Noblis has received a U.S. patent for a machine learning-based method generating models that can imitate different security checkpoint setups and show how each configuration responds to various kinds of threat.
The Reston, Virginia-based nonprofit organization said Thursday its iterative modeling system employs algorithms based on concepts from rule-based adversarial games.
Noblis researchers particularly utilized chess game rules to allow the system to determine the possible outcomes of various potential attacks through security checkpoints.
Brian Lewis, team lead and co-inventor at Noblis, said the approach could help decision-makers calculate return on investment should they decide to implement changes to a certain security configuration.
The tool was developed under the organization’s internally sponsored research program.