We already have that - variational inference based algorithms like BBVI use grad... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		RJTrolo on Oct 20, 2017 \| parent \| context \| favorite \| on: Edward – A Turing-Complete Language for Deep Proba... We already have that - variational inference based algorithms like BBVI use gradient descent for training

eli_gottlieb on Oct 20, 2017 [–]

Variational inference also only works for continuous probability models, so it can't be used for most interesting use-cases of probabilistic programming.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact