FedSmart: An Auto Updating Federated Learning Optimization Mechanism
This work addresses privacy and robustness challenges in federated learning for applications with heterogeneous data, though it appears incremental as it builds on existing methods for non-IID data.
The paper tackles the problem of federated learning under non-IID data distributions and potential data poisoning, introducing FedSmart, a performance-based optimization mechanism that adjusts client weights based on local validation accuracy, resulting in improved model performance by allocating greater weight to clients with similar data distributions.
Federated learning has made an important contribution to data privacy-preserving. Many previous works are based on the assumption that the data are independently identically distributed (IID). As a result, the model performance on non-identically independently distributed (non-IID) data is beyond expectation, which is the concrete situation. Some existing methods of ensuring the model robustness on non-IID data, like the data-sharing strategy or pretraining, may lead to privacy leaking. In addition, there exist some participants who try to poison the model with low-quality data. In this paper, a performance-based parameter return method for optimization is introduced, we term it FederatedSmart (FedSmart). It optimizes different model for each client through sharing global gradients, and it extracts the data from each client as a local validation set, and the accuracy that model achieves in round t determines the weights of the next round. The experiment results show that FedSmart enables the participants to allocate a greater weight to the ones with similar data distribution.