Who co-authored the 1986 paper that helped popularize backpropagation for neural networks?

David Rumelhart, Geoffrey Hinton, and Ronald Williams co-authored the 1986 paper that helped popularize backpropagation for neural networks. Their article, “Learning representations by back-propagating errors,” described how multilayer networks could adjust internal weights using error information.

Backpropagation calculates how a change in each weight affects the final error, applying the chain rule of calculus through the network. Combined with gradient descent, this makes it practical to train networks with hidden layers rather than only single-layer models.

The 1986 paper did not invent every idea associated with backpropagation. Related mathematical methods had appeared earlier, including work by Paul Werbos and others. The importance of Rumelhart, Hinton, and Williams was their clear demonstration and influential presentation of the method for connectionist learning, helping establish multilayer neural networks as a major research direction.