Deep Learning
A free video path covering GAN fundamentals, implementation in Keras and TensorFlow, and advanced variants like DCGAN, StyleGAN, StackGAN, CycleGAN, and SRGAN.
Learn from: the original creator. Original content published on YouTube. DigitalSkillX organizes these public resources into a structured learning path. DigitalSkillX does not claim ownership or partnership.
Now playing: Generative Adversarial Networks: A Beginner's Guide to GANs
This learning path follows a YouTube playlist by Code With Aarohi. It starts with the basic idea behind GANs, including the generator and discriminator and how they are trained. It then moves into hands-on coding with Keras and TensorFlow, building models for face generation and MNIST digits. Later videos cover specific GAN architectures: DCGAN, StyleGAN, StyleGAN2 ADA, StackGAN, CycleGAN, and SRGAN. The path is best for learners who already know some Python and basic deep learning, since the implementation videos assume that background. Each lesson is a video, and the order follows the original playlist.
Self-paced learners who want a clear free path with creator-credited YouTube lessons.
Lesson 1: Generative Adversarial Networks: A Beginner's Guide to GANs
This beginner's guide introduces the architecture of GANs, explaining how generators and discriminators work together. It covers how GANs create data that mimics real examples and discusses key loss functions such as binary cross-entropy. The video is a good starting point before any coding.
Open on YouTubeLesson 2: What Are GANs? | Generative Adversarial Networks Explained
This video answers the question 'What are GANs?' by covering generative models, the basic GAN concept, how GANs work, and the training phases. It provides a conceptual overview that complements the first lesson.
Open on YouTubeLesson 3: GANs Implementation: Creating Faces that Don't Exist
This implementation video uses the CelebA dataset to generate realistic human faces that do not exist. It walks through coding the generator and discriminator from scratch, the step-by-step training process, and how to calculate losses. It is a practical follow-up to the fundamentals.
Open on YouTubeLesson 4: Generative Adversarial Networks (GAN) - implementation in Keras
This lesson implements a GAN in Keras for generating MNIST handwritten digits. It covers the generator, discriminator, and loss function, with a GitHub link for the full code. It is a shorter, focused coding exercise.
Open on YouTubeLesson 5: StyleGAN Explained
This video explains StyleGAN and how it differs from a standard GAN. It discusses why StyleGAN produces high-quality images and what improvements it brings. No coding is involved; it is a conceptual lesson.
Open on YouTubeLesson 6: StyleGAN Implementation
This lesson shows how to implement StyleGAN using a pretrained model. It includes a GitHub link and refers back to the StyleGAN explanation video. It is a practical, shorter session focused on using an existing model.
Open on YouTubeLesson 7: StyleGAN2 ADA on Custom Dataset |StyleGAN2 with adaptive discriminator augmentation (ADA)
This video covers StyleGAN2 with adaptive discriminator augmentation (ADA), released in 2020. It shows how to prepare a custom dataset, train StyleGAN2 ADA, and generate images from the trained model. A GitHub link is included.
Open on YouTubeLesson 8: StackGAN | Text to Image Generation with Stacked Generative Adversarial Networks
This video explains StackGAN, a model for text-to-image generation using stacked generative adversarial networks. It is a conceptual overview that pairs with the following implementation video.
Open on YouTubeLesson 9: StackGAN Implementation| Text to Image Generation with Stacked Generative Adversarial Networks
This is a TensorFlow implementation of StackGAN for text-to-photo-realistic image synthesis. It follows the StackGAN explanation and includes a GitHub link. The video is longer and goes through the code in detail.
Open on YouTubeLesson 10: CycleGAN | Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
This short video introduces CycleGAN for unpaired image-to-image translation. It explains the idea of translating an image from one domain to another, such as night to day or sunny to rainy. It is a conceptual overview.
Open on YouTubeLesson 11: SRGAN Explained | Super-Resolution Generative Adversarial Network
This video explains SRGAN, a Super-Resolution Generative Adversarial Network. It covers how SRGAN upscales images by a factor of 4, for example from 172x208 pixels to 688x832 pixels. It is a conceptual lesson on super-resolution.
Open on YouTubeLesson 12: DCGAN | Deep Convolutional Generative Adversarial Network
This video implements a Deep Convolutional GAN (DCGAN) on the CelebA dataset using TensorFlow 2.3 and Keras. It explains what DCGAN is and shows how to generate fake faces. A GitHub link is provided for the code.
Open on YouTube0 of 12 lessons completed (0%).
Complete this learning path to become eligible for your DigitalSkillX certificate.
Deep Learning
PyTorch for Deep Learning: A Complete Beginner Path
A 22-part video course by Patrick Loeber that takes you from installing PyTorch to building, training, and deploying neural networks.
Generative AI
Generative AI and LLMs: From Basics to AI Agents
A 31-part beginner-to-advanced course that explains how large language models work and builds up to RAG, AI agents, and production AI systems, without heavy math.
Generative AI
Generative AI Course: From Basics to Building with LLMs
A free video learning path covering generative AI fundamentals, major AI chatbots, local LLM tools, and frameworks for building AI applications. Taught by Amit Diwan.
Generative AI
Generative AI Course: Beginner to Advanced
A 10-part video series that takes you from AI basics through Python, machine learning, transformers, image generation, LLM apps, RAG, and agentic AI.