Hands-On Automated Machine Learning

The fully connected layer

There can be multiple convolution, ReLU, and pooling operations in a CNN. However, there is always a last single stage which is a fully connected layer. The fully connected layer is the feed-forward neural network that we discussed earlier. The purpose of this step is to make different predictions on the image dataset, such as classifying images.

The following diagram illustrates the final picture of a CNN's basic architecture:

Now that we understand some basics about CNNs, let's create a CNN using Keras. We will use the MNIST dataset that is present in Keras itself. The MNIST dataset is a well-known dataset of written ...