Meta Knowledge Condensation for Federated Learning

Ping Liu, Xin Yu, Joey Tianyi Zhou

Existing federated learning paradigms usually extensively exchange distributed models at a central solver to achieve a more powerful model. However, this would incur severe communi…