In the current global threat environment, Homeland security depends on both domain and situational awareness. The probability of a secure homeland is based on conditional probabilities describing the domain and situational awareness. These probabilities increase with the ability to deploy and utilize powerful data-driven models that make sense of the large volume of information available. With data continuing to grow in scope and complexity, organizations require innovative strategies, services, and technologies to unlock the value of their data analytics potential. Data can be used to make more insightful, forward-looking decisions about readiness, logistics, personnel, intelligence, and a host of other critical mission concerns.
The use of highly skilled technological capabilities coupled with a new generation of advanced predictive analytics offers government organizations the opportunity to take advantage of one of their most valuable resources, data, and then turn that data into action.
Our paper describes our approach for building predictive models using predictive analytics supporting faster and better critical decision making, using a set of modern machine learning operations (MLOps) best practices geared to help a user fulfill its various missions and can easily adapt to accommodate changes in mission priorities. A direct result of predictive model research is providing real-time predictive analytics to decision-makers for faster and better models. This is achieved via a suite of deployed predictive models, which are trained and continuously improved using open-source custom AI/ML pipelines utilizing supervised and deep model learning on large and varied data sets. The models could also be used as stand-alone tools to predict future trends/events and provide support for tactical, resource allocation decisions.
Keywords
AI, ANALYTICS, DATA, MACHINE LEARNING, MODELING, OPERATIONAL ENVIRONMENT, PROBABILITY, SECURITY, TECHNOLOGY
Additional Keywords
Predictive Analytics, Predictive Models