Artificial intelligence · 1989

Convolutional neural network

LeCun trains a convolutional net by backprop to read handwritten ZIP codes.

Yann LeCun and colleagues at Bell Labs combined weight-shared convolution with backpropagation to recognise digits from US mail. The architecture made vision learnable end to end and became the workhorse of computer vision for 25 years.

See it in the living tree Wikipedia

Arcs above join it to what it stood on, arcs below to what it led to. The bar places this stretch on the tree's whole timeline.

What it built on 2

Backpropagation 1986
Neocognitron 1980

What it led to 4

LeNet-5 1998
Deep learning (AlexNet) 2012
AlphaGo 2016
WaveNet 2016

Related measurements

The dashed line marks when it first worked.

Training compute of notable AI models Source: Epoch AI, Notable AI Models (CC BY 4.0)

Computation per constant dollar, 1850-2006 Latest: 7.3B million units of computer power per 2006 $ (2003) Source: Nordhaus 2007, Two Centuries of Productivity Growth in Computing, J. Econ. Hist. 67(1), Table 6