Learning Generative Adversarial Networks
暫譯: 學習生成對抗網絡

Key Features

• Understand the buzz surrounding Generative Adversarial Networks and how they work, in the simplest manner possible
• Develop generative models for a variety of real-world use-cases and deploy them to production.
• Contains intuitive examples and real-world cases to put theoretical concepts explained in this book to practical use

Book Description

Generative models are gaining a lot of popularity among the data scientists, mainly because they facilitate the building of AI systems that consume raw data from a source and automatically builds an understanding from it. Unlike supervised learning methods, generative models do not require labelling of the data which makes it an interesting system to use. This book will teach you all you need to know about generative models and the basics of implementing a generative adversarial network from scratch.

The book begins with the basics of generative models, as you get to know the theory behind generative adversarial networks and it's building blocks. You will understand how conditional GAN can automatically generate compatible colors for a sketch and is capable of painting hand-draw sketch with proper colors. Discover the latest approach of stacking Generative Adversarial Networks into multiple stages to decompose the problem of text to image synthesis, and develop intelligent and creative applications from a wide variety of datasets, mainly focusing on images. You will also see how to use DiscoGAN successfully transfers style from one domain to another using Tensorflow and Keras. Through this book you will be trained to build GAN models and use them in a production environment. You will be well versed with the basics of generative modelling, and learn how to use it effectively and accurately.

By the end of this book, you will be well versed with the basics of generative modelling, and learn how to use it effectively and accurately.

What you will learn

• Generate images and how to build semi-supervised model using Generative Adversarial Network(GAN)
• Use stacking with Deep Learning architecture to run and generate images from text.
• Tune GAN models by addressing the high dependency between input examples of a mini batch using Virtual Batch Normalization.
• Create data and "feed" the models by using the appropriate GAN models with python libraries Tensorflow and Keras.
• Explore the steps to deploy deep models in production