liveBook · Manning

Diffusion model

As explained in Machine Learning Algorithms in Depth

A diffusion model is a generative model that learns to reverse a gradual noise-adding process to generate samples from a data distribution.

A diffusion model is a generative model that learns to reverse a gradual noise-adding process, producing samples that match a target data distribution. section 11.5

The model uses a parameterized Markov chain with two passes: a forward pass progressively adds Gaussian noise to an image, while a reverse pass transforms noise back into a target-distribution sample. section 11.5

Core structure:

  • Train the Markov chain using variational inference. section 11.5
  • Use additive Gaussian noise to make images progressively noisier. section 11.5
  • Reverse the noising process to generate samples, including high-resolution images. section 11.5

How Machine Learning Algorithms in Depth covers it

How deep it goes
Defines Markov-chain structure and forward and reverse passes, but does not derive or implement training.
Examples
Uses progressively noised images and denoising diffusion probabilistic models as concrete reference points.
What you can do after
Evaluate/understand how noisy images become samples matching a target distribution.
How it connects
Connects diffusion models to variational inference, Markov chains, additive Gaussian noise, and likelihood-based generative models.
Cover of Machine Learning Algorithms in Depth
liveBookmanning.com
Read Machine Learning Algorithms in Depth on liveBook Get the book at manning.com

Also explained in

Compare how 10 books explain diffusion model →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machine Learning Algorithms in Depth. The section links point to the parts of the book it draws on.