https://arxiv.org/pdf/1906.02691.pdf

https://arxiv.org/pdf/1606.05908.pdf

Setting

$\huge f:\mathcal{Z}\times\mathcal{\Theta}\to\mathcal{X}$

$\huge f(z;\theta)\in\mathcal{X}$

MLE

typically,

or Bernoulli

• How to choose $\mathcal{Z}$?
• How to calculate the integration over $z$?

Distribution mapping

So $\large Z\sim\mathcal{N}(0, I)$ is good enough.

Measure likelihood

Naïve integrate