Convolutional Neural Networks

Convolutional Neural Network (CNN) is the name given to a neural network that has convolutional layers. These convolutional layers handle the high dimensionality of raw images efficiently with the help of convolutional filters. CNNs allow us to recognize highly complex patterns in images, which would be impossible with a simple neural network. CNNs can also be used for natural language processing.

The first few layers of a CNN are convolutional, where the network applies different filters to the image to find useful patterns in the image; then there's the pooling layers, which help down-sample the output of the convolutional layers. The activation layer controls which signal flows from one layer to the next, emulating the neurons in our brain. The last few layers in the network are dense layers; these are the same layers we used for the previous exercise.

Convolutional Layer

The convolutional layer consists of multiple filters that learn to activate when they...