book

GANs in Action

Name: GANs in Action
ISBN: 9781617295560

by Jakub Langr, Vladimir Bok

September 2019

Intermediate to advanced

240 pages

7h 30m

English

Manning Publications

Read now

Unlock full access

GANs in Action: Deep learning with Generative Adversarial Networks
Jakub Langr and Vladimir Bok
Copyright
Dedication
Brief Table of Contents
Table of Contents
Preface
Jakub LangrVladimir Bok
Acknowledgments
Jakub LangrVladimir Bok
About this book
Who should read this bookAbout the codeliveBook discussion forumOther online resourcesHow this book is organized: a roadmapPart 1, “Introduction to GANs and generative modeling”Part 2, “Advanced topics in GANs”Part 3, “Where to go from here”About the authors
About the cover illustration
Saint-Sauveur
Part 1. Introduction to GANs and generative modeling

Chapter 1. Introduction to GANs
1.1. What are Generative Adversarial Networks?1.2. How do GANs work?1.3. GANs in action1.3.1. GAN training1.3.2. Reaching equilibrium1.4. Why study GANs?Summary
Chapter 2. Intro to generative modeling with autoencoders
2.1. Introduction to generative modeling2.2. How do autoencoders function on a high level?2.3. What are autoencoders to GANs?2.4. What is an autoencoder made of?2.5. Usage of autoencoders2.6. Unsupervised learning2.6.1. New take on an old idea2.6.2. Generation using an autoencoder2.6.3. Variational autoencoder2.7. Code is life2.8. Why did we try aGAN?Summary
Chapter 3. Your first GAN: Generating handwritten digits
3.1. Foundations of GANs: Adversarial training3.1.1. Cost functions3.1.2. Training process3.2. The Generator and the Discriminator3.2.1. Conflicting objectives3.2.2. Confusion matrix3.3. GAN training algorithm3.4. Tutorial: Generating handwritten digits3.4.1. Importing modules and specifying model input dimensions3.4.2. Implementing the Generator3.4.3. Implementing the Discriminator3.4.4. Building the model3.4.5. Training3.4.6. Outputting sample images3.4.7. Running the model3.4.8. Inspecting the results3.5. ConclusionSummary
Chapter 4. Deep Convolutional GAN
4.1. Convolutional neural networks4.1.1. Convolutional filters4.1.2. Parameter sharing4.1.3. ConvNets visualized4.2. Brief history of the DCGAN4.3. Batch normalization4.3.1. Understanding normalization4.3.2. Computing batch normalization4.4. Tutorial: Generating handwritten digits with DCGAN4.4.1. Importing modules and specifying model input dimensions4.4.2. Implementing the Generator4.4.3. Implementing the Discriminator4.4.4. Building and running the DCGAN4.4.5. Model output4.5. ConclusionSummary
Part 2. Advanced topics in GANs
Chapter 5. Training and common challenges: GANing for success
5.1. Evaluation5.1.1. Evaluation framework5.1.2. Inception score5.1.3. Fréchet inception distance5.2. Training challenges5.2.1. Adding network depth5.2.2. Game setups5.2.3. Min-Max GAN5.2.4. Non-Saturating GAN5.2.5. When to stop training5.2.6. Wasserstein GAN5.3. Summary of game setups5.4. Training hacks5.4.1. Normalizations of inputs5.4.2. Batch normalization5.4.3. Gradient penalties5.4.4. Train the Discriminator more5.4.5. Avoid sparse gradients5.4.6. Soft and noisy labelsSummary
Chapter 6. Progressing with GANs
6.1. Latent space interpolation6.2. They grow up so fast6.2.1. Progressive growing and smoothing of higher-resolution layers6.2.2. Example implementation6.2.3. Mini-batch standard deviation6.2.4. Equalized learning rate6.2.5. Pixel-wise feature normalization in the generator6.3. Summary of key innovations6.4. TensorFlow Hub and hands-on6.5. Practical applicationsSummary
Chapter 7. Semi-Supervised GAN
7.1. Introducing the Semi-Supervised GAN7.1.1. What is a Semi-Supervised GAN?7.1.2. Architecture7.1.3. Training process7.1.4. Training objective7.2. Tutorial: Implementing a Semi-Supervised GAN7.2.1. Architecture diagram7.2.2. Implementation7.2.3. Setup7.2.4. The dataset7.2.5. The Generator7.2.6. The Discriminator7.2.7. Building the model7.2.8. Training7.3. Comparison to a fully supervised classifier7.4. ConclusionSummary
Chapter 8. Conditional GAN
8.1. Motivation8.2. What is Conditional GAN?8.2.1. CGAN Generator8.2.2. CGAN Discriminator8.2.3. Summary table8.2.4. Architecture diagram8.3. Tutorial: Implementing a Conditional GAN8.3.1. Implementation8.3.2. Setup8.3.3. CGAN Generator8.3.4. CGAN Discriminator8.3.5. Building the model8.3.6. Training8.3.7. Outputting sample images8.3.8. Training the model8.3.9. Inspecting the output: Targeted data generation8.4. ConclusionSummary
Chapter 9. CycleGAN
9.1. Image-to-image translation9.2. Cycle-consistency loss: There and back aGAN9.3. Adversarial loss9.4. Identity loss9.5. Architecture9.5.1. CycleGAN architecture: building the network9.5.2. Generator architecture9.5.3. Discriminator architecture9.6. Object-oriented design of GANs9.7. Tutorial: CycleGAN9.7.1. Building the network9.7.2. Building the Generator9.7.3. Building the Discriminator9.7.4. Training the CycleGAN9.7.5. Running CycleGAN9.8. Expansions, augmentations, and applications9.8.1. Augmented CycleGAN9.8.2. ApplicationsSummary
Part 3. Where to go from here
Chapter 10. Adversarial examples
10.1. Context of adversarial examples10.2. Lies, damned lies, and distributions10.3. Use and abuse of training10.4. Signal and the noise10.5. Not all hope is lost10.6. Adversaries to GANs10.7. ConclusionSummary
Chapter 11. Practical applications of GANs
11.1. GANs in medicine11.1.1. Using GANs to improve diagnostic accuracy11.1.2. Methodology11.1.3. Results11.2. GANs in fashion11.2.1. Using GANs to design fashion11.2.2. Methodology11.2.3. Creating new items matching individual preferences11.2.4. Adjusting existing items to better match individual preferences11.3. ConclusionSummary
Chapter 12. Looking ahead
12.1. Ethics12.2. GAN innovations12.2.1. Relativistic GAN12.2.2. Self-Attention GAN12.2.3. BigGAN12.3. Further reading12.4. Looking back and closing thoughtsSummary
Training Generative Adversarial Networks (GANs)
Index
ABCDEFGHIJKLMNOPRSTUVWZ
List of Figures
List of Tables
List of Listings

Content preview from GANs in Action

Preface

Jakub Langr

When I first discovered GANs in 2015, I instantly fell in love with the idea. It was the kind of self-criticizing machine learning (ML) system that I always missed in other parts of ML. Even as humans, we constantly generate possible plans and then discriminate that just naively running into a door is not the best idea. GANs really made sense to me—to get to the next level of AI, we should take advantage of automatically learned representations and a machine learning feedback loop. After all, data was expensive, and compute was getting cheap.

The other thing I loved about GANs—though this realization came later—was its growth curve. No other part of ML is so “new.” Most of computer vision was invented before 1998, whereas ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781617295560Publisher Support Publisher Website Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

GANs in Action

by Jakub Langr, Vladimir Bok