Case Study: U.S. Navy reduces ML model update time by 97% with Fiddler AI

U.S. Navy reduces ML model update time by 97% with Fiddler

The U.S. Navy's NAVSEA PMS 408 office faced a challenge maintaining the performance of machine learning models used for underwater explosive ordnance disposal. Their automatic target recognition (ATR) models, essential for unmanned underwater vehicles, were difficult to monitor and update, risking operational readiness as threats evolved. They partnered with Fiddler and the Defense Innovation Unit to address this.

With the Fiddler AI Observability platform, they built a robust MLOps pipeline for the AMMO project. This solution provided monitoring and explainability, drastically speeding up retraining and deployment. Fiddler's work helped reduce ML model update time by 97%, improved model explainability, and enabled the successful transition of the prototype to a production system with the Naval Information Warfare Center Pacific.

U.S. Navy