5,337
edits
SinGAN: Learning a Generative Model from a Single Natural Image: Difference between revisions
SinGAN: Learning a Generative Model from a Single Natural Image (view source)
Revision as of 19:33, 5 November 2019
, 5 November 2019→Training and Loss Function
| Line 21: | Line 21: | ||
==Training and Loss Function== | ==Training and Loss Function== | ||
<math>\min_{G_n} \max_{D_n} \mathcal{L}_{adv}(G_n, D_n) + \alpha \mathcal{L}_{rec}(G_n)</math><br> | |||
They use a combination of the standard GAN adversarial loss and a reconstruction loss. | They use a combination of the standard GAN adversarial loss and a reconstruction loss. | ||
===Reconstruction Loss=== | ===Reconstruction Loss=== | ||
<math> | <math>\mathcal{L}_{rec} = \Vert G_n(0,(\bar{x}^{rec}_{n+1}\uparrow^r) - x_n \Vert ^2</math><br> | ||
The reconstruction loss ensures that the original image can be built by the GAN.<br> | The reconstruction loss ensures that the original image can be built by the GAN.<br> | ||
Rather than inputting noise to the generators, they input | Rather than inputting noise to the generators, they input | ||
<math>\{z_N^{rec}, z_{N-1}^{rec}, ..., z_0^{rec}\} = \{z^*, 0, ..., 0\}</math> | <math>\{z_N^{rec}, z_{N-1}^{rec}, ..., z_0^{rec}\} = \{z^*, 0, ..., 0\}</math> | ||
where the initial noise <math>z^*</math> is drawn once and then fixed during the rest of the training. | where the initial noise <math>z^*</math> is drawn once and then fixed during the rest of the training. | ||