DDPM stands for: Denoising Diffusion Probabilistic Model

DDPM stands for Denoising Diffusion Probabilistic Model

As of 2024, DDPMs are primarily used for computer vision tasks, including image generation and enhancement.

Example

A trained DDPM can generate new images that resemble a given dataset of photographs.

Remember this

Understanding DDPMs is crucial for advancing generative models in computer vision tasks.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

denoising score matching does: learns to denoise, which equals learning the score

Propensity score matching (PSM) reduces bias in treatment effect estimates

Langevin dynamics does: adds noise to gradient descent to sample from a distribution

Langevin dynamics uses stochastic differential equations

Stable Diffusion

Stable Diffusion generates images from text descriptions

DDIM does: deterministic sampling for faster generation with fewer steps

DDIM accelerates image generation by deterministically sampling intermediate steps

Diffusion model

q(x_t|x_{t-1}) adds Gaussian noise at each step

Markov chain Monte Carlo

MCMC samples from complex posterior distributions

Swipe through 100 ML concepts daily