Diffusion Models: Difference between revisions
No edit summary |
|||
| Line 1: | Line 1: | ||
==Background== | ==Background== | ||
By Sohl-Dickstein ''et al.''[https://arxiv.org/pdf/1503.03585.pdf]. | |||
The goal is to define a mapping between a complex distribution <math>q(\mathbf{x}^{(0)})</math> (e.g. set of realistic images) to a simple distribution <math>\pi(\mathbf{y})=p(\mathbf{x}^{(T)})</math>(e.g. multivariate normal).<br> | |||
This is done by defining a forward trajectory <math>q(\mathbf{x}^{(0...T)})</math> and optimizing a reverse trajectory <math>p(\mathbf{x}^{(0 ... T)})</math>.<br> | |||
The forward trajectory is repeatedly applying a Markov diffusion kernel (i.e. a function with a steady distribution <math>\pi(\mathbf{y})</math>), performing T steps of diffusion.<br> | |||
The reverse trajectory is again applying a diffusion kernel but with an estimated mean and variance.<br> | |||
==Image Generation== | ==Image Generation== | ||
Revision as of 16:00, 29 March 2022
Background
By Sohl-Dickstein et al.[1].
The goal is to define a mapping between a complex distribution \(\displaystyle q(\mathbf{x}^{(0)})\) (e.g. set of realistic images) to a simple distribution \(\displaystyle \pi(\mathbf{y})=p(\mathbf{x}^{(T)})\)(e.g. multivariate normal).
This is done by defining a forward trajectory \(\displaystyle q(\mathbf{x}^{(0...T)})\) and optimizing a reverse trajectory \(\displaystyle p(\mathbf{x}^{(0 ... T)})\).
The forward trajectory is repeatedly applying a Markov diffusion kernel (i.e. a function with a steady distribution \(\displaystyle \pi(\mathbf{y})\)), performing T steps of diffusion.
The reverse trajectory is again applying a diffusion kernel but with an estimated mean and variance.
Image Generation
Super-resolution
Resources
- Google AI Blog High Fidelity Image Generation Using Diffusion Models - discusses SR3 and CDM