Generating a New Reality : From Autoencoders and Adversarial Networks to Deepfakes

The emergence of artificial intelligence (AI) has brought us to the precipice of a new age where we struggle to understand what is real, from advanced CGI in movies to even faking the news. AI that was developed to understand our reality is now being used to create its own reality. In this book we look at the many AI techniques capable of generating new realities. We start with the basics of deep learning. Then we move on to autoencoders and generative adversarial networks (GANs). We explore variations of GAN to generate content. The book ends with an in-depth look at the most popular generator projects. By the end of this book you will understand the AI techniques used to generate different forms of content. You will be able to use these techniques for your own amusement or professional career to both impress and educate others around you and give you the ability to transform your own reality into something new. What You Will Learn Know the fundamentals of content generation from autoencoders to generative adversarial networks (GANs) Explore variations of GAN Understand the basics of other forms of content generation Use advanced projects such as Faceswap, deepfakes, DeOldify, and StyleGAN2 Who This Book Is For Machine learning developers and AI enthusiasts who want to understand AI content generation techniques Table of Contents About the Author About the Technical Reviewer Acknowledgments Introduction Chapter 1: The Basics of Deep Learning Prerequisites The Perceptron The Multilayer Perceptron Backpropagation Stochastic Gradient Descent PyTorch and Deep Learning Understanding Regression Over- and Underfitting Classifying Classes One-Hot Encoding Classifying MNIST Digits Conclusion Chapter 2: Unleashing Generative Modeling Unsupervised Learning with Autoencoders Extracting Features with Convolution The Convolutional Autoencoder Generative Adversarial Networks Deep Convolutional GAN Conclusion Chapter 3: Exploring the Latent Space Understanding What Deep Learning Learns Deep Learning Function Approximation The Limitations of Calculus Deep Learning Hill Climbing Over- and Underfitting Building a Variational Autoencoder Learning Distributions with the VAE Variability and Exploring the Latent Space Conclusion Chapter 4: GANs, GANs, and More GANs Feature Understanding and the DCGAN Unrolling the Math of GANs Resolving Distance with WGAN Discretizing Boundary-Seeking GANs Relativity and the Relativistic GAN Conditioning with CGAN Conclusion Chapter 5: Image to Image Content Generation Segmenting Images with a UNet Uncovering the Details of a UNet Translating Images with Pix2Pix Seeing Double with the DualGAN Riding the Latent Space on the BicycleGAN Discovering Domains with the DiscoGAN Conclusion Chapter 6: Residual Network GANs Understanding Residual Networks Cycling Again with CycleGAN Creating Faces with StarGAN Using the Best with Transfer Learning Increasing Resolution with SRGAN Conclusion Chapter 7: Attention Is All We Need! What Is Attention? Understanding the Types of Attention Applying Attention Augmenting Convolution with Attention Lipschitz Continuity in GANs What Is Lipschitz Continuity? Building the Self-Attention GAN Improving on the SAGAN Conclusion Chapter 8: Advanced Generators Progressively Growing GANs Styling with StyleGAN Version 2 Mapping Networks Style Modules Removing Stochastic/Traditional Input Stochastic Variation (Noisy Inputs) Mixing Styles Truncation of W Hyperparameter Tuning Frechet Inception Distance StyleGAN2 Weight Demodulation Path Length Regularization Lazy Regularization No Growing Large Networks DeOldify and the New NoGAN Colorizing and Enhancing Video Being Artistic with ArtLine Conclusion Chapter 9: Deepfakes and Face Swapping Introducing the Tools for Face Swapping Gathering the Swapping Data Downloading YouTube Videos for Deepfakes Understanding the Deepfakes Workflow Extracting Faces Sorting and Trimming Faces Realigning the Alignments File Training a Face Swapping Model Creating a Deepfake Video Encoding the Video Conclusion Chapter 10: Cracking Deepfakes Understanding Face Manipulation Methods Techniques for Cracking Fakes Handcrafted Features Learning-Based Features Artifacts Identifying Fakes in Deepfakes Conclusion Appendix A: Running Google Colab Locally Appendix B: Opening a Notebook Appendix C: Connecting Google Drive and Saving Index "The emergence of artificial intelligence (AI) has brought us to the precipice of a new age where we struggle to understand what is real, from advanced CGI in movies to even faking the news. AI that was developed to understand our reality is now being used to create its own reality. In this book we look at the many AI techniques capable of generating new realities. We start with the basics of deep learning. Then we move on to autoencoders and generative adversarial networks (GANs). We explore variations of GAN to generate content. The book ends with an in-depth look at the most popular generator projects. By the end of this book you will understand the AI techniques used to generate different forms of content. You will be able to use these techniques for your own amusement or professional career to both impress and educate others around you and give you the ability to transform your own reality into something new. You will: Know the fundamentals of content generation from autoencoders to generative adversarial networks (GANs); Explore variations of GAN; Understand the basics of other forms of content generation; Use advanced projects such as Faceswap, deepfakes, DeOldify, and StyleGAN2."-- Provided by publisher