Skip to main content
. 2020 Jun 10;120(16):8066–8129. doi: 10.1021/acs.chemrev.0c00004

Figure 21.

Figure 21

Example of the use of a CNN. One slides convolution layers (red) over an image, which for example can be a two-dimensional diffraction pattern.131 Usually, one then uses a pooling layer to compress the matrices after convolution. After flattening, the output can be used for conventional hidden NN layers.