Algorithmic Tools for Adversarial Games

Counteracting the actions of groups that direct and implement violent behavior towards the United States and its citizens is one of the major issues of our time. Elimination of individuals identified as threats is apparently of marginal long-term utility: other adversaries arise to replace fallen comrades. The conflict that is the global war on terror exists not only on an individual level but also on group and societal levels. The interplay of the dynamics among these levels is an important process to understand if we are to defeat terrorist agendas. A systems-theoretic approach to the social network dynamics of such groups provides an alternative to direct individual-based intervention, and there is significant potential in taking such an approach. We propose the development of cutting-edge algorithms for dynamic, stochastic games, integrating dynamic models of social networks with game-theoretic approaches to decision and our unique tools for handling uncertainty, intent estimation, and robustness to deception, for the improvement of decision-making within such contexts.