Hands-On Generative Adversarial Networks with PyTorch 2.x

With examples that cover computer vision, NLP and computer graphics, this book aims to become the practical guide for beginners in machine learning to understand the fundamental principles of GANs and learn the efficient applications of PyTorch library. The explanations of underlying mathematics behind the models also deliver in-depth knowledge to the advanced readers who wish to use GANs and PyTorch professionally.

Key Features

Explain the fundamental structures and principles of more than 22 GAN models
Showcase various image/audio/text synthesis and image-to-image/text-to-image translation tasks with GANs
More than just GAN: enriched with presentation of traditional approaches as well as similar deep learning techniques

Book DescriptionComputer vision is one of the most actively researched fields in deep learning. Aside from image classification and object recognition, the tasks of image synthesis have become an intriguing way to attract the attention of beginners in machine learning. GANs have been proved to be successful in generating realistic images over the last few years. As PyTorch is becoming the most popular open-source library in deep learning, a practical guide for building GAN models with PyTorch can be very useful for readers.
In this book, we start with a quick example of building a simple GAN model with pure Python and thorough demonstrations of key features in PyTorch. We dive into a classical GAN model for image synthesis while giving clear explanation of its mechanism, before we move on to more interesting applications including image-to-image translation, image restoration, text synthesis, text-to-image translation, audio synthesis and 3D model generation. We also give thorough explanations of underlying mathematical principles, related traditional approaches and useful techniques in deep learning along the course, which may help advanced readers professionally.What you will learn

Build a simple GAN model with pure Python
Key features in the latest version of PyTorch 2.x
Understand how computational graphs are built for neural networks
Perform image synthesis, image style transfer and image restoration
Generate images based on label prompts or text descriptions
Handle NLP tasks by generating audio and text with GANs
Synthesize texture image and produce 3D models
Perform adversarial attack and data augmentation with GANs
Useful tricks and techniques in deep learning

Who this book is forThis book is for anyone looking to do creative work in machine learning, especially in deep learning. It is required for the reader to have programming experience in Python. Those who are familiar with the concepts of machine learning, deep learning and computer vision may find it much easier to follow our course. Do not worry if you haven't used PyTorch before, because we will explain everything you need to know about PyTorch. It is highly recommended to have access to medium-to-high end NVIDIA graphics card to save you much time waiting for the model results.