Research breakthrough · curated entry
The variational autoencoder gives deep generative models a probability frame
Auto-Encoding Variational Bayes
- Date
- December 20, 2013
- Category
- Research breakthrough
- Organization / Model
- University of Amsterdam
- Evidence
- 1 source · 1 evidence family
Kingma and Welling combined variational inference with the reparameterisation trick so that deep generative models with latent variables could be trained end to end by backpropagation. Together with the GAN of the same period it formed one of two main generative lines, and its latent-space idea spread through image, speech and representation learning.
Curator's note: Cross-checked against the arXiv preprint; the paper was published at ICLR in 2014.
Curated sources · 1
Cite this page
This page is a curated static snapshot and can be cited directly. Preferred format GB/T 7714—2025 (electronic resource [EB/OL]); the other 5 styles open below, together with the machine-readable files.
GB/T 7714—2025 (electronic resource [EB/OL])
AI 纪年. 变分自编码器 VAE 提出，深度生成模型有了概率框架[EB/OL]. (2013-12-20)[2026-11-03]. https://aihist.net/e/vae-2013.html.
APA 7 · American Psychological Association
AI 纪年. (2013, December 20). 变分自编码器 VAE 提出，深度生成模型有了概率框架 [Web page]. https://aihist.net/e/vae-2013.html
MLA 9 · Modern Language Association
"变分自编码器 VAE 提出，深度生成模型有了概率框架." AI 纪年, 20 Dec. 2013, aihist.net/e/vae-2013.html. Accessed 3 Nov. 2026.
Chicago 17 · notes-bibliography
AI 纪年. "变分自编码器 VAE 提出，深度生成模型有了概率框架." Published Dec. 20, 2013. Accessed Nov. 3, 2026. https://aihist.net/e/vae-2013.html.
IEEE · citation-reference list
"变分自编码器 VAE 提出，深度生成模型有了概率框架," AI 纪年, 20 Dec. 2013. [Online]. Available: https://aihist.net/e/vae-2013.html. [Accessed: 3 Nov. 2026].
ACM · author-date variant
AI 纪年. December 20, 2013. 变分自编码器 VAE 提出，深度生成模型有了概率框架. Retrieved November 3, 2026 from https://aihist.net/e/vae-2013.html.
All six styles come from one template and can be compared word for word. [引用日期], Accessed, urldate and Y2 all carry this page's build date, 2026-11-03 — a static page is an immutable snapshot, so substitute the date you actually consulted it. The address in every style is the archive's canonical page: this English page translates the same snapshot, and the work keeps its name, so one record cites both. For one-click copying use the same entry in the timeline, which takes the day you opened it. Expand the other styles before printing.