Federated Learning on Non-iid Data via Local and Global Distillation

TL;DR

This paper introduces FedND, a federated learning approach utilizing local and global knowledge distillation to improve performance and communication efficiency in non-iid data scenarios.

Contribution

It proposes a novel federated learning method combining self-distillation and noisy sample distillation to enhance model training on non-iid data.

Findings

Achieves superior performance over existing methods.

Reduces communication costs.

Effective in non-iid data environments.

Abstract

Most existing federated learning algorithms are based on the vanilla FedAvg scheme. However, with the increase of data complexity and the number of model parameters, the amount of communication traffic and the number of iteration rounds for training such algorithms increases significantly, especially in non-independently and homogeneously distributed scenarios, where they do not achieve satisfactory performance. In this work, we propose FedND: federated learning with noise distillation. The main idea is to use knowledge distillation to optimize the model training process. In the client, we propose a self-distillation method to train the local model. In the server, we generate noisy samples for each client and use them to distill other clients. Finally, the global model is obtained by the aggregation of local models. Experimental results show that the algorithm achieves the best performance and is more communication-efficient than state-of-the-art methods.