Federated Learning Model with Adaptive Differential Privacy Protection in Medical IoT

Algorithm 1

Different private federated learning deep neural network with adaptive gradient descent (DPFLAGD-DNN).

Input: Training dataset, loss function , privacy budget , gradient normal clipping , budget increase rate

Output: Weight parameters

1:  Receive from server

2:  Initialization: , random

3:  for do

4:    Compute the average relevance

5:    Compute the relevance ratio

6:    Calculate the privacy budget

7:    

8:    

9:    

10:  end for

11:  while

12:    

13:    

14:    

15:    

16:    

17:  end while

18:  return