Detailed explanation of the anomaly detection algorithm of unsupervised learning

Anomaly Detection

Motivation of the question

In the next series of videos, I will introduce you to the problem of anomaly detection . This is a common application of machine learning algorithms. An interesting thing about this algorithm is that although it is mainly used for unsupervised learning problems, from some perspectives, it is similar to some supervised learning problems.

What is anomaly detection? To explain this concept, let me give you an example:

Imagine you are an aircraft engine manufacturer. When the aircraft engines you produce come off the production line, you need to conduct QA (quality control testing), and as part of this test, you measure some characteristic variables of the aircraft engine, such as The heat generated during operation, or the vibration of the engine, etc.

In this way, you have a data set starting from x (1) x^{(1)}