AI Seminar

Combating Adversaries under Uncertainties in Real-world Security Problems: Advanced Game-theoretic Behavioral Models and Robust Algorithms

Thanh NguyenPhD CandidateUniversity of Southern California

Security is a global concern. Real-world security problems range from domains such as the protection of ports and airports from terrorists to protecting forests and wildlife from smugglers and poachers. A key challenge in solving these security problems is that security resources are limited; not all targets can be protected all the time. Therefore, security resources must be deployed intelligently, taking into account responses of attackers and potential uncertainties over their types, preference, and knowledge. Stackelberg Security Games (SSG) have drawn a significant amount of interest from security agencies. SSG-based decision aids are in widespread use for the protection of assets such as major ports in the US and airport terminals.

My research focuses on addressing uncertainties in SSGs — one recognized area of weakness in SSGs. For example, adversary payoff values can be extremely difficult to assess and are generally characterized by significant uncertainty. My talk will cover three major contributions of my research, including: 1) new behavioral models of attackers grounded in data from both real-world domains and human subject experiments; 2) new robust planning algorithms for security agencies developed for a variety of domain uncertainty settings; and 3) deployment of my algorithms in the PAWS system, which is currently being used by NGOs in a conservation area in Malaysia.

Sponsored by

Strategic Reasoning Group

Faculty Host

Michael Wellman