Generatives Deep Learning

Let your deep learning models get creative! This book demonstrates how the most innovative deep learning algorithms, such as Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), work. Aimed at creative data scientists and programmers who enjoy experimenting with code, it utilizes Python, Keras, and TensorFlow. Generative models have become one of the most exciting areas in artificial intelligence: with generative deep learning, it is now possible to teach a machine to paint, write, or even compose music—creative skills that were previously reserved for humans. With this hands-on book, data scientists can replicate some of the most impressive generative deep learning models, such as Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), encoder-decoder models, and world models. David Foster first covers the fundamentals of deep learning with Keras and illustrates how each method works before delving into some of the most advanced algorithms in the field. The numerous practical examples and tips will help you discover how your models can learn more efficiently and become even more creative.

From the content:

• Discover how Variational Autoencoders can change facial expressions in photos.
• Create practical GAN examples from scratch and use CycleGAN for style transfer and MuseGAN for music generation.
• Use recurrent generative models to generate text and learn how to enhance these models with the attention mechanism.
• Find out how generative deep learning can assist agents in performing tasks within reinforcement learning.
• Learn about the architecture of transformers (BERT, GPT-2) and image generation models like ProGAN and StyleGAN.

This book is an accessible introduction to the deep learning toolkit for generatives.