FLUID Framework Enables Federated Learning for Biomedical AI Between Earth and ISS Using Real and Synthetic Data

Jan 19, 2025, 05:51 PM

A new framework named FLUID (Federated Learning Using In-space Data) has been developed to enable federated learning for training biomedical machine learning models in a spaceflight context. This foundational architecture facilitates the training and updating of classifier models between Earth and the International Space Station (ISS) using both real biomedical research data and synthetically generated data. The initiative represents a significant advancement in the application of AI technologies in space, allowing for decentralized data training without the need for centralized data storage.