A consortium of researchers dedicated to improving the understanding of the human causes and consequences of terrorism

Advancing machine learning algorithms that could predict terrorist threats

DHS follow-on funding provides opportunity for improvement

Dr. Prabhakar Misra, START research affiliate and Howard University Professor, recently published a report updating his research on machine learning algorithms that could one day predict terrorist threats.

Utilizing his background as a physicist, Misra has been working with the National Consortium for the Study of Terrorism and Reponses to Terrorism (START) for three years. Misra is a recipient of the Excellence in Research Productivity award and was elected as a fellow of the American Physical Society (APS). His research focuses on the detailed characterization of a variety of nanomaterials (e.g. carbon nanotubes, graphene, and metal oxides) using Raman spectroscopy and Molecular Dynamics simulation techniques.

To expand on his previous research from the summer of 2014, the primary aim of this START project involved making improvements to machine processing capabilities. Part of the follow-on research sought to address missing and incomplete data, in addition to improving the pattern recognition in START databases.

Misra along with his Howard University graduate students Raul Garcia-Sanchez and Daniel Casimir were able to modify the machine learning code used for importing the Global Terrorism Database (GTD) so that they could easily import all START-formatted databases.

Additionally, the research project successfully resolved previous issues thus improving the algorithm’s misclassification error by 8 percent. Computation also reduced to one hour.

The findings are explored in the report “Development and Optimization of Machine Learning Algorithms and Models of Relevance to START Databases.”

The follow-on project was supported by the Department of Homeland Security Research Team Follow-On Funding Program for Minority Serving Institutions.