Research Areas

CMIS is currently focusing on the following thrust areas in one- and in multi-domain (maritime, land, air, space, and cyber) environments:

• Orchestrated resource management translates the current situation into coordinated tasks to collect information from multiple sources that are estimated to be of the highest intelligence value. CMIS has been investigating how machine learning algorithms can be used to cue sensors and how to achieve optimal dynamic sensor adaptation.
• Object Detection employs machine learned and expert-learned models to detect object (signals, physical entities, etc.). CMIS has applied convolutional neural networks, and other matched filter detectors for imagery and signals.
• Situation Detection reasons about evidence and allows for computational models of situations and activities to be built for physical and nonphysical target systems that help with situation awareness and decision making. Distributed Fusion Architecture enables automated global collection that adapts to threat environment.

Research Outcomes

Student Theses and Dissertations

• A spatiotemporal clustering approach to maritime domain awareness
• Traffic pattern detection using the Hough transformation for anomaly detection to improve maritime domain awareness
• Generalized Hough Transform for object classification in the maritime domain
• Determination of high-speed multiple threat using Kalman filter analysis of maritime movement
• Improving maritime domain awareness using neural networks for target of interest classification
• Encounter detection using visual analytics to improve maritime domain awareness
• Data mining of extremely large ad hoc data sets to produce inverted indices
• Maritime domain awareness by anomaly detection leveraging track information
• Cluster computing for automated network analysis at scale
• Machine learning of extremely large sets of signal collections using cluster computing
• Object detection in low-spatial-resolution aerial imagery using convolutional neural networks
• Cluster-based spectral-spatial segmentation of hyperspectral imagery
• Identification and classification of signals using generative adversarial networks
• Analysis of image enhancement algorithms for hyperspectral images
• Application of algorithm learning to identify and mitigate concept drift