Vessel traffic flow prediction through multi-scale spatiotemporal attention in dual-graph networks

Key Points

• The outcome highlights improved prediction accuracy of vessel traffic flows, fostering better navigation strategies.
• Key evidence indicates a 20% increase in accuracy for traffic predictions across multiple scenarios.
• Analysis utilizing multi-scale spatiotemporal attention on dual-graph networks was performed to optimize flow predictions.
• Enhancing vessel traffic management systems may significantly reduce congestion in busy maritime areas.