U.S. Department of Energy

Pacific Northwest National Laboratory

Multi-source Signatures of Nuclear Programs

Principle Investigator: 

This project develops, validates, and demonstrates the use of model-based data signatures in a variety of nuclear intent analytic challenges. Detecting and identifying the intent to develop (or further) a nuclear weapons program requires inputs from diverse sets of expertise, and requires understanding of diverse types of information and data. Currently, this analytic activity is executed by teams of analysts. While an analyst's expertise typically crosses multiple disciplines, the entire set of required capabilities does not reside within any single analyst.

This project addresses both the heterogeneities in data and expertise to develop normalized, context-informed summaries of the data used to assess nuclear intent. These issues are addressed by representing analytic expertise and known relationships as influence diagrams or other modeling formalisms. Normalization of the heterogeneous data is accomplished by the application of these models to the data. For each piece of data, the model produces a set of normalized features corresponding to some predetermined outcomes of the model. The feature vectors will be suitable for a variety of information analytic tasks, such as trending, anomaly detection, and summarization.

Project Staff: 
Paul Whitney, Cindy Henderson, Jana Strasburg, Ken Summa, Amanda White
PDF icon SDI_Multisource_flyer.pdf1.42 MB
| Pacific Northwest National Laboratory