Conventional approaches to radiation measurements are sometimes insufficient to address the challenges of nuclear threat detection in real-world conditions. Threat-aware signal processing, statistical methods, machine learning, and artificial intelligence are fast becoming an important part of the analysis of large data sets acquired by radiation detectors and supporting contextual sensors. These computational methods enable us to unmask data correlations, trends, and patterns that provide valuable insights of the physics phenomena under study. The methods and insights can then be used to fuse data from different measurement modalities and analyze measurements with greatly enhanced performance. This talk will describe some of the difficulties in nuclear threat detection and how machine learning can be applied to address these challenges. Problems specific to applying machine learning to nuclear threat detection, some of which can be catastrophic, will be discussed, and solutions will be presented that can avoid these issues.
Simon Labov is an expert in nuclear detection systems, advanced spectral and multisource analysis algorithms, and distributed detector systems. He currently leads the development of the machine learning-enabled Enhanced Radiological Nuclear Inspection and Evaluation (ERNIE) System. ERNIE is now operating at some of the busiest U.S. seaports to increase sensitivity and reduce false alarms for radiation portal monitors. He also leads an algorithm development team for the ubiquitous detection SIGMA program and other mobile detection systems. In 2001, he initiated the Cellular-Telephone-Based Radiation Sensor and Wide-Area Detection Network Project, which was awarded three patents and has been aggressively supported by multiple funding agencies. Simon Labov joined LLNL in 1987 and helped initiate a program to develop high-resolution, energy-dispersive x-ray detectors that operate at very low temperatures, and then founded and directed the LLNL Radiation Detection Center. Prior to joining LLNL, Labov developed instrumentation for soft-x-ray astrophysics and received a B.S. at Stanford University and a Ph.D. at the University of California, Berkeley.