Exercises

Try answering and working on the following exercises to expand the knowledge that you've gained from this chapter:

1. What is the adversarial training process?
2. Write the value function of the min-max game that the Discriminator and Generator are playing.
3. Explain why the min-max value function formulation can saturate in the early training step of training.
4. Write and explain the non-saturating value function.
5. Write the rules of the adversarial training process.
6. Are there any recommendations on how to feed a condition to a GAN?
7. What does it mean to create a conditional GAN?
8. Can only the fully connected neural networks be used to create GANs?
9. Which neural network architecture works better for the image generation problem?
10. Update the code of the Unconditional GAN: Log the Generator and Discriminator loss value on TensorBoard, and also log matplotlib plots.
11. Unconditional...