Security Informatics: An Artificial Intelligence Approach to Security
Friday Theory Seminar at BU presented by the CAS Computer Science Department and BU RISCS Center
The ability to model, understand, and respond to the behavior of other agents has applications in many contexts, from gaming to financial investing. One particularly salient domain is the field of international security and development where artificial intelligence models can be leveraged to analyze complex and uncertain security situations. These behavioral models can form the foundation for decision-support technologies to help policymakers looking for conflict management, international development, or national security strategies.
Of course, behavioral modeling is a computationally challenging problem; real world datasets can contain 10^10,000 possible behaviors requiring efficient techniques to manage the analytic complexity. In this talk, I will present an overview of my research in this area, looking at the probabilistic logic formalisms and scalable reasoning algorithms I have used to study the behavioral dynamics of the Afghan drug trade and violent ethnopolitical conflict in the Middle East. I will then discuss new results on using abductive reasoning to identify possible counter strategies and the extension to temporal domains.