Build GANs and Diffusion Models with TensorFlow and PyTorch
2h 22mAdvanced2022-09-15
Authors
Janani Ravi
Certified Google Cloud Architect and Data Engineer
Course details
If you’re looking for a crash course in generative modeling, this course was made for you. Generative adversarial networks (GANs) and diffusion models are some of the most important components of machine learning infrastructure. Join instructor Janani Ravi to find out more about how to get started building GANs with both dense neural as well as deep convolutional networks. Javani shows you the basics of how to train a deep convolutional GAN on multichannel images. Along the way, she gives you tips on how to get up and running with GANs using TensorFlow and diffusion models using PyTorch.
Skills covered
PulumiKubernetesMachine LearningGenerative AIArtificial Intelligence FoundationsAdvancedAmazon Web Services (AWS)AmazonArtificial Intelligence (AI)Open Source
Concepts
Introduction
- Overview of generative models
- Applications of generative models
Getting Started with Generative Adversarial Networks
- Introducing GANs and diffusion models
- Generator and discriminator
- Architectural overview of a GAN
- Training the generator and discriminator
- Common problems with GANs
Building a GAN Using a Dense Neural Network
- Getting set up with Google Colab
- Loading the fashion MNIST data set
- The generator network
- The discriminator network
- Adversary loss functions
- Training the generative adversarial network
- Generating images using the GAN
Building a GAN Using a Deep Convolutional Network
- Overview of CNNs
- Transposed convolutional layer
- Deep Convolutional GANs
- Greyscale images - Generator and discriminator in a Deep Convolutional GAN
- Greyscale images - Training a Deep Convolutional GAN
Training a Deep Convolutional GAN on Multichannel Images
- Color images - Loading multichannel image data
- Color images - Generator and discriminator in a Deep Convolutional GAN
- Color images - Training a Deep Convolutional GAN
Getting Started with Diffusion Models
- Generative learning trilemma
- Introducing denoising diffusion probabilistic models
- How do denoising diffusion probabilistic models work
- Forward diffusion process
- Reverse diffusion process
- Training a diffusion model - Intuition
Running a Diffusion Model
- Denoising diffusion probabilistic models - Exploring implementation on GitHub
- Denoising diffusion probabilistic models - Code overview
- Denoising diffusion probabilistic models - Code tweaks
- Denoising diffusion probabilistic models - Generating images
Conclusion
- Summary and next steps