Keras pooling layers

Keras implements a pooling operation as a layer that can be added to CNNs between other layers. In this exercise, you will construct a convolutional neural network similar to the one you have constructed before:

Convolution => Convolution => Flatten => Dense

However, you will also add a pooling layer. The architecture will add a single max-pooling layer between the convolutional layer and the dense layer with a pooling of 2x2:

Convolution => Max pooling => Convolution => Flatten => Dense

A Sequential model along with Dense, Conv2D, Flatten, and MaxPool2D objects are available in your workspace.

This exercise is part of the course

Image Modeling with Keras

View Course

Exercise instructions

Add an input convolutional layer (15 units, kernel size of 2, relu activation).
Add a maximum pooling operation (pooling over windows of size 2x2).
Add another convolution layer (5 units, kernel size of 2, relu activation).
Flatten the output of the second convolution and add a Dense layer for output (3 categories, softmax activation).

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Add a convolutional layer
____(____(15, kernel_size=2, activation='relu', 
                 input_shape=(img_rows, img_cols, 1)))

# Add a pooling operation
____

# Add another convolutional layer
____

# Flatten and feed to output layer
model.add(____)
model.add(____(3, activation='softmax'))
model.summary()

Edit and Run Code