The input dataset is defined and loaded. The create_conv_layer function presented in the recipe takes the following five input parameters and needs to be defined while setting-up a convolution layer:
- Input: This is a four-dimensional tensor (or a list) that comprises a number of (input) images, the height of each image (here 32L), the width of each image (here 32L), and the number of channels of each image (here 3L : red, blue, and green).
- Num_input_channels: This is defined as the number of color channels in the case of the first convolution layer or the number of filter channels in the case of subsequent convolution layers.
- Filter_size: This is defined as the width and height of each filter in the convolution layer. Here, the filter is assumed to be a square.
- Num_filters: This is defined as the number of filters in a given convolution layer.
- Use_pooling: This is a binary variable that is used perform 2 x 2 max pooling.