A convolutional neural network perfoms the same transformation to every small patch duirng image processing or to consecutve tokens in a time series or sequential data. This may be acheived by having lots of different nodes for exach parch, but where the weights are 'clamped' to be the same, or by reusing a small network. In bith cases backpropogatio or otger learning rules have to be modified. The technique derives its name from mathematical {{convolutions]} used in linear time-series analysis.
Defined on page 277
Used on Chap. 8: page 172; Chap. 12: pages 276, 277, 286; Chap. 19: page 472
Convolutional neural network