Chen, Lei (Southern University of Science and Technology), Wei, Xuetao (Southern University of Science and Technology), Zhang, Shiyao (Great Bay University)

ST-TransNet: An Adaptive Spatio-Temporal Transfer Network for City-Wise Traffic Speed Prediction

Scheduled for presentation during the Regular Session "S34b-Data-Driven Optimization and Governance in Intelligent Urban Mobility" (FR-EA-T34), Friday, November 21, 2025, 14:10−14:30, Surfers Paradise 1

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Traffic speed prediction is a core task of Intelligent Transportation Systems (ITS), providing crucial support for urban traffic management. However, due to economic and infrastructure limitations, data scarcity in several cities adversely affects traffic prediction accuracy. To address this issue, we propose a transfer learning-based framework, SpatioTemporal Transfer Network (ST-TransNet), for city-wise traffic speed prediction. ST-TransNet employs graph partitioning to obtain fine-grained traffic graph representations, leveraging Graph Attention Network (GAT) and Mamba to capture dynamic spatiotemporal features of traffic data. By adopting a parameter-sharing transfer learning strategy, ST-TransNet is pretrained on data-rich cities and fine-tuned on limited data from data-scarce cities, enabling knowledge transfer of traffic patterns. Experimental results on real-world traffic datasets show that ST-TransNet outperforms state-of-the-art baselines in data-scarce target domains. In addition, the accurate, robust, and generic traffic prediction results demonstrate the reliable application in data-limited cities.