What term refers to the process of aggregating updates from local models to a central model in Federated Learning?

The process of aggregating updates from local models to a central model in Federated Learning is referred to as Model Aggregation. In this context, federated learning allows multiple devices to train models locally on their own data while only sharing the model weights or gradients with a central server. The central server then combines these updates, which are the key contributions from each local model, into a single updated global model. This collaborative approach leverages local training data without needing to share it directly, ensuring privacy and security.

Choosing Model Aggregation highlights the importance of the collective improvement of the global model through the integration of diverse local updates, which can enhance the overall performance and adaptability of the model in real-world applications. Other terms presented do not specifically define this crucial step in the federated learning process.