An Adaptive Communication-Efficient Federated Learning to Resist Gradient-Based Reconstruction Attacks

Table 5

The evaluation metrics of the GAN-based attacks with the MNIST dataset in different compression.

Method	Iterations	CE	Communication frequency	SSIM	MSE
GAN	500	1	1	0.6248	37.3750
GAN	500	0.97	1	0.6138	40.1095
GAN	500	0.95	1	0.6107	45.3234
GAN	500	0.93	1	0.5861	45.6194
GAN	500	0.90	1	0.6144	39.9441