A Review on the State of the Art of Machine Learning and Satellite Imaging: Detecting Scene Changes in Selected Nuclear Fuel Cycle Datasets

Shiloh Elliott - Idaho National Laboratory (INL) BEA
Ashley Shields - Idaho National Laboratory (INL) BEA
Eduardo Trevino - Idaho National Laboratory
Gustavo Reyes - Idaho National Laboratory (INL) BEA
Mark Schanfein - Idaho National Laboratory
The timely detection of clandestine nuclear facilities through satellite imagery (SI) analysis is one of the greatest challenges faced by IAEA’s SGIS satellite imagery processing capabilities. INL, through DHS, is currently applying machine learning (ML) to existing SI databases to find critical infrastructure facilities across the USA, with primary focus placed on identifying critical predecessor and successor facilities. For example, in the case of a clandestine research reactor, predecessor and successor facilities would include steel forging plants, water treatment services, and specialized logistic distribution centers (inter alia). This project explores the use of available public SI datasets for ML, applying mature and proven classification algorithms and scene change detection software to satellite imagery. This capability addresses IAEA’s needs to detect undeclared nuclear materials and activities within a State while encompassing the entire nuclear fuel cycle.