So far, we have covered a variety of unsupervised deep learning methodologies that can lead to many interesting applications, such as feature extraction, information compression, and data augmentation. However, as we move toward supervised deep learning methodologies that can perform classification or regression, for example, we have to begin by addressing an important question related to neural networks that might be in your mind already: what is the difference between wide and deep neural networks?
In this chapter, you will implement deep and wide neural networks to see the difference in the performance and complexities of both. As a bonus, we will cover the concepts of dense networks and sparse networks in terms of the connections between neurons. We will also optimize the dropout rates in our networks to maximize the generalization ability of...
