Hands-On Artificial Intelligence for Cybersecurity

Dimensionality reduction with principal component analysis (PCA)

PCA makes it possible to identify the representative variables (also called principal components) of a dataset, selecting those along which the data is more spread out.

To understand why we need to perform dimensionality reduction of high-dimensional data (such as images) and how we can achieve dimensionality reduction using PCA, we can consider the following descriptive example.

Let's imagine we need to distinguish the nutritional value of foods; which nutrients should we consider among vitamins, proteins, fats, and carbohydrates?

To answer this question, we must be able to identify which nutrient acts as the main component, that is, which nutrient (or combination of nutrients) ...