Green Security Games

Опубликовано 13 июня 2016, 17:26

Security is a global concern, requiring efficient, randomized allocation and scheduling of limited security resources. To that end, we have used computational game theory to build decision aids for security agencies around the world, leading to the creation of a new research area of "security games". Today security-games based decision aids for infrastructure security are deployed in the US and internationally for protection of ports, airports, air traffic, trains and campuses. This talk will focus on our recent work on "green security games" that builds on this previous work. Green security games and their applications are focused on deploying limited security resources to protect forests, fish, wildlife, e.g., assisting security agencies against poaching, illegal fishing or illegal logging. Unlike previous deployments of single-shot games for infrastructure protection, green security games are repeated games where security agencies must act against adversaries that repeatedly engage in illegal activities, and where adversaries act with more of a limited surveillance and limited ability to plan each activity. After providing a quick overview of early work in security games, I will discuss our experience testing green security game software in protected areas, and discuss key research challenges in scaling up to large-scale problems, in modeling bounded rationality of human adversaries and machine learning in green security games.