As the use of artificial intelligence (AI) and machine learning (ML) continues to grow across industries, their applications in the defense community become more and more profound. Specifically relating to adaptive military training, applying the insights gleaned through AI and ML makes training today’s warfighters more effective than ever. However, like any AI or ML application, the data set has to be accurate and relevant to do its job properly.
A recent whitepaper titled “The Value of Cognitive Workload in Machine Learning Predictive Analytics” explores the development of deep neural networks based on training data sets to predict future states of student performance. This whitepaper was authored by Amy Dideriksen, Global Training Research Manager in Mission Systems at Collins Aerospace; Joseph Williams, Senior Software Engineer in Collins Aerospace’s Advanced Technologies Data Analytics group; Dr. Thomas “Mach” Schnell, Professor in Industrial and Mechanical Engineering at the University of Iowa and director and chief test pilot of the Operator Performance Laboratory (OPL); and Gianna Avdic-McIntire, Senior Project Manager for Data Analytics in Mission Systems at Collins Aerospace.
The study’s authors observed that “when cognitive workload is included in our deep neural networks, it increased the performance prediction to an extremely high level of accuracy.” This means that the ability to capture cognitive workload data points and incorporate them into an AI/ML training algorithm sheds a lot of light on a warfighter’s training patterns and predictability. In turn, those insights power a greater ability for robust adaptive learning programs. With the ability to predict student performance in real-time, training content can be more effectively modified to maintain student engagement, aiding in the transfer and retention of requisite skills.
In the whitepaper’s introduction, the authors note, “It is believed that by enhancing training in real-time, through adaptive learning techniques, training facilities can personalize training to increase the quality and speed of learning. The industry is beginning to include human system metrics in the adaptive learning solution. With the increased interest in big data, we have seen an increase in commercial sensors to collect and report data in real-time. However, understanding what data to collect and how to appropriately use that data is still being proven through evidence-based research.”
This paper moves to fill in those blanks about appropriate training data capture and how AI and ML can be optimized in those methods.
Download the full whitepaper in the Modern Military Training Resource Center here.