Jargon-Aware Artificial Intelligence

SSC will leverage the ability for deep neural networks to perform feature extraction along with transparent classifier architectures, which are innately explainable to create an AI assistant for passive air anti-submarine warfare applications. Deep neural networks will be used to learn the features operators use when reviewing sonar displays. This embedded expert knowledge will enable the AI assistant to analyze a sonar display and explain it in terms, which are understandable by domain practitioners. A transparent classifier can then take those features and provide fuzzy classifications, and can explain why the classification was made based on features already understood by domain practitioners. This technology will be extended to be used as an self-updating tool for operators.