Generative Adversarial Networks (GAN): The Complete Guide

Name: Generative Adversarial Networks (GAN): The Complete Guide
Rating: 3.6 (42 reviews)

Generative Adversarial Networks in Python

Created byHoang Quy La

Last updated 3/2022

English

English [Auto],

What you'll learn

Learn the basic principles of generative models
Build a GAN (Generative Adversarial Network) in Tensorflow
Tensorflow
DCGAN
WGAN

Course content

4 sections • 20 lectures • 3h 47m total length

Course Structure1:01
Outline the course structure and costs, introduce yarn and code, and map progress through GAN topics with planned updates and future content.
How to make the course out of this course1:52
Watch all video content, follow step-by-step explanations, and engage in the Q&A to maximize learning and ask questions throughout the GAN course.
Introduction to GAN20:54
Master the basics of generative adversarial networks, where the generator and discriminator compete to distinguish real from fake images, and tackle the core training challenges.
GAN Project: Importing libraries and data11:15
Begin a GAN project by importing the required libraries and loading the dataset, applying transforms and normalization, and configuring image size and learning parameters.
GAN Project: Generator Construction17:08
In this lecture, we construct the generator for the GAN project, configuring nz, ngf, and nc, building a sequential convolutional generator and implementing its forward pass.
GAN Project: Discriminator Construction13:31
Iterate on discriminator construction for GANs, copy and adapt code, compare activation choices such as leaky relu, and implement global average pooling to streamline the network.
GAN Project: Defining optimizer and loss4:59
Define the loss criterion and set up the optimizer for the GAN project, wiring the generator and discriminator, preparing noise inputs, and starting the training run.
GAN Project: Fully Connected Network and results17:56
Train a fully connected network as a generative adversarial network, with a generator producing samples and a discriminator learning to distinguish real and fake images.

Introduction to DCGAN and project implementation Part 112:34
Explore the dcgan architecture, including generator and discriminator with convolutional layers and activation functions, and implement a TensorFlow-based project with data preparation, training, and evaluation.
Introduction to DCGAN and project implementation Part 29:13
Continue building the DCGAN generator by feeding noise, applying batch normalization, and using transposed convolutions, then set up the discriminator in the next video.
Introduction to DCGAN and project implementation Part 39:57
Learn to configure a dcgan with discriminator and generator, trainable parameters, binary cross-entropy loss, and epoch-based optimization to produce image samples.
Introduction to DCGAN and project implementation Part 418:43
Train a dcgan by generating images from noise and iteratively updating the generator and discriminator (critic) across epochs, monitoring g_loss and d_loss.
Introduction to DCGAN and project implementation Final Part19:05
This part introduces dcgan project implementation, coding and training, showing generator and discriminator, loss and accuracy plots, and generating images while exploring upgrades like a larger model and data augmentation.

Introduction to WGAN5:58
Learn how wgan extends generative adversarial networks to stabilize training with a loss that correlates with the quality of generated images and a critic, reducing sensitivity to architecture and hyperparameters.
WGAN Implementation Part 114:51
In this WGAN implementation part 1, learners implement the generator using a sequential, fully connected network with dense layers, batch normalization, and activation, preparing the model for training with TensorFlow.
WGAN Implementation Part 216:49
Continue building the wgan with a generator, a critic, and a combi network, training in alternating steps with clip value and scaling data to -1 to 1.
WGAN Implementation Part 37:47
Implement a wgan with a generator and a critic, using simple fully connected networks, activation functions, and a clipping step for the critic during training.
WGAN Implementation Part 49:11
Continue the WGAN implementation by generating images from random noise and plotting the results, configuring image dimensions and the generator’s output, and setting up main function for the next video.
WGAN Implementation final Part13:41
Implement the final part of the WGAN by building the generator and critic, configuring the optimizer, and training across 10,000 to 40,000 steps.

Requirements

Calculus
Probability
Object-oriented programming
Python coding: if/else, loops, lists, dicts, sets
Basic deep learning

Description

GANs have been one of the most interesting developments in deep learning and machine learning recently.

Yann LeCun, a deep learning pioneer, has said that the most important development in recent years has been adversarial training, referring to GANs.

GAN stands for generative adversarial network, where 2 neural networks compete with each other.

What is unsupervised learning?

Unsupervised learning means we’re not trying to map input data to targets, we’re just trying to learn the structure of that input data.

This course is a comprehensive guide to Generative Adversarial Networks (GANs). The theories are explained in-depth and in a friendly manner. After each theoretical lesson, we will dive together into a hands-on session, where we will be learning how to code different types of GANs in PyTorch and Tensorflow, which is a very advanced and powerful deep learning framework!

In this first course, You will learn

GAN
DCGAN
WGAN

"If you can't implement it, you don't understand it"

Or as the great physicist Richard Feynman said: "What I cannot create, I do not understand".
My courses are the ONLY courses where you will learn how to implement machine learning algorithms from scratch
Other courses will teach you how to plug in your data into a library, but do you really need help with 3 lines of code?
After doing the same thing with 10 datasets, you realize you didn't learn 10 things. You learned 1 thing, and just repeated the same 3 lines of code 10 times...

Who this course is for:

Anyone who wants to improve their deep learning knowledge

Generative Adversarial Networks (GAN): The Complete Guide

What you'll learn

Explore related topics

Course content

Introduction8 lectures • 1hr 29min

Introduction to DCGAN5 lectures • 1hr 10min

WGAN6 lectures • 1hr 8min

Thank you1 lecture • 1min

Requirements

Description

Who this course is for: