Advanced Graph Neural Networks: urban traffic forecasting

Project Type: Academic

Main Challenge:

In the realm of urban traffic forecasting, advanced Graph Neural Networks (GNNs) incorporating attention mechanisms have emerged as a cutting-edge approach. These sophisticated models utilize GNNs to effectively capture and analyse complex spatial and temporal relationships in urban traffic networks. By incorporating attention mechanisms, they can focus on critical information within the graph, improving prediction accuracy and enhancing the understanding of traffic patterns and congestion dynamics. This innovative combination of GNNs and attention mechanisms has shown great promise in advancing the field of urban traffic forecasting, offering more accurate and insightful predictions for optimizing transportation systems and mitigating traffic congestion.

Funding: Internal

DA&AI tools:

• Deep Learning
• Multivariate data processing

Publications:

Perevozcikova, J., Pavlyuk, D., 2023. Attention-Based Spatio-Temporal Graph Convolutional Networks – A Systematic Review, in: Reliability and Statistics in Transportation and Communication, Lecture Notes in Networks and Systems. Springer International Publishing, Cham, pp. 26–33. https://doi.org/10.1007/978-3-031-26655-3_3