Variational Autoencoder - Model, ELBO, loss function and maths explained easily!

Umar Jamil · Beginner ·🧬 Deep Learning ·3y ago

About this lesson

A complete explanation of the Variational Autoencoder, a key component in Stable Diffusion models. I will show why we need it, the idea behind the ELBO, the problems in maximizing the ELBO, the loss function and explain the math derivations step by step. Link to the slides: https://github.com/hkproj/vae-from-scratch-notes Chapters 00:00 - Introduction 00:41 - Autoencoder 02:35 - Variational Autoencoder 04:20 - Latent Space 06:06 - Math introduction 08:45 - Model definition 12:00 - ELBO 16:05 - Maximizing the ELBO 19:49 - Reparameterization Trick 22:41 - Example network 23:55 - Loss function

Original Description

A complete explanation of the Variational Autoencoder, a key component in Stable Diffusion models. I will show why we need it, the idea behind the ELBO, the problems in maximizing the ELBO, the loss function and explain the math derivations step by step. Link to the slides: https://github.com/hkproj/vae-from-scratch-notes Chapters 00:00 - Introduction 00:41 - Autoencoder 02:35 - Variational Autoencoder 04:20 - Latent Space 06:06 - Math introduction 08:45 - Model definition 12:00 - ELBO 16:05 - Maximizing the ELBO 19:49 - Reparameterization Trick 22:41 - Example network 23:55 - Loss function
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Related Reads

📰
Understanding Deep Learning Through Four Interactive Experiments
Explore deep learning concepts through interactive experiments to gain hands-on understanding
Medium · Data Science
 📰
Understanding Deep Learning Through Four Interactive Experiments
Explore deep learning through interactive experiments to gain hands-on understanding
Medium · Deep Learning
 📰
Optimizers in Deep Learning: From Gradient Descent to Adam
Learn how optimizers in deep learning work, from basic Gradient Descent to advanced Adam optimizer, to improve model training
Medium · Deep Learning
 📰
The Meta-Architecture of Interface Fracture: High-Dimensional Logical Stress and Systemic Collapse…
Learn about the meta-architecture of interface fracture and its relation to high-dimensional logical stress and systemic collapse in deep learning systems
Medium · Deep Learning

Chapters (11)

Introduction
0:41 Autoencoder
2:35 Variational Autoencoder
4:20 Latent Space
6:06 Math introduction
8:45 Model definition
12:00 ELBO
16:05 Maximizing the ELBO
19:49 Reparameterization Trick
22:41 Example network
23:55 Loss function
Up next
Image Classification with ml5.js
The Coding Train
Watch →