ADT: i-Group Learning and i-Detect for Dynamic Real Time Anomaly Detection with Applications in Maritime Threat Detection

点击翻译按钮获取中文摘要

Threats to national and global security through the maritime transportation system can be multi-faceted and serious, ranging from human and drug trafficking, smuggling, transport of nuclear material and dirty bombs, to garbage dumping and illegal fishing. Hence it is important to design an efficient early detection and risk assessment system for maritime traffic over space and time. With today's data gathering capabilities and global regulation and agreements, it is now possible to achieve such a goal with sophisticated and advanced statistical tools. The project develops novel statistical tools for individualized grouping and baseline distribution formation and for subsequent individualized detection of abnormal deviations from the baseline distribution, with a quantified risk assessment. The new developments are target-oriented and more precise, thus more effective than conventional methods. The project utilizes the Automated Identification System (AIS) data that, by international convention, is transmitted by over a million vessels worldwide, and combines this data with geological, geographical, and geophysical data about oceans and river systems, and coastlines and ports. It provides a timely and effective tool for the improvement of national security. The project also provides a rich training ground for next generation data scientists with interdisciplinary research experiences in maritime security and risk assessment.The project investigates novel statistical approaches for threat detection, utilizing a wide range of data sources. The developed iGroup method and iDetect tools are general statistical methods that form a useful framework for many threat detection and risk assessment problems. They enrich the theory and methods of a new statistical analytical toolkit. The project focuses on utilizing the developed methods in one important application, maritime traffic threat detection and risk assessment. It advances the science of threat detection by developing new tools for dealing with heterogeneous data, extending the data-depth method to complex space, finding new learning tools for handling features that differ in importance, and producing anomaly detection methods for situations where data is missing, incorrect, or deliberately misleading. The development of a real-time threat detection and risk assessment system for maritime traffic can enhance maritime safety and at the same time detect human and group movements on vessels and more generally criminal and terrorist activities of various kinds. The methods to be developed also offer a general framework for threat detection in other areas, such as cell phone monitoring, cyber security, anti-money laundering, market analysis, information retrieval, and personalized medicine. The project also has a strong interdisciplinary educational component for training the next generation of data scientists.