Analytic Posterior Classification

[scottperkins]

It would be nice to able to use a simple analytic form for the representation of the posterior. If, for example, we used a multi-normal distribution for the posterior and a Gaussian kernel KDE, you can analytical calculate the relevant integrals.