Improved Anti-Submarine Warfare Training Through Interpretable Machine Learning

Training Acoustic Warfare Operators (AWO) for Anti-Submarine Warfare (ASW) is a time-consuming process that requires extensive human support. But instructors must divide their time between helping students and hunting for the signal in sonograms themselves. To improve training outcomes, nou Systems, Inc. (nSI) will deliver an AI training system that is interpretable, not explainable after the fact. While many AI models' decisions are notoriously hard to explain, recent advances allow models that directly show what aspect of data they're using to decide. In this innovative Phase I effort, the nou Systems team will (a) architect the first interpretable AI model to detect submarines in sonograms; (b) combine it with an interpretable model that captures the instincts that skilled AWOs use when considering sonograms; and (c) architect the entire machine learning (ML) pipeline using modern best practices. The result: an AI model that detects subs and explains its reasoning to support training AWOs.