What is machine learning and why does it work?

Speaking of expert systems, we said that machine learning was conceived to avoid the time-consuming process of extracting knowledge from human experts. There are cases where knowledge can’t be extracted at all from humans, such as computer vision and the recognition of some activity patterns. Each human has the knowledge needed for computer vision but this knowledge is stored on a subconscious level so it can’t be extracted. Moreover, by empowering software with the capability to learn, we might also furnish it with capabilities that no human has. This is the case of machine learning used in fraud detection, spam detection, and hacker attack prevention.

In the next subsection, we give some definitions and analyze machine learning’s main features.

Machine learning basics

The purpose of machine learning can be defined as follows:

Given a set F of functions and a set I of available information...