Introduction

In the previous chapter, we learned about how a neural network can be trained to predict values and how a recurrent neural network (RNN), based on its architecture, can prove to be useful in many scenarios. In this chapter, we will discuss and observe how convolutional neural networks (CNNs) work in a similar way to dense neural networks (also called fully-connected neural networks, as mentioned in Chapter 2, Introduction to Computer Vision).

CNNs have neurons with weights and biases that are updated during training time. CNNs are mainly used for image processing. Images are interpreted as pixels and the network outputs the class it thinks the image belongs to, along with loss functions that state the errors with every classification and every output.

These types of networks make an assumption that the input is an image or works like an image, allowing them to work more efficiently (CNNs are faster and better than deep neural networks). In the following sections, you will learn...