Centre the comparisons

A journey through kernel principal components with Botie.

Classical Machine Learning · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

What If the Useful Direction Is Not Straight in the Original Picture?

Part 1 of 3

Ordinary PCA looks for straight directions through a cloud of observations. If the observations follow a curved structure, projecting onto a straight direction can place very different parts of that structure on top of one another.

Sources for this history

Schölkopf, Smola and Müller (1998), Nonlinear Component Analysis as a Kernel Eigenvalue Problem

← Previous lesson: Kernel Ridge Regression Read the full lesson and derivations