M22.3 CONNECT THE MECHANISM
Learn a latent representation and a way back to data
Squeeze a doodle into two numbers and rebuild it. Then pick two random numbers and decode a doodle nobody drew. Learn the one-line trick and the one penalty that make the second part work.
LESSON OVERVIEW13 min lesson
Lesson overview
Squeeze a doodle into two numbers and rebuild it. Then pick two random numbers and decode a doodle nobody drew. Learn the one-line trick and the one penalty that make the second part work.
What you’ll explore
- Autoencoders reconstruct inputs through a latent representation; variational autoencoders learn a probabilistic encoder and decoder with a prior-matching term that supports principled latent sampling.
GO TO THE SOURCE
Original explanations, connected to the research.
Auto-Encoding Variational Bayes (Kingma & Welling, 2013)Stochastic Backpropagation and Approximate Inference in Deep Generative Models (Rezende, Mohamed & Wierstra, 2014)An Introduction to Variational Autoencoders (Kingma & Welling, 2019)Deep Learning, chapter 14: Autoencoders (Goodfellow, Bengio & Courville, 2016)Suggest a correction
A precise note can make an explanation better.
Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.