Federated Learning

Overview

Federated learning is a machine learning approach that enables model training across multiple data-holding institutions without requiring direct transfer of the underlying data. Each participant trains on local data and contributes model updates (gradients or weights) rather than raw records. A central coordinator aggregates the updates into an improved global model. In the pharmaceutical context this architecture is particularly valuable because it preserves proprietary compound libraries, clinical records, and patient data while enabling cross-institution model improvement.