Researchers developed a system that streamlines the process of federated learning, a technique where users collaborate to train a machine-learning model in a way that safeguards each user’s data. The system reduces communication costs of federated learning and boosts accuracy of a machine-learning model trained using this method, which would make federated learning more feasible to implement in real-world settings.
Click here for original story, Collaborative machine learning that preserves privacy
Source: ScienceDaily