Document Type

Article

Publication Date

August 2024

Patent Number

12067857

CPC

G08B 21/02 , G06F 18/217 , G06N 3/04 , G06N 3/08 , G06Q 90/20

Abstract

A method for reducing a threat footprint for an area-of-interest by, determining a base risk score for an area-of-interest, deriving a graphical representation of the area-of-interest and one or more defense strategies for the area-of-interest, training an adversarial reinforcement learning agent to identify one or more successful attack paths on the graphical representation of the area-of-interest that are effective in bypassing the one or more defense strategies for the area-of-interest, determining an active risk score based upon the one or more successful attack paths identified by the adversarial reinforcement learning agent and determining one or more mitigation actions that will reduce the active risk score for the area-of-interest based upon the base risk score and one or more predetermined resource constraint.

Application Number

17/559401

Assignees

University of South Florida

Filing Date

12/22/2021

Share

COinS