Work

From Routine to Reflection: Pruning Neural Networks in Communication-Efficient Federated Learning