Neural networks are exceptionally effective at getting good characteristics for highly structured data. Physical phenomena are conditioned by numerous variables that can be easily measured through modern sensors. In this way, big data is produced that is difficult to deal with using classic techniques. Neural networks lend themselves to simulating complex environments.
In this chapter, we will learn how to develop models based on artificial neural networks (ANNs) to simulate physical phenomena. We will start by exploring the basic concepts of neural networks, and then we will examine their architecture and main elements. We will demonstrate how to train a network to update its weights. Then, we will apply these concepts to a practical use case to solve a regression problem. In the last part of the chapter, we will analyze deep neural networks.
In this chapter, we're going to cover the following topics: