Contrastive encoder pre-training-based clustered federated learning for heterogeneous data