An Adaptive Spatio-Temporal Traffic Flow Prediction Using Self-Attention and Multi-Graph Networks DOI Creative Commons

Basma Alsehaimi,

Ohoud Alzamzami, Nahed Alowidi

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(1), P. 282 - 282

Published: Jan. 6, 2025

Traffic flow prediction is a pivotal element in Intelligent Transportation Systems (ITSs) that provides significant opportunities for real-world applications. Capturing complex and dynamic spatio-temporal patterns within traffic data remains challenge prediction. Different approaches to effectively modeling correlations have been proposed. These often rely on single model capture temporal dependencies, which neglects the varying influences of different time periods flow. Additionally, these models frequently utilize either static or graphs represent spatial limits their ability address overlapping relationships. Moreover, some struggle fully variations, leading exclusion critical information ultimately resulting suboptimal performance. Thus, this paper introduces Adaptive Spatio-Temporal Attention-Based Multi-Model (ASTAM), an architecture designed dependencies data. The ASTAM employs multi-temporal gated convolution with multi-scale input segments non-linear correlations. It utilizes parallel multi-graphs facilitate dependencies. Furthermore, incorporates self-attention mechanism adaptively long-term variations Experiments conducted four datasets reveal proposed outperformed 13 baseline approaches, achieving average reductions 5.0% MAE, 13.28% RMSE, 6.46% MAPE across datasets.

Language: Английский

An Adaptive Spatio-Temporal Traffic Flow Prediction Using Self-Attention and Multi-Graph Networks DOI Creative Commons

Basma Alsehaimi,

Ohoud Alzamzami, Nahed Alowidi

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(1), P. 282 - 282

Published: Jan. 6, 2025

Traffic flow prediction is a pivotal element in Intelligent Transportation Systems (ITSs) that provides significant opportunities for real-world applications. Capturing complex and dynamic spatio-temporal patterns within traffic data remains challenge prediction. Different approaches to effectively modeling correlations have been proposed. These often rely on single model capture temporal dependencies, which neglects the varying influences of different time periods flow. Additionally, these models frequently utilize either static or graphs represent spatial limits their ability address overlapping relationships. Moreover, some struggle fully variations, leading exclusion critical information ultimately resulting suboptimal performance. Thus, this paper introduces Adaptive Spatio-Temporal Attention-Based Multi-Model (ASTAM), an architecture designed dependencies data. The ASTAM employs multi-temporal gated convolution with multi-scale input segments non-linear correlations. It utilizes parallel multi-graphs facilitate dependencies. Furthermore, incorporates self-attention mechanism adaptively long-term variations Experiments conducted four datasets reveal proposed outperformed 13 baseline approaches, achieving average reductions 5.0% MAE, 13.28% RMSE, 6.46% MAPE across datasets.

Language: Английский

Citations

0