TOPMAST: Training Operational Performance via Measure Automation and Scenario-generation Technology

As the number of emerging warfare capabilities increases, adaptive training is required in order to keep pace with the corresponding increase in the scale and complexity of training regimes. To enable truly adaptive training, training systems need to both (1) automatically generate training scenarios and (2) automatically assess trainee performance. Lastly, (3) diagnostic visualization technologies must enable instructors and analysts to validate the outputs and processes of the complex automation technologies. TOPMAST (Training Operational Performance via Measure Automation and Scenario-generation Technology) is an adaptive training solution that will repurpose tactical training data to enable adaptive training at scale. TOPMAST will enhance existing simulation technologies to automate requisite adaptive training capabilities (challenge 1). Bayesian statistics and a supervised reinforcement learning framework will automatically discover which Measures of Performance are most predictive of a trainees mission performance (challenge 2). Advanced, high-dimensional data visualizations will provide a user-friendly view of the automated measurement discovery processes and will empower instructors and analysts with the knowledge of the automated systems reasoning (challenge 3). This understanding will ensure instructors remain in control of tactical training despite their usage of automated technologies.