Towards Model-Agnostic Federated Learning over Networks
We present a model-agnostic federated learning method for networks of heterogeneous data and models. The network structure reflects similarities between the (statistics of) local datasets and, in turn, their associated local("personal") models. Our method is an instance of empirical risk minimization, with the regularization term derived from the network structure of data. In particular, we require well-connected local models, forming clusters, to yield similar predictions on a common test set. The proposed method allows for a wide range of local models. The only restriction on these local models is that they allow for efficient implementation of regularized empirical risk minimization (training). For a wide range of models, such implementations are available in high-level programming libraries including scikit-learn, Keras or PyTorch.
PDF Abstract
