Throughout this book, we have made use of machine learning techniques, with topic modeling, clustering and classifying algorithms, as well as what we call shallow learning – word embeddings. Word embeddings were our first glimpse into neural networks and the kind of semantic information they can learn.
Neural networks can be understood as a computing system or machine learning algorithm whose architecture is vaguely inspired by biological neurons in the brain. We say vaguely here because of the lack of thorough understanding we have of the human brain – through the neural connections and structure of the brain was certainly influential in some of the basic building blocks of neural networks, such as the perceptron [1] and single-layer neural network [2].
A neural network generally consists of a number of nodes that perform mathematical operations and interact with each other via connections. This model resembles a brain in the sense that the nodes tend to represent neurons and...
