
2021 – Present
This project addresses the significant gap in how agencies and stakeholders allocate resources to defend against threats to soft targets, as current methods often rely on intuition and outdated practices rather than data-driven strategies. Current practices lead to inefficiencies in deploying countermeasures and increase the vulnerability of soft targets to strategic attacks. The primary objective is to develop game-theoretic models, which strategically predict how humans will act and make decisions based on several factors. These models aid organizations in optimizing the allocation of security resources against strategic attackers based on these predictive models, which incorporate stakeholder-informed parameters and data, offering a tool for better decision-making and resource distribution to defend soft targets.
The project employs a combined approach, using game theory, experimental psychology, behavioral economics, and risk analysis to construct models to capture the strategic choices considered before an attack at a soft target venue. The increased understanding of the attackers’ decision-making process allows for the optimization of countermeasures and resource allocation in the event of an attack. The project seeks to integrate its game-theoretic models into real-world applications by developing a graphical user interface (GUI) tool. This tool visualizes vulnerabilities in defense systems and helps defenders optimally allocate resources. The practical applications include improving security plans for schools, houses of worship, and public venues while fostering information sharing between soft targets and local law enforcement.
This project improves the state of the art by creating more realistic game-theoretic models that account for multiple layers of defense, multiple targets, and uncertainties, such as attackers’ lack of knowledge about certain defenses. It also incorporates behavioral game studies to validate the models using data from real-world decision-making experiments. By combining these validated models with machine-learning tools, the project advances theoretical understanding and provides practical solutions for optimal resource allocation against strategic adversaries, significantly improving upon traditional, simplistic approaches.
In 2024, the project has improved its models based on stakeholder engagement and survey results. It has developed partnerships with schools, houses of worship, sports venues, and law enforcement. Additionally, the project has created a machine-learning model to predict optimal resource allocation.
The project works closely with SENTRY’s Passenger-based Surface Transportation Case Study and partners with venue stakeholders, including the Williamsville Central School District, local houses of worship, the University at Buffalo Athletics, and the Amherst Police Department. These collaborative efforts aid in gathering real-world data, validating the models, and helping refine the project’s recommendations for improving the safety and security of soft targets.