feature request: gradients of expected values about probability HOT 3 OPEN

Hi Jackson! This is a great request. Others might have more to say here, but I think the current state of things is that walking the TF graph is strongly discouraged, and code that tries to do so using unpublished APIs is subject to breakage without notice. The brittleness of graph-walking in Edward was a primary motivation for the development of Edward2, which uses its own tracing mechanism to avoid directly walking the TF graph.

Subject to this restriction, you might find that the expectation utility (https://github.com/tensorflow/probability/blob/master/tensorflow_probability/python/monte_carlo.py#L29) does some of what you're asking for: given an explicit source of randomness, it returns a Monte Carlo expectation with unbiased stochastic gradient, using the reparametrization or score-function estimators as appropriate. You can use this to effectively construct stochastic computation graphs, albeit in perhaps a slightly lower-level way than you're thinking of. We'd certainly be excited about designs to make this sort of functionality more cleanly accessible.

from probability.

Is there a white paper somewhere outlining to scope of the API edward2 is planning to compass? Or is the main idea just to rewrite Edward in a less brittle way?

from probability.

@dustinvtran want to take this?

from probability.