Effectiveness of Incorporating Adversary Probability Perception Modeling in Security Games
Abstract
We attempt to further improve the Subjective Utility Quantal Response (SUQR) model used to handle boundedly rational human adversaries in Stackelberg security game (SSG) algorithms. Given recent work on human decision-making, we adjust the existing subjective utility function to account for adversary probability perceptions of defender coverage. We then compare the predictive accuracy of this expanded model to the existing SUQR model utilizing data from previous security game experiments with human subjects. Our results show the incorporation of probability perceptions into the SUQR can provide improvements in the ability to predict probabilities of attack in certain games.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 30, 2015
- Accession Number
- AD1015545
Entities
People
- Adam D. Ackerman
- Elizabeth Carpenter
- Robert Gutierrez
- Sierra Kelly
Organizations
- United States Air Force Academy