A multi-modal attention neural network for traffic flow prediction by capturing long-short term sequence correlation.

Area of Science:

• Artificial Intelligence
• Transportation Engineering
• Data Science

Background:

• Accurate traffic flow prediction is crucial for efficient transportation management and driver decision-making.
• Existing methods often overlook temporal correlations, focusing primarily on temporal continuity.
• This limitation hinders the development of truly effective traffic prediction models.

Purpose of the Study:

• To propose a novel multi-modal attention neural network for enhanced traffic flow prediction.
• To effectively capture both spatio-temporal dependencies and long-short term sequence correlation (LSTSC).
• To improve the accuracy and reliability of traffic flow forecasting, particularly for long-term predictions.

Main Methods:

• Development of a multi-modal attention neural network architecture.
• Utilizing attention mechanisms to capture complex spatio-temporal correlations within traffic data.
• Incorporating a novel approach to model long-short term sequence correlation (LSTSC).

Main Results:

• The proposed model demonstrated superior accuracy and reliability in traffic flow prediction.
• Significant improvements were observed, especially in long-term prediction scenarios.
• Validation on PeMS08 and PeMSD7(M) datasets confirmed the model's effectiveness.

Conclusions:

• The multi-modal attention neural network effectively addresses limitations of existing methods by considering temporal correlations.
• The model's ability to capture LSTSC leads to more accurate and reliable traffic flow predictions.
• This approach offers a promising advancement for intelligent transportation systems.