Training Effective Neural Networks on Structured Data with Federated Learning

Federated Learning decreases privacy risks when training Machine Learning (ML) models on distributed data, as it removes the need for sharing and centralizing sensitive data. However, this learning paradigm can also influence the effectiveness of the obtained prediction models. In this paper, we specifically study Neural Networks, as a powerful and popular ML model, and contrast the impact of Federated Learning on the effectiveness compared to a centralized approach - when data is aggregated at one place before processing - to assess to what extent Federated Learning is suited as a replacement. We also analyze the effect of non-independent and identically distributed (non-iid) data on effectiveness and convergence speed (efficiency) of Federated Learning. Based on this, we show in which scenarios (depending on the dataset, the number of nodes in the setting and data distribution) Federated Learning can be successfully employed.