Interactive demo · Foundations of Deep Learning · Chapter 5

Generative classifiers: Gaussian classes, linear and quadratic boundaries

Model each class by a Gaussian p(x | Ck) = N(x | μk, Σk) and a prior p(Ck); Bayes' theorem gives the posterior as a softmax of ak = ln p(x | Ck)p(Ck). If the classes share one covariance, the quadratic terms cancel, ak = wkTx + wk0 is linear and every boundary is a straight line; changing a prior only shifts the lines in parallel. If the covariances differ, the boundaries are quadratic curves. Fitting by maximum likelihood uses class fractions, class means and (for a shared covariance) the pooled covariance.

Classes

Covariances

Prior of the red class p(C₁)

Parameters

Scoreboard

Try these

In the lecture notes