Gregory A. Godfrey

Director, ADA

Dr. Godfrey joined Metron in 1998, and has served as a Principal Investigator (PI) or Technical Lead on a wide range of research projects for organizations such as the Defense Advanced Research Projects Agency (DARPA), Office of Naval Research (ONR), National Aeronautics and Space Administration (NASA), National Geospatial Intelligence Agency (NGA), and the Intelligence Community. His technical emphasis has been in developing algorithms in multiple domains to support data analytics, machine learning, Bayesian inference, and related optimization problems.

As the Director of Metron’s Advanced Data Analytics division, he manages a team of computer scientists and mathematicians with a wide range of software language skills (e.g., MATLAB, Java, Python, MapReduce, Spark, C++) and framework expertise (e.g., Service-Oriented Architectures, Hadoop). Dr. Godfrey oversees Metron’s Cloudera-based computing cluster of twenty machines with more than 600 cores and over one TB of shared memory to support customized software solutions for Customs and Border Protection (CBP), ONR, NGA, and NASA. For ONR and NGA, Metron supplements processing capabilities as needed with Amazon Web Services (AWS)-based virtual machines.

Dr. Godfrey is the PI on a three-year effort with ONR to design more robust deep neural network (DNN) training methods to protect against deception and other adversarial attacks. He is leading an internal Metron effort to combine deep reinforcement learning (DRL) techniques with Approximate Dynamic Programming (ADP)-based algorithms to design more effective hybrid learning algorithms. For DARPA’s Fundamental of Design (FUN DESIGN) program, Dr. Godfrey leads the development of novel mathematical representations, algorithms and software to optimize the conceptual design of mechanical systems from functional building blocks. He also serves as an optimization consultant on multiple DARPA programs, including CASCADE, PROTEUS, and CODA.

Dr. Godfrey has led NASA SBIR Phase I, II, and III efforts on flight anomaly detection to improve safety and efficiency in the national airspace. Metron demonstrated its Normalcy Score Broker (NSB) technology, which uses historical flight data to develop models of normal behavior and then applies statistical methods in Spark to combine multiple features into a single score to identify outliers. Metron has used this same NSB technique to develop operational systems for customers in the land and maritime domains. These algorithms are being integrated and evaluated within NASA’s Air Traffic Management – eXploration (ATM-X) distributed simulation testbed environment prior to operational transition.

In support of ONR, Dr. Godfrey has developed activity and anomaly detection algorithms using vehicle tracks derived from wide-area surveillance sources. The approach extends classical likelihood ratio detection theory, which has had tremendous success detecting physical objects for military applications, to detect activities and events. Metron has transitioned the resulting Threat Detection Suite (TDS) software to NGA Research, where it is used to support Activity-Based Intelligence studies using archival repositories of wide-area derived track data. Dr. Godfrey is currently managing a related project with ONR / NGA called “Cloud-Based Fusion of Multi-INT Data” to integrate and to extend Metron’s track analytics library into NGA’s distributed AWS-based cloud computing environment.

Under DARPA funding, Dr. Godfrey designed the algorithms for and managed the software development of multi-agent technology in two separate domains: coordinated unmanned aerial vehicle (UAV) search and surveillance and collaborative airlift planning. In the UAV domain, he developed negotiation mechanisms that enable distributed coordination of a fleet of autonomous UAVs performing search (detection) and surveillance (monitoring) on mobile ground targets.

Education

  • Doctor of Philosophy, Operations Research, Princeton University, 2000
  • Master of Arts, Operations Research, Princeton University, 1994
  • Bachelor of Science, Mathematics, Yale University, 1991
  • Bachelor of Science, Applied Physics, Yale University, 1991