Generative Models

Mix a clean signal with noise

A journey through diffusion models with Botie.

Generative Models · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

Could Learning to Remove Noise Teach a Model to Generate?

Part 1 of 3

Creating a detailed image in one operation is a difficult task. Adding a little noise to an existing image is easy, and learning to account for a small corruption gives us a more manageable question. What if we could organise generation as many such steps?

Sources for this history

Sohl-Dickstein and colleagues (2015), Deep Unsupervised Learning Using Nonequilibrium Thermodynamics
Ho, Jain and Abbeel (2020), Denoising Diffusion Probabilistic Models

← Previous lesson: Generative Adversarial Networks Read the full lesson and derivations