About the project
This project will develop advanced machine learning and stochastic optimisation methods to identify anomalous behaviours in maritime traffic using Automatic Identification System (AIS) data. It will enhance maritime situational awareness and security while addressing uncertainty and complexity in vessel behaviour modelling.
The maritime domain is critical for global trade and economic stability, yet faces threats from anomalous behaviours such as illegal fishing, smuggling, and piracy. Automatic Identification System (AIS) data offers a rich source of real-time vessel movement information. However, detecting anomalies within such high-dimensional, noisy, and incomplete datasets remains a challenge.
This project aims to bridge machine learning (ML) and stochastic programming for robust maritime anomaly detection. You will:
- develop ML-based models (deep learning, unsupervised clustering, probabilistic models) to classify and predict vessel behaviours from AIS data
- formulate stochastic optimisation models to address uncertainty and incomplete data, enabling decision-support for anomaly identification
- explore hybrid ML–optimisation approaches to enhance detection accuracy, interpretability, and scalability
- validate methods on real-world AIS datasets, benchmarking against existing anomaly detection methods, and collaborate with maritime stakeholders
The project will contribute to complex integrated systems research, providing a robust framework for maritime domain awareness, supporting sustainable shipping, and enhancing global maritime security.
In addition to the University of Southampton supervisor, this project includes the following external supervisors:
- Professor Peter Schütz - NHH Norwegian School of Economics
- Dr Wissam Albukhanajer - Roke
