Advanced Data Analytics (ADA)
ADA excels at developing advanced decision aid technologies for customers overwhelmed by the size and complexity of their data.
ADA’s research scientists and software engineers persistently extend the state of the art in machine learning, network analytics, activity-based intelligence, and optimization. We develop innovative solutions tailored for customers to better understand and exploit their data.
Deep Machine Learning
ADA researchers have developed a variety of deep neural networks with architectures tailored to specific data and capabilities. They have been applied to detect aircraft flight anomalies, to classify objects in images and video when adversaries attempt to deceive, and to correlate dynamically changing network structures and activities with world events. We have developed:
- Recurrent neural networks (RNN) to exploit sequential data and assess dynamic phenomena
- Bayesian neural networks (BNN) to manage uncertainty and counter adversarial attacks
- Graph convolutional networks (GCN) to exploit entity-link data
- Hybrid models that combine NN architectures to exploit rich data (e.g., RNN-GCN for dynamic networks)
Shallow Machine Learning
Metron has developed a variety of supervised and unsupervised learning models and deployed them in operational systems to analyze trends, detect anomalies, discover patterns and indicators, and cue customers to investigate. Application domains of our most impactful shallow ML decision aids are cargo shipping and passenger travel (by air, sea and land).
In the early 2000’s, Metron researchers in ADA pioneered a theory of detection on networks that extends classical detection theory (of physical objects by fusing sensor measurements) to subnetworks or transactional patterns buried in entity-link data. Ever since, we have been extending those results and developing capabilities to detect complex patterns and track dynamically changing networks by fusing information gleaned from vast, noisy, incomplete data sources.
ADA Division Leadership
Christopher M. Boner, Ph.D.
I started the Advanced Data Analytics division after over a decade at Metron leading project teams to develop innovative technologies in network science, machine learning, trend analysis and anomaly detection. Several have been deployed in operational systems to enable customers to better understand, exploit, and act on large, complex data.
Gregory A. Godfrey, Ph.D.
I design novel algorithms using my training in mathematics, physics and operations research from Yale and Princeton. For over twenty years, I have led talented teams at Metron to solve diverse, challenging problems in dynamic planning / scheduling, Bayesian inference and data analytics. My recent interests use Bayesian methods to train innovative machine learning models that address uncertainty.
Data Science Career Opportunities
Metron hires scientists with experience researching novel approaches that advance the state of the art. Our researchers apply these innovations to new problem domains working alongside subject matter experts. They are familiar with machine learning tools and pipelines and work with software engineers to integrate solutions into client systems.