Skip to yearly menu bar Skip to main content


Poster

Rethinking Variational Inference for Probabilistic Programs with Stochastic Support

Tim Reichelt · Luke Ong · Thomas Rainforth

Hall J (level 1) #719

Keywords: [ Variational Inference ] [ Probabilistic Programming ] [ Stochastic Support ]


Abstract:

We introduce Support Decomposition Variational Inference (SDVI), a new variational inference (VI) approach for probabilistic programs with stochastic support. Existing approaches to this problem rely on designing a single global variational guide on a variable-by-variable basis, while maintaining the stochastic control flow of the original program. SDVI instead breaks the program down into sub-programs with static support, before automatically building separate sub-guides for each. This decomposition significantly aids in the construction of suitable variational families, enabling, in turn, substantial improvements in inference performance.

Chat is not available.