Deep Feedforward Networks

In this chapter, you will build our first deep learning network—deep deedforward networks (DFN). We will begin by discussing the evolutionary history of deep feedforward networks and then discuss the architecture of DFN. In any classification task, DFN plays an integral role. Apart from supporting the classification tasks, DFN standalone can be used both for regression and classification. Any deep learning network has a lot of elements like loss function, gradients, optimizers, and so on coming together to train the network. In this chapter, we will discuss these essential elements in detail. These elements will be common to all kinds of deep learning networks we are going to see in this book. We will also be demonstrating how to bring up and preprocess the data for training a deep learning network. You may find things a little difficult to understand...