A few weeks ago I asked this question which was about how to understand GANs discriminative distribution. Now I am trying to draw a Figure like the one below, but for CycleGAN instead.

GAN distributions

In CycleGAN there are 2 Generators GAB and GBA. Generator GAB translate an image from domain A to domain B and GBA the reverse. The green solid line is the original image in domain B. The image that GAB generates is visualized as the black dashed line. The blue line is the original image in domain A and the red dashed line is the image generated by GBA (reconstructed). So the basic idea is that the generated image from A->B are approaching the distribution of the original image in B. And the same idea for the blue and red lines. This is what I draw so far:

enter image description here

Is this a correct way of visualizing the distributions? How can I add the discriminator's discriminative line? And am I missing something important?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.