Table 5 Optimizers & FL architecture.
From: Privacy preserving skin cancer diagnosis through federated deep learning and explainable AI
Category | Parameter/method | Description |
|---|---|---|
Optimizer 1 | AdamW | weight decay control |
Optimizer 2 | SGD | Standard optimizer with momentum |
Optimizer 3 | RMSprop | moving average of squared gradients |
No. of clients | 3 | Each client trains on its local skin cancer dataset |
Communication rounds | 25 | Number of global aggregation rounds |
Local training epochs | 20 | Training epochs per client before sending updates |
Aggregation algorithm | FedAvg | Average model weights from clients to update the global model each round |
