APCSMA: Adaptive Personalized Client-Selection and Model-Aggregation Algorithm for Federated Learning in Edge Computing Scenarios.

From the analysis of the comparative data on accuracy and loss function values for each algorithm under three different data-distribution conditions, as presented in the above figures, a clear trend can be observed: the greater the degree of dispersion in the data distribution among clients, the more significant the impact on the training of the global model, resulting in a notable performance degradation in edge computing scenarios compared to an ideal environment.

also in 19,460 other papers