In previous chapters, we covered the idea of using adversarial learning to generate simple signals with NumPy and learned about the new features and capabilities of PyTorch 1.3. It's time for us to use PyTorch to train a GAN model for generating interesting samples.
In this chapter, we will introduce you to a classic and well-performing GAN model, called DCGAN, to generate 2D images. You will learn the following:
- The architecture of DCGANs
- The training and evaluation of DCGANs
- Using a DCGAN to generate handwritten digits, human faces
- Having fun with the generator network by performing image interpolation and arithmetic calculation on the latent vectors to change the image attributes
By the end of this chapter, you will have grasped the core architecture design of GAN models for generating image data and have a better understanding of...
