// ✦ Diffusion, Flow Matching & Generative Models
Variational Autoencoder (VAE)
Encodes data as a distribution in latent space and decodes samples into new data.
变分自编码器 VAE 2013 Intermediate core
IN PLAIN WORDS
The encoder outputs mean and variance, samples via the reparameterization trick and decodes; loss = reconstruction + KL to a standard normal. Stable Diffusion's latent space comes from a VAE.
Used in: Latent encoders in Stable Diffusion
TIMELINE · prerequisites → this → what builds on it
01Original paper
Auto-Encoding Variational Bayes
Diederik P Kingma, Max Welling · 2013
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Learn first
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
04Where the course teaches it
On learning paths:Generative Models & Frontier Research
For AI / data
- Verification: verified, checked 2026-10-21.
- JSON: /api/learn/algorithms.json (id:
vae)