Enhancing Effectiveness through the Fusion of Machine Learning, VC Simulation, and Agent-Based Modeling

We introduce an approach that synergizes Artificial Intelligence (AI) and Virtual Constructive (VC) simulation to improve training and mission design. The first phase is centered on the simulation of a jet fighter pilot's performance, emphasizing scenario variability and sensitivity analysis, and identifying critical success factors, demonstrating the value of VC simulation and its ability to generate data. Simulations consisted of one to three jet fighters (there is always one jet fighter with a human pilot, and the other are AI-based), engaged in attempting to destroy control towers that are defended by a radar-guided missile defense system that can launch counterstrikes. Comprehensive data collection is accomplished which is instrumental in training a Convolutional Neural Network (CNN). The second phase uses the developed CNN to augment the decision-making process. A second agent-based modeling environment is introduced for evaluation purposes and includes various agents (AI-driven fighter pilots, radars, defense systems, and missiles). The developed environment that has a degree of similarity with the original one facilitates comprehensive testing across various scenarios and making decisions. The integration of VC simulations and machine learning captures an operator's capabilities. It facilitates the transfer of these capabilities into other environments, allowing for the development of a more complex environment to evaluate new training and resource allocation.