Generative AI is the most spoken of AI direction in media nowadays, and this book is aimed at assisting you in becoming an expert in its most well-established class of models - Generative Adversarial Nets.
With the help of this book, you will work your way up from understanding the basic components and architecture of GANs, building your first model from scratch to designing, building, training and optimizing a wide variety of these powerful models. You will go way beyond theoretical knowledge and gain hands-on experience in finding the right type of GAN for each specific problem using PyTorch examples provided in every chapter.
You will cover important image-generation and translation architectures such as classic and conditional GANs, DCGANs, StyleGANs, CycleGANs, and pix2pix. Learn to synthesize sequences, text and audio, and generate videos. Finally, we will dive into the state-of-the-art hybrid models of GANs with other generative models.
By the end of this book, you will be an expert in practical applications of GANs to real-world problems.