Federated Learning Model with Adaptive Differential Privacy Protection in Medical IoT

Algorithm 3

Differential private federated learning (DP-FL).

Input:

  Number of users , rounds of communication , privacy parameters set , number of users participating in each epoch of communication .

Output:

  The weight parameters of the server model.

  //Server executes.

1: Random initialize

2: for do

3:    random select n users.

4:   for in parallel do

5:    

6:    Send the updated parameters to Server

7:   end for

8:   

9:   

10:   The server broadcasts global parameters to all participants

11: end for

12: return

  //Users execute

13: function

14:   

15:   DPAGD-DNN

16:   

17: return