Using CNN together with GANs results in a faster capacity to learn. Let's look at a number of practical examples of different epochs to understand how the machine will learn to improve its writing process. For instance, compare the results achieved after four iterations in the following recipe with the results achieved after four iterations in the previous recipe. Do you see the difference? I wish that I could learn this art myself!
Epoch 0 |
Epoch 2 |
Epoch 4 |
Epoch 8 |
Epoch 16 |
Epoch 32 |
Epoch 64 |
Epoch 128 |
Epoch 200 |
Example of forged MNIST-like with DCGAN