The interactive film is in the book's Companion

Play it, pause it, turn it and drag it with Mathematics for AI and Machine Learning.

Get the book

Whitening: PCA and ZCA

Chapter 13 · Mathematics for AI and Machine Learning · Advanced · 1:41

An interactive live film: a correlated Gaussian cloud and its bell are rotated into the eigenbasis, scaled to unit variance (PCA) and rotated back (ZCA). Rotate the white cloud and find the whitening that moves points least.

Narration

Here are samples of a Gaussian vector with covariance two, one, one, two, the example in the book. The two coordinates are correlated, so the cloud and its bell are tilted ellipses.

The spectral decomposition of the covariance gives its axes: the eigenvectors, with variances three and one along them.

The rest of the narration is in the book's Companion.

More from Chapter 13

Interactive1:35Normalizing Flows: Chaining Invertible MapsChapter 13Preview1:38Independence as an Outer ProductChapter 13Preview1:40Conditional Gaussian: A Slice of the BellChapter 13Preview1:22Random Fourier FeaturesChapter 13Preview1:46Change of Variables: Same Probability, Different AreaChapter 13Preview1:35The von Mises-Fisher Distribution: A Gaussian for DirectionsChapter 13Preview1:42The Dirichlet Distribution: A Density on the SimplexChapter 13Preview1:14Gaussian Mixture Model: Multimodal TerrainChapter 13Preview1:36Beta and Bernoulli: Updating with Pseudo-CountsChapter 13Preview1:43ROC Curve and AUC: the Mass Above the DiagonalChapter 13Preview1:40Hoeffding's Inequality: How Fast a Sample Mean SettlesChapter 13Preview1:25Temperature Scaling: Calibrating a SoftmaxChapter 13Preview1:34Mahalanobis Distance: The Ellipsoids of a GaussianChapter 13Preview2:09Random Directions Are Nearly OrthogonalChapter 13