From VAEs to Diffusion Models | Angus Turner

Machine Learning and Data Science.

\[\require{cancel}\] Introduction Recently I have been studying a class of generative models known as diffusion probabilistic models. These models were proposed by Sohl-Dickstein et al. in 2015 [1] , however they first caught my attention last year when Ho et al. released “Denoising Diffusion Probabilistic Models” [2] . Building on [1] , Ho et al. showed that a model trained with a stable variational objective could match or surpass GANs on image generation. Since the release of DDPM there has been a wave of renewed interest in diffusion models. New works have extended their success to the dom

saved by

• João Araújo

related reading

• Generative Modeling by Estimating Gradients of the Data Distribution | Yang Songyang-song.github.io
• What are Diffusion Models? | Lil'Loglilianweng.github.io
• The Principles of Diffusion Modelsarxiv.org
• The Principles of Diffusion Modelsarxiv.org
• Diffusion model - Wikipediaen.wikipedia.org
• Step-by-Step Diffusion: An Elementary Tutorialarxiv.org
• [2006.11239] Denoising Diffusion Probabilistic Modelsarxiv.org
• ⭐️ Diffusion Modelsandrewkchan.dev
• [2010.02502] Denoising Diffusion Implicit Modelsarxiv.org
• [2406.08929] Step-by-Step Diffusion: An Elementary Tutorialar5iv.labs.arxiv.org
• Tutorial - What is a variational autoencoder? – Jaan Lı 李jaan.io
• Diffusion models from scratchchenyang.co