IET Signal Processing

View full text Add to dashboard Cite

As an essential part of intelligent transportation system (ITS), traffic forecasting has provided crucial role for traffic management and risk assessment. However, complex spatial–temporal dependencies, heterogeneity, dynamicity, and periodicity of traffic data influence the traffic forecasting performance. Consequently, we propose a novel effective gated spatial–temporal merged transformer (GSTMT) inspired by multimask and dual branch for accurate traffic forecasting in this paper. Specifically, we first conduct a concatenation of gated spatial static mask transformer (GSSMT) and gated spatial dynamic mask transformer (GSDMT) with residual network. The GSSMT and GSDMT evolve from the traditional transformer by making preferable modifications that include gated linear unit (GLU), multimask mechanism including static mask matrix (SMM) and dynamic mask matrix (DMM), and spatial attention (SA). Among them, GLU is to promote the performance of capturing spatial dependency, dynamicity, and heterogeneity due to advanced performance for controlling information flow through layers. Additionally, by developing multimask mechanism including two novel SMM and DMM, the proposed GSTMT can precisely model the static and dynamic spatial structure for effectively highlighting static dependency and dynamicity. And SA is injected for enhancing the ability of capturing spatial dependency of GSSMT and GSDMT. Secondly, we develop a dual‐branch gated temporal transformer (DBGTT) for capturing temporal dependency, heterogeneity, dynamicity, and periodicity via incorporating the GLU and mixed time series decomposition (MTD) into traditional transformer. Similarly, we also introduce the GLU for empowering DBGTT with capability of capturing temporal dependency, dynamicity, and heterogeneity. In addition, MTD, which brings dual‐branch mechanism, can enhance the DBGTT for capturing more detailed temporal information via exploiting global and periodic profile of traffic data. At last, some experiments, which are performed on several real‐world traffic datasets, demonstrate the better results over classic traffic forecasting methods.

show abstract

Dynamic Jacobi graph and trend-aware flow attention convolutional network for traffic forecasting

Cited by 2 publications

References 25 publications

A GNN Approach for Turn-Level Traffic Prediction: Dynamic Relation Awareness and Hypergraph Modeling

A GNN Approach for Turn-Level Traffic Prediction: Dynamic Relation Awareness and Hypergraph Modeling

Gated Spatial–Temporal Merged Transformer Inspired by Multimask and Dual Branch for Traffic Forecasting

Contact Info

Product

Resources

About