Training neural networks with backpropagation
Used the chain rule to pass errors back from the output layer, layer by layer, to train multi-layer networks.
Why it matters
Solved the problem of how to train multi-layer networks. Every large model today is still trained this way.
Costs and controversy
Similar methods had been proposed before; this Nature paper is what made it widely known.
Built on
Led to
LeNetAlexNetTuring AwardNobel Prize
Source: doi.org ↗