Deep learning networks

Terry Benzschawel

The emergence of deep learning network applications has created a new era in machine learning. Speech recognition, language translation, image processing, autonomous driving and chatbots are examples of the potential of deep learning neural networks. And these are just the beginning. This chapter presents an overview of the general types of deep learning networks and their potential and considers examples of each of these applications. Such networks include multilayer perceptrons, convolutional networks, recurrent neural networks and long short-term memory (LSTM) networks.

13.1 AN OVERVIEW OF DEEP LEARNING NETWORKS

What is meant by “deep learning” neural networks? Leaders and experts in the field have various ideas of what deep learning is. The term deep learning originated with Hinton et al (2006), who used the phrase “deep” to describe the development of large artificial neural networks. Hinton and Salakhutdinov (2006) stuck with the term “deep” to describe their approach to developing networks with many more hidden layers than was previously typical. Thus, the name deep originated to describe neural networks with more than one hidden layer, as shown in Figure 13.1. Deep neural