Context Independent Anomaly Detection for Enhanced Decision Making

Today's intelligence, surveillance, and reconnaissance sensors offer a diverse set of raw information that spans the electromagnetic spectrum. While the goal of these sensors is to increase situational awareness resulting in effective threat recognition, the realization of this objective is often impossible to achieve because the process of searching the data for a potential threat is predicated on the analyst having some predefined understanding of how a threat is represented in the ISR data. The challenge of Context Independent Anomaly Detection is to replace the current paradigm of analyst-intensive review of vast amounts of ISR data with an innovative approach that processes the ISR data in an unsupervised manner to identify anomalies that can be reported to the war fighter in real-time with a minimum amount of processing power. The proposed research will leverage Hilbert Technology's patented Hilbert Engine to develop an automated method for converting ISR data into numerical representations, mathematically processing the transformed data to identify anomalies, and then converting the anomalous data back to the ISR data space for use with tactical decision aids. The solution will operate in real-time and require modest processing resources.