By Rob Albritton, Senior Director/AI Practice Lead, Ted Hallum, Senior Defense ML Engineer, and John Bellamy, Principal ML Engineer
In Part I of this series, we talked about the challenges warfighters face while accessing large quantities of data in the field, from locations that are often remote and offer limited connectivity. We also discussed how AI is transforming the way we manage data at the tactical edge. In Part II of this series, we looked specifically at Octo’s AI-powered tool, CXSearchTM, and how it transforms data dissemination in the field. Now, let’s take a look at Hatteras™, Octo’s solution that’s revolutionizing adaptive machine learning operations (MLOps) for the ever-changing landscape, especially on the battlefield.
The Issue
Warfighters, first responders, operatives, and humanitarian aids can’t afford to be without models that are in sync with the current state of the world. What’s standing in their way? We call it “drift,” something that happens when machine learning (ML) models’ performance degrades—which it does quickly, especially when conditions don’t match the patterns the models were trained to detect. To be most effective, models must be continuously retrained. With traditional ML models, retraining requires engineers to perform the model updates and DevOps Teams to redeploy them, making them unavailable to users for varying amounts of time.
The Solution
Understanding the concept of drift and the need to constantly retrain and redeploy ML models, Octo’s oLabs™ developed Hatteras, a truly revolutionary tool that saves lives and gives users an edge. Hatteras is a low-cost MLOps solution that was built with the tactical edge in mind. By embracing open source frameworks and integrating open source tools with permissive licenses, Hatteras empowers users to securely build, deploy, monitor, and maintain ML models from one single platform. It continuously monitors ML models for drift or pattern deviations which trigger automated model retraining without having to involve ML engineers or DevOps teams.
The Benefits
Hatteras keeps ML models from decaying. Since Hatteras was designed for users on the tactical edge, where data scientists and ML engineers aren’t available, Hatteras remains highly available even in austere environments. It delivers an intuitive interface that allows non-technical users to operate it wherever they may be. It provides a safety net for anyone that depends on updated, accurate ML models…on the battlefield and beyond.
What does this mean for close combat operations? Few environments are as dynamic as the battlefield where there are no data scientists, ML engineers, or DevOps engineers to work with ML models. Hatteras ensures models are maintained and redeployed, giving warfighters the advantage of timely, accurate situational awareness even under adverse, changing conditions. The result is more missions met with fewer lives lost.
For more information on how Hatteras can help your Federal Government end users or for a demonstration, contact a member of our team.