Abstract: Conventional federated learning (FL) assumes the homogeneity of models, necessitating clients to expose their model parameters to enhance the performance of the server model. However, this ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results