Vessel Traffic Flow Prediction on Sparse Data via Spatio-Temporal Graph Neural Networks with a Learnable Tweedie Head
arXiv:2606.07694v1 Announce Type: new Abstract: Accurate vessel traffic flow prediction is crucial for smart port operations and navigational safety. However, maritime traffic flow data are often highly sparse with intermittent bursts, making robust forecasting challenging. Under such conditions, conventional spatio-temporal graph neural networks (ST-GNNs) can degrade toward conservative near-zero predictions and fail to capture non-zero activity. Although zero-inflated negative binomial (ZINB) models partially address excess zeros, their two-part formulation can still remain conservative arou
The increasing complexity of maritime logistics and the growing data availability from smart port initiatives are driving the demand for more robust prediction models capable of handling real-world data challenges.
Accurate vessel traffic prediction is critical for optimizing global supply chains, improving port efficiency, and enhancing navigational safety, areas of significant economic and strategic importance.
This research introduces a novel deep learning approach that significantly improves the ability to predict maritime traffic flow even with sparse data, overcoming limitations of previous models.
- · Smart port operators
- · Logistics and shipping companies
- · Maritime AI/ML developers
- · Supply chain management platforms
- · Ports reliant on inefficient manual scheduling
- · Older, less adaptive forecasting software vendors
More efficient port operations lead to reduced vessel waiting times and lower fuel consumption.
Improved traffic predictions contribute to a reduction in maritime accidents and better environmental management within ports.
Enhanced logistical predictability could support the development of more resilient and adaptive global supply chains, reducing system shocks.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG