A Data Augmentation Method for Vertical Federated Learning

Algorithm 1

FedDA algorithm of part A. We use default values of =10, =5, and =0.0001.

Input datasets and , neural networks , , , a random number [0,1], epoch number , the batch size , Adam hyperparameters , intitial cirtic parameters , initial generator parameters .

1 fordo.

2  fordo.

3   fordo.

4    Sample ,a batch from the ;

5    Sample , a batch from the ;

6    Learn through ;

7    ;

8    =;

9    ;

10      Sample real data ;

11      Latent variable ;

12      ;

13      

14   end for

15   

16  end for

17  fordo

18   Sample , a batch from the ;

19   Learn through ;

20   ;

21   ;

22 end for

23

24 end for