Abstract
Given the real-world applications of Stackelberg security games (SSGs), addressing uncertainties in these games is a major challenge. Two competitive approaches have been pursued by previous work on addressing uncertainties in SSGs, namely: (1) applying robust optimization techniques without requiring a prior distribution; and (2) using probabilistic models to capture uncertainties. In general, the decision of which approach to use is based on the availability of data. While the first approach suits data-sparse domains, the second approach works better for data-rich domains. My thesis will focus on addressing uncertainties in SSGs following these two leading approaches. In particular, with regards to robust methods, I attempt to develop new maximin/minimax regret-based robust algorithms for computing a defender's optimal strategy given uncertainties. I also aim to contribute to probabilistic modeling techniques by developing a new computational model of human decision making to capture the adversary's bounded rationality.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have