liveBook · Manning

Kernel principal component analysis

As explained in Timeless Algorithms, The Foundational Ideas

Kernel principal component analysis applies principal component analysis in a feature space induced by a kernel to capture nonlinear structure.

Kernel principal component analysis (kernel PCA) represents relationships among observations with a kernel, then finds principal components in the resulting feature space. Unlike ordinary PCA’s linear directions, this can capture nonlinear structure without explicitly constructing every transformed feature. section 8.8.1

Kernel PCA belongs to the broader kernel-method family, not as a direct extension of support vector machines: it uses neither margins nor support vectors. section 8.8.1

How Timeless Algorithms, The Foundational Ideas covers it

How deep it goes
Explains kernel representations and feature-space components, but does not derive the method or discuss implementation.
What you can do after
Understand how kernel representations let principal component analysis capture nonlinear structure without explicitly constructing transformed features.
How it connects
Connects principal component analysis and kernel methods, while distinguishing kernel PCA from support vector machines and their margins.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Timeless Algorithms, The Foundational Ideas. The section links point to the parts of the book it draws on.