Wikinventia — Atlas of discoveries and inventions · Global Age

Backpropagation — efficient training of multilayer neural networks — Rumelhart, Hinton, and Williams

1986 AD · Transmission: Global
AIMethodBritish

In 1986 Geoffrey Hinton, David Rumelhart, and Ronald Williams published in Nature the practical demonstration that the backpropagation algorithm could efficiently train multilayer neural networks. Backpropagation calculates how to adjust a network's weights by propagating the error from the output back to earlier layers via partial derivatives. The article pulled neural networks out of the second AI winter and laid the technical foundation for deep learning. Hinton continued for decades as a marginal figure in a community dominated by other approaches, until AlexNet (2012) proved him right. In 2024 he received the Nobel Prize in Physics together with John Hopfield for their fundamental contributions to machine learning.

InstitutionUniversity of California San Diego / Carnegie Mellon / University of Toronto
Historical regionUnited States / Canada
Primary sourceRumelhart, D.E., Hinton, G.E. & Williams, R.J. — "Learning representations by back-propagating errors" (Nature, 323, 533–536, 1986)
Secondary sourceHinton, G. — Nobel Prize in Physics 2024 (with John Hopfield) "for foundational discoveries and inventions that enable machine learning with artificial neural networks"
Original languageEnglish
View this entry in the interactive atlas → View in graph →