Artificial Intelligence · Generative Models
A topic within Generative Models, itself one of 10 topics in that field and part of Artificial Intelligence.
3 works
Sohl-Dickstein, Weiss, Maheswaranathan & Ganguli
Destroy structure in data with a slow noising process, then learn to reverse it: the result is a generative model with a tractable likelihood, and the basis of diffusion.
link checked 17 Sept 2026Ho, Jain & Abbeel
Reduces the diffusion objective to plain noise prediction, at which point sample quality catches up with adversarial models without the training instability.
link checked 17 Sept 2026Rombach et al.
Running the diffusion process in a compressed latent space rather than in pixels cuts the cost enough that high-resolution image models fit on ordinary hardware.
link checked 17 Sept 2026