Non-Independent and Identically Distributed

A setting in which data distributions differ significantly across participating parties. It is a core challenge in federated learning, making aggregation of a global model difficult.