ITR: Research on the Profiling Problem in Cybersecurity and Anti-Terrorism

Louisiana State University, Baton Rouge LA

Investigators

Abstract

This project proposes to perform research on new and theoretically sound profiling techniques for detection and identification of terrorists and cyber attacks. A common thread of problems in such attacks is how can we detect "bad guys" from good guys quickly? Various scientific techniques such as data modeling and integration, meta-data management, discrete mathematical models and algorithms, data mining, machine learning, visualization and statistical techniques will be investigated. For example, new mathematical models and algorithms for profiling "bad guys/transactions" will be developed and tested. One research direction is to formulate the problem as a more generalized formulation of the current set covering problem. Efficient algorithms will be developed, and performance bounds of certain algorithms will be studied. Simulations will be done to compare the performance of the new algorithms with conventional algorithms, and the pros and cons of these algorithms will be identified. Making the models more realistic and testing the algorithms/techniques against more realistic data sets will be done by teaming with LSU's National Center for Security Research and Training (NCSRT) , FBI, AFRL, and others. The project team consists of diverse ethnic groups and plans to organize a workshop dedicated to women and under-represented minority. . If successful, this project can have significant impact on the scientific community and on the society.