AI ChronicleAI 纪年 Full timeline →

Research breakthrough · curated entry

Deep belief networks make deep layers trainable again

Reducing the Dimensionality of Data with Neural Networks

Date
July 28, 2006
Category
Research breakthrough
Organization / Model
University of Toronto
Evidence
1 source · 1 evidence family

Hinton and Salakhutdinov reported a layer-wise pretraining method for deep belief networks in Science, making deep architectures trainable once more. This is usually taken as the start of the deep-learning revival.

Curator's note: Cross-checked against the Science article page.

Curated sources · 1

  1. Original paper · Science2006-07-28

    Reducing the Dimensionality of Data with Neural Networks

    science.org

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. 深度信念网络让深层网络重新可训练，深度学习复兴[EB/OL]. (2006-07-28)[2026-11-03]. https://aihist.net/e/dbn-2006.html.

APA 7 · American Psychological Association

AI 纪年. (2006, July 28). 深度信念网络让深层网络重新可训练，深度学习复兴 [Web page]. https://aihist.net/e/dbn-2006.html

MLA 9 · Modern Language Association

"深度信念网络让深层网络重新可训练，深度学习复兴." AI 纪年, 28 July 2006, aihist.net/e/dbn-2006.html. Accessed 3 Nov. 2026.

Chicago 17 · notes-bibliography

AI 纪年. "深度信念网络让深层网络重新可训练，深度学习复兴." Published July 28, 2006. Accessed Nov. 3, 2026. https://aihist.net/e/dbn-2006.html.

IEEE · citation-reference list

"深度信念网络让深层网络重新可训练，深度学习复兴," AI 纪年, 28 July 2006. [Online]. Available: https://aihist.net/e/dbn-2006.html. [Accessed: 3 Nov. 2026].

ACM · author-date variant

AI 纪年. July 28, 2006. 深度信念网络让深层网络重新可训练，深度学习复兴. Retrieved November 3, 2026 from https://aihist.net/e/dbn-2006.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.