Practical Artificial Intelligence with Swift

Chapter 10. AI and ML Methods

In Part I and Part II we focused on the very practical and largely task-focused side of artificial intelligence. The other side of AI is theory, and the processes involved in making machine learning actually work. Thus, now that we have covered the how to implement a collection of AI and machine-learning features, this chapter will briefly discuss some of the many underlying methods that are used to give a system the appearance of intelligence.

Be aware there is no need to memorize the mathematics sometimes featured in this chapter; this is a practical book, as promised on the cover. A deep understanding of data science or statistics is not necessary, and skipping this chapter entirely would not make you any less equipped to follow the steps in later chapters to make an application with functional AI features. But to follow our mantra of effective, ethical, and appropriate use of AI, having an understanding of these basic principles is crucial.

Scary-looking algorithms are worked around with visual examples and metaphors that communicate the fundamental principles with small working examples. These demonstrate at a low level how each method goes about making decisions.

When applying this knowledge in future projects, having a broad understanding of how different methods work and what kinds of data they are most and least compatible with can allow better decisions to be made. Decisions that can be made in the design phase—when the cost of changes is ...