Figure 1
From: A comparative study of federated learning methods for COVID-19 detection
Illustration of FL models and algorithms: (a) Federated averaging, where clients train on a local batch of data. (b) FedSGD, in which a subset of clients is selected, and each performs a single step of SGD before sending model updates to the server. (c) Cyclic Weight Transfer (CWT), where clients train locally and pass the model to the next client, repeating the cycle. (d) Single Weight Transfer (SWT), where the model passes through each client only once. (e) Stochastic Weight Transfer (STWT), in which the model is sequentially passed through clients, with participating clients in each round being sampled randomly.
