Introduction
Backpropagation allows hidden layers to learn internal representations useful for a task. It is not itself a technique for visual generation, but becomes one of the mechanisms underlying later deep-learning vision systems. Features need no longer be entirely designed by hand; they can form through training.
View full image ↗Original network diagram from Rumelhart, Hinton and Williams’s 1986 backpropagation paper, PDF page 2. The authors’ connections, units and labels are preserved.
David E. Rumelhart; Geoffrey E. Hinton; Ronald J. Williams · Rights in the original image and document remain with their respective holders
Source · Learning representations by back-propagating errors ↗RESEARCH ACCOUNT
Back-propagating error makes internal representations trainable.
ANTI-AI ARCHIVE · Revised 2026-10-02
Method and contribution
Rumelhart, Hinton and Williams describe repeatedly adjusting connection weights to reduce the discrepancy between actual and desired outputs. Hidden units acquire features useful to the task rather than receiving a complete inventory of features from the designer. The 1986 paper demonstrates and explains this procedure; its publication should not be treated as the sole origin of every gradient-based learning method.
Section sources: Learning representations by back-propagating errors
Art-historical interpretation
This is a history of training mechanisms, not an artwork. Compared with explicitly programming drawing rules, learning intermediate features shifts attention towards datasets, objectives and the organisation of training. Later image models’ visual regularities can be investigated through these mechanisms without equating statistical learning with an artist’s lived experience.
Section sources: Learning representations by back-propagating errors
Sources for this account
Author or conference abstract and version record read; full proofs and experiments were not independently reproduced.
This account expands the archive within the cited scope. Historical interpretation is not presented as the original author’s statement or as a proven causal influence. The account and translation are AI-assisted, pending independent human review. Section references identify evidence without claiming independent verification of every historical statement.
Continue with a comparative question
Compare the methods, forms of control or presentation contexts of this entry and The Perceptron.
Compare the methods, forms of control or presentation contexts of this entry and AlexNet.
Dates & version record
Date displayed for this node: 1986 · Historical date recorded for the source: 1986
These dates refer to the historical event or recorded version, not this page’s publication date. The original date precision and unresolved questions are retained.
Original sources & further reading
These links lead to the cited paper, article, institution or conference page. External texts retain their source languages.
Learning representations by back-propagating errors
https://doi.org/10.1038/323533a0
Current cited URL
Provenance, translation & verification
Timeline entry #049 · Historical node
Research materials & supplement references · 1
AI Art Genealogy Research Master · p. 11 · 030
Date as recorded: 1986
Learning representations by back-propagating errors ↗
Learning representations by back-propagating errors
Archive account based on the listed research materials, not a full translation of the linked work. AI-assisted translation; independent human review pending.
The master’s research claims and dates await independent verification.
Cite this node
ANTI-AI ARCHIVE. Learning representations by back-propagating errors. Art-history node #049. https://salondesrefuses.cn/en/art-history/030
For specific historical claims, also cite the original sources above and include your access date. This account is not a full translation of the linked work.
Adjacent nodes follow chronological order; adjacency does not establish direct influence or causation.