Jane Edwards
IBM’s government technology subsidiary Octo will continue to conduct research and develop a machine learning operations prototype for the U.S. Army through an option year exercised by the service branch under a previously awarded broad agency announcement.
Octo said Tuesday it will work with the Army Combat Capabilities Development Command Army Research Laboratory to build methods for developing an active learning pipeline designed to automatically retrain ML models in austere environments.
The company will perform R&D work onsite with ARL and at its innovation center, oLabs, in Reston, Virginia.
Rob Albritton, vice president of oLabs, said Octo’s MLOps platforms work to improve the survivability of soldiers in environments where cloud-based systems are hard to field due to connectivity issues.
“Our solution also addresses ML model drift by automatically retraining ML models to adjust to shifting conditions at the tactical edge. This is not a ‘nice to have’ but a necessity that helps enable our warfighters succeed no matter where they serve,” added Albritton.
Register here to attend the Potomac Officers Club’s 10th Annual Defense R&D Summit on Jan. 31 to hear government leaders and industry experts discuss the latest developments in military technology.
