An Experimental Machine Learning Model on Soldier Information Representation

The future landscape of military operations is anticipated to become more complex, driven by trends such as operating in urban environments and extreme climate conditions. Information technology advancements raise questions about their actual benefits versus potential cognitive overload for soldiers in the information-rich battlefield of the near future. Designing effective systems for soldiers requires understanding of information factors, modality factors, device factors, and contextual factors that all affect the technology’s effectiveness in transferring relevant information. This work proposes a digital sandbox, the
Information-Concept Experimentation Dashboard, as a potential early-stage check for innovative information concepts within
the Dutch Ministry of Defense’s innovation process, specifically focusing on soldier navigation technology. The presented approach consists of a machine learning model for information management that is grounded in a theoretical framework about human cognition and the impact of militaryrelated stressors. It addresses the brain’s limitations in processing multi-sensory information, working memory constraints, attention diversion, and the influence of stress on information processing. The model aims to evaluate various information concepts’ effectiveness in aiding soldiers efficiently in a navigation task. Reinforcement learning aims to find optimal behaviour through multiple simulations. Inspecting the learned behaviour provides
valuable insights in the potential of information concepts.