Research Article
Generation of Human Micro-Doppler Signature Based on Layer-Reduced Deep Convolutional Generative Adversarial Network
Table 1
Parameters of generator network.
| Row | Layer name | Layer type | Attribute |
| 1 | Input noise | Image input | 1 × 1 × 100 noise vector | 2 | TConv 1 | Transposed convolutional | 512 tconv filters of size 4 × 4 with stride [2, 2] and cropping [0, 0] | 3 | BN1 | Batch normalization | — | 4 | ReLU 1 | ReLU | — | 5 | TConv 2 | Transposed convolutional | 512 tconv filters of size 4 × 4 with stride [2, 2] and cropping [0, 0] | 6 | BN2 | Batch normalization | — | 7 | ReLU 2 | ReLU | — | 8 | TConv 3 | Transposed convolutional | 512 tconv filters of size 4 × 4 with stride [2, 2] and cropping [0, 0] | 9 | BN3 | Batch normalization | — | 10 | ReLU 3 | ReLU | — | 11 | TConv 4 | Transposed convolutional | 512 tconv filters of size 4 × 4 with stride [2, 2] and cropping [0, 0] | 12 | BN4 | Batch normalization | — | 13 | ReLU 4 | ReLU | — | 14 | TConv 4 | Transposed convolutional | 512 tconv filters of size 4 × 4 with stride [2, 2] and cropping [0, 0] | 15 | tanh | Hyperbolic tangent | — |
|
|