In CycleGANs, we have two generators and two discriminators. Generators learn to translate images from one domain to another and a discriminator tries to discriminate between the translated images.

So, we can say the loss function of discriminator can be represented as follows:

Similarly, the loss function of discriminator can be represented as follows:

The loss function of generator can be represented as follows:

The loss function of generator can be given as follows:

Altogether, the final loss can be written as follows: