Skip to main content
. 2020 Aug 11;20(16):4485. doi: 10.3390/s20164485

Table 3.

Detailed architecture for the discriminator.

Layers Type Filter Size Output Dimension Activation Note
Input (3,64,64)
Convolution 1 3 × 3 (128,32,32) Leaky ReLU Dropout rate = 0.25
Momentum = 0.8
Dropout (128,32,32)
Convolution 2 3 × 3 (256,16,16) Leaky ReLU
Dropout (256,16,16)
Batch norm (256,16,16)
Convolution 3 3 × 3 (512,8,8) Leaky ReLU
Dropout (512,8,8)
Batch norm (512,8,8)
Convolution 4 3 × 3 (1024,4,4) Leaky ReLU
Dropout (1024,4,4)
Flatten (16384)
Output 5 (1) Sigmoid