Artificial Intelligence (AI) is rooted in mathematics and statistics. When creating an Artificial Neural Network (ANN), we're conducting mathematical operations on data represented in linear space; it is, by nature, applied mathematics and statistics. Machine learning algorithms are nothing but function approximations; they try and find a mapping between an input and a correct corresponding output. We use algebraic methods to create algorithms that learn these mappings.
Almost all machine learning can be expressed in a fairly straight-forward formula; bringing together a dataset and model, along with a loss function and optimization technique that are applicable to the dataset and model. This section is intended as a review of the basic mathematical tools and techniques that are essential to understanding what's under the hood in AI.
In this chapter...
