User-Level Membership Inference for Federated Learning in Wireless Network Environment

Algorithm 1

Participant’s attack procedure.

Input: The GANs iteration round , the federated learning model (), generator , discriminator .

Output: The generated dataset :(x,y) and the inference result ‘IN’ or ‘OUT’.

1: Procedure Adversary Execution.

2: Initialize

3: Set

4: for (;;) do

5: Run to generate sample

6: Update based on Eq (8)

7: end for

8:

9: Output:

10:

11: Attack Phase:

12: Train CNN model using dataset.

13: Perform membership inference attack against dataset.

14: Compare the inference results with the claimed information.

15: Output: Mark every record as ‘IN’ or ‘OUT’, where ‘IN’represents the Victim’s training sample.