| `self.D_B_loss_real` | 在对域`B`中的真实图像进行分类时,鉴别符`D[B]`的二进制交叉熵损失。(This loss is to be minimized with respect to the parameters of the discriminator `D[B]`) |
| `self.D_B_loss_fake` | 在对域`B`中的伪造图像进行分类时,鉴别符`D[B]`的二进制交叉熵损失。(This loss is to be minimized with respect to the parameters of the discriminator `D[B]`) |
| `self.D_A_loss_real` | 在对域`A`中的真实图像进行分类时,鉴别符`D[A]`的二进制交叉熵损失。(This loss is to be minimized with respect to the parameters of the discriminator `D[A]`) |
| `self.D_A_loss_fake` | 在对域`A`中的伪造图像进行分类时,鉴别符`D[A]`的二进制交叉熵损失。(This loss is to be minimized with respect to the parameters of the discriminator `D[A]`) |
| `self.loss_GABA` | 通过两个生成器`G[AB]`和`G[BA]`将域`A`中的图像映射到`B`,然后再映射回`A`的重建损失 )加上假图片`G[AB](x[A])`的二进制交叉熵,并由域`B`中的鉴别器标记为真实图像。(This loss is to be minimized with respect to the parameters of the generators `G[AB]` and `G[BA]`) |
| `self.loss_GBAB` | 通过两个生成器`G[BA]`和`G[AB]`将域`B`中的图像映射到`A`,然后再映射回`B`的重建损失 加上伪图片`G[BA](x[B])`的二元交叉熵,由域`A`中的鉴别器标记为真实图像。(This loss is to be minimized with respect to the parameters of the generators `G[AB]` and `G[BA]`) |