Creating the first convolution layer

In this recipe, let's create the first convolution layer.

Getting ready

The following are the inputs to the function create_conv_layer defined in the recipe Using functions to create a new convolution layer.

• Input: This is a four-dimensional reshaped input placeholder variable: x_image
• Num_input_channels: This is the number of color channels, namely num_channels
• Filter_size: This is the height and width of the filter layer filter_size1
• Num_filters: This is the depth of the filter layer, namely num_filters1
• Use_pooling: This is the binary flag set to TRUE

How to do it...

1. Run the create_conv_layer function with the preceding input parameters:

# Convolutional Layer 1
conv1 <- create_conv_layer(input=x_image,
num_input_channels=num_channels,
filter_size=filter_size1,
num_filters=num_filters1,
use_pooling=TRUE)

2. Extract the layers of the first convolution layer:

layer_conv1 <- conv1$layer
conv1_images <- conv1$layer$eval(feed_dict = dict(x = train_data$images...