Maritime Situational Awareness - Deep Learning Aggregated Reference Knowledgebase (MSA-DARK)

The Maritime Domain, specifically in the littoral zones, is becoming increasingly congested with cooperative and non-cooperative maritime traffic making it difficult for a commander to obtain a Recognized Maritime Picture (RMP) or maintain Maritime Situational Awareness (MSA). Machine Learning (ML) can be applied to help rapidly and decisively make informed decisions on the tactical relevance of AoR and theater events but the shifting contextual relevance of the input parameters still puts a significant cognitive burden on operators.  The solution proposed here, the Maritime Situational Awareness - Deep Learning Aggregated Reference Knowledgebase (MSA-DARK) will use ML in a layered approach to execute supporting big data analytics and ML algorithms against tactically relevant Situational Awareness (SA) data sources available to Minotaur Family of Systems (MFoS) and other C5ISR applications. Through innovative ML algorithms coupled with automation, algorithms, and big data analytics, MSA-DARK will use ML to govern and monitor contributing ML models that mine big data sources for information that MSA applications can then use as reference knowledge to assess the tactical relevance of threads of information from imagery, acoustics, environmental systems, sensor systems, signals intelligence, and even historical habits (patterns-of-life). The resulting solution will aggregate and extract curated knowledge from tactical and theater -level sources by fusing tactical sensors with national data to improve analysts’ operating efficiency to support synchronized actions in the maritime environment.  This will provide coherent battlespace awareness that will help commanders understand their operational environments, discover and predict the actions of targets or threats, and help devise Courses of Action (COA) to assess or defeat them. To accelerate development and increase the chances of success, this team consisting of PMAT and Mosaic ATM will start by using platforms and ML algorithms that have already achieved a demonstrated level of integration with the MFoS.